Compare commits

...

28 Commits

Author SHA1 Message Date
bb2d9a3afe Release 0.5.5
- Transition to CMake build system
- Fix Monitor being unable to change PG sizes
- Fix PG optimizer not using some OSDs in some cases
- Fix inability to change PG count online
- Improve journal flusher performance
- Add a little better systemd unit generator
- Use w=8 with jerasure (breaking change for EC pools)
2021-02-26 01:59:18 +03:00
e899ed2c25 Make OSDs with 256 flushers (as they are now dynamic) 2021-02-26 01:59:18 +03:00
e21b14b72c Fix rpm specs for building with CMake 2021-02-26 01:59:18 +03:00
5af8eddaa9 Add the remaining build script for Debian 2021-02-26 01:59:18 +03:00
4f5a94c07a Modify instructions for the CMake build 2021-02-26 00:28:57 +03:00
e16b87ecc8 Rename random_combinations() parameter from "unordered" to "ordered" as it's more correct 2021-02-25 23:59:34 +03:00
fcb4aa0a11 Fix Monitor being unable to change PG sizes 2021-02-25 23:59:34 +03:00
12adfa470c Add a test for changing PG size 2021-02-25 23:59:33 +03:00
7f15e0c084 Add a simple test for the PG optimizer 2021-02-25 23:59:33 +03:00
08d4bef419 Fix PG optimizer removing PGs without adding new ones
This happened when the distribution was already valid for the current OSD tree,
but didn't use all OSDs. For example, OSDs 1 2 3 and all PGs equal to [ 1, 2 ]
remained unchanged.
2021-02-25 23:59:33 +03:00
2d73b19a6c Fix online PG count change bugs 2021-02-25 23:59:33 +03:00
69c87009e9 Add a test for changing PG count 2021-02-25 23:59:33 +03:00
c974cb539c Make flusher_count adaptive and limit write iodepth 2021-02-25 23:59:33 +03:00
00e98f64f3 A little better systemd unit generator 2021-02-25 23:59:33 +03:00
91a70dfb1b Add a test for the no_same_sector_overwrites mode 2021-02-25 23:59:33 +03:00
178388ac8c Use packages/ subdir instead of build/ for Docker package builds 2021-02-25 23:59:04 +03:00
bf9a175efc Move C/C++ sources to src subdirectory 2021-02-25 23:59:03 +03:00
08aed962de Use CMake 2021-02-25 23:58:08 +03:00
8c65e890b9 Slightly clean up the build script 2021-02-25 23:56:54 +03:00
8cda70b889 Allow to enable AddressSanitizer with "ASAN=1 make" 2021-02-25 23:55:33 +03:00
61ab22403a Use w=8 with jerasure 2021-02-25 23:55:33 +03:00
16da663a66 Add another test for failure domains 2021-02-25 23:55:33 +03:00
4a2dcf7b6b Update the license to VNPL 1.1
VNPL 1.1 is slightly reworded to make it clear that proprietary software
interacting with Vitastor and providing some kind of service to end users isn't
a "Proxy Program" if it's not specially designed to be used with Vitastor.

For example, Windows OS running in a virtual machine stored in a Vitastor
cluster clearly isn't.
2021-02-25 23:55:33 +03:00
8d48cc56b0 Generate randomly permutated OSD combinations when optimizing for compressed chunks 2021-02-25 23:55:33 +03:00
9f58f01425 Mirror afr.js from /vitalif/ceph-afr-calc 2021-02-25 23:55:33 +03:00
b9e7d31aa1 Release v0.5.4
- Fix a rare hang, more or less reproducible with very slow drives
- Fix a hang with the no_same_sector_overwrites mode
2021-02-24 01:40:30 +03:00
2d9f09dcb6 Attempt forced trim when stopping an overrun flusher
Fixes a rare hang happening in the event of journal space running out without
new work to do for flushers except the current sector.
The hang could be reproduced more or less consistently with very slow drives.
2021-02-24 01:33:01 +03:00
7cc59260c5 Fix no_same_sector_overwrites related bug 2021-02-23 18:50:51 +03:00
126 changed files with 1025 additions and 772 deletions

View File

@@ -1,5 +1,6 @@
.git
build
packages
mon/node_modules
*.o
*.so
@@ -15,3 +16,4 @@ fio
qemu
rpm/*.Dockerfile
debian/*.Dockerfile
Dockerfile

5
CMakeLists.txt Normal file
View File

@@ -0,0 +1,5 @@
cmake_minimum_required(VERSION 2.8)
project(vitastor)
add_subdirectory(src)

View File

@@ -1,46 +0,0 @@
#!/usr/bin/perl
use strict;
my $deps = {};
for my $line (split /\n/, `grep '^#include "' *.cpp *.h`)
{
if ($line =~ /^([^:]+):\#include "([^"]+)"/s)
{
$deps->{$1}->{$2} = 1;
}
}
my $added;
do
{
$added = 0;
for my $file (keys %$deps)
{
for my $dep (keys %{$deps->{$file}})
{
if ($deps->{$dep})
{
for my $subdep (keys %{$deps->{$dep}})
{
if (!$deps->{$file}->{$subdep})
{
$added = 1;
$deps->{$file}->{$subdep} = 1;
}
}
}
}
}
} while ($added);
for my $file (sort keys %$deps)
{
if ($file =~ /\.cpp$/)
{
my $obj = $file;
$obj =~ s/\.cpp$/.o/s;
print "$obj: $file ".join(" ", sort keys %{$deps->{$file}})."\n";
print "\tg++ \$(CXXFLAGS) -c -o \$\@ \$\<\n";
}
}

195
Makefile
View File

@@ -1,195 +0,0 @@
BINDIR ?= /usr/bin
LIBDIR ?= /usr/lib/x86_64-linux-gnu
QEMU_PLUGINDIR ?= /usr/lib/x86_64-linux-gnu/qemu
BLOCKSTORE_OBJS := allocator.o blockstore.o blockstore_impl.o blockstore_init.o blockstore_open.o blockstore_journal.o blockstore_read.o \
blockstore_write.o blockstore_sync.o blockstore_stable.o blockstore_rollback.o blockstore_flush.o crc32c.o ringloop.o
# -fsanitize=address
CXXFLAGS := -g -O3 -Wall -Wno-sign-compare -Wno-comment -Wno-parentheses -Wno-pointer-arith -fPIC -fdiagnostics-color=always -I/usr/include/jerasure
all: libfio_blockstore.so osd libfio_sec_osd.so libfio_cluster.so stub_osd stub_uring_osd stub_bench osd_test dump_journal qemu_driver.so nbd_proxy rm_inode
clean:
rm -f *.o libblockstore.so libfio_blockstore.so osd libfio_sec_osd.so libfio_cluster.so stub_osd stub_uring_osd stub_bench osd_test dump_journal qemu_driver.so nbd_proxy rm_inode
install: all
mkdir -p $(DESTDIR)$(LIBDIR)/vitastor
install -m 0755 libfio_sec_osd.so $(DESTDIR)$(LIBDIR)/vitastor/
install -m 0755 libfio_cluster.so $(DESTDIR)$(LIBDIR)/vitastor/
install -m 0755 libfio_blockstore.so $(DESTDIR)$(LIBDIR)/vitastor/
install -m 0755 libblockstore.so $(DESTDIR)$(LIBDIR)/vitastor/
mkdir -p $(DESTDIR)$(BINDIR)
install -m 0755 osd $(DESTDIR)$(BINDIR)/vitastor-osd
install -m 0755 dump_journal $(DESTDIR)$(BINDIR)/vitastor-dump-journal
install -m 0755 nbd_proxy $(DESTDIR)$(BINDIR)/vitastor-nbd
install -m 0755 rm_inode $(DESTDIR)$(BINDIR)/vitastor-rm
mkdir -p $(DESTDIR)$(QEMU_PLUGINDIR)
install -m 0755 qemu_driver.so $(DESTDIR)$(QEMU_PLUGINDIR)/block-vitastor.so
dump_journal: dump_journal.cpp crc32c.o blockstore_journal.h
g++ $(CXXFLAGS) -o $@ $< crc32c.o
libblockstore.so: $(BLOCKSTORE_OBJS)
g++ $(CXXFLAGS) -o $@ -shared $(BLOCKSTORE_OBJS) -ltcmalloc_minimal -luring
libfio_blockstore.so: ./libblockstore.so fio_engine.o json11.o
g++ $(CXXFLAGS) -Wl,-rpath,'$(LIBDIR)/vitastor',-rpath,'$$ORIGIN' -shared -o $@ fio_engine.o json11.o libblockstore.so -ltcmalloc_minimal -luring
OSD_OBJS := osd.o osd_secondary.o msgr_receive.o msgr_send.o osd_peering.o osd_flush.o osd_peering_pg.o \
osd_primary.o osd_primary_subops.o etcd_state_client.o messenger.o osd_cluster.o http_client.o osd_ops.o pg_states.o \
osd_rmw.o json11.o base64.o timerfd_manager.o epoll_manager.o
osd: ./libblockstore.so osd_main.cpp osd.h osd_ops.h $(OSD_OBJS)
g++ $(CXXFLAGS) -Wl,-rpath,'$(LIBDIR)/vitastor',-rpath,'$$ORIGIN' -o $@ osd_main.cpp $(OSD_OBJS) libblockstore.so -ltcmalloc_minimal -luring -lJerasure
stub_osd: stub_osd.o rw_blocking.o
g++ $(CXXFLAGS) -o $@ stub_osd.o rw_blocking.o -ltcmalloc_minimal
osd_rmw_test: osd_rmw_test.o
g++ $(CXXFLAGS) -o $@ osd_rmw_test.o -lJerasure -fsanitize=address
STUB_URING_OSD_OBJS := stub_uring_osd.o epoll_manager.o messenger.o msgr_send.o msgr_receive.o ringloop.o timerfd_manager.o json11.o
stub_uring_osd: $(STUB_URING_OSD_OBJS)
g++ $(CXXFLAGS) -o $@ -ltcmalloc_minimal $(STUB_URING_OSD_OBJS) -luring
stub_bench: stub_bench.cpp osd_ops.h rw_blocking.o
g++ $(CXXFLAGS) -o $@ stub_bench.cpp rw_blocking.o -ltcmalloc_minimal
osd_test: osd_test.cpp osd_ops.h rw_blocking.o
g++ $(CXXFLAGS) -o $@ osd_test.cpp rw_blocking.o -ltcmalloc_minimal
osd_peering_pg_test: osd_peering_pg_test.cpp osd_peering_pg.o
g++ $(CXXFLAGS) -o $@ $< osd_peering_pg.o -ltcmalloc_minimal
libfio_sec_osd.so: fio_sec_osd.o rw_blocking.o
g++ $(CXXFLAGS) -ltcmalloc_minimal -shared -o $@ fio_sec_osd.o rw_blocking.o
FIO_CLUSTER_OBJS := cluster_client.o epoll_manager.o etcd_state_client.o \
messenger.o msgr_send.o msgr_receive.o ringloop.o json11.o http_client.o osd_ops.o pg_states.o timerfd_manager.o base64.o
libfio_cluster.so: fio_cluster.o $(FIO_CLUSTER_OBJS)
g++ $(CXXFLAGS) -ltcmalloc_minimal -shared -o $@ $< $(FIO_CLUSTER_OBJS) -luring
nbd_proxy: nbd_proxy.o $(FIO_CLUSTER_OBJS)
g++ $(CXXFLAGS) -ltcmalloc_minimal -o $@ $< $(FIO_CLUSTER_OBJS) -luring
rm_inode: rm_inode.o $(FIO_CLUSTER_OBJS)
g++ $(CXXFLAGS) -ltcmalloc_minimal -o $@ $< $(FIO_CLUSTER_OBJS) -luring
qemu_driver.o: qemu_driver.c qemu_proxy.h
gcc -I qemu/b/qemu `pkg-config glib-2.0 --cflags` \
-I qemu/include $(CXXFLAGS) -c -o $@ $<
qemu_driver.so: qemu_driver.o qemu_proxy.o $(FIO_CLUSTER_OBJS)
g++ $(CXXFLAGS) -ltcmalloc_minimal -shared -o $@ $(FIO_CLUSTER_OBJS) qemu_driver.o qemu_proxy.o -luring
test_blockstore: ./libblockstore.so test_blockstore.cpp timerfd_interval.o
g++ $(CXXFLAGS) -Wl,-rpath,'$(LIBDIR)/vitastor',-rpath,'$$ORIGIN' -o test_blockstore test_blockstore.cpp timerfd_interval.o libblockstore.so -ltcmalloc_minimal -luring
test_shit: test_shit.cpp osd_peering_pg.o
g++ $(CXXFLAGS) -o test_shit test_shit.cpp -luring -lm
test_allocator: test_allocator.cpp allocator.o
g++ $(CXXFLAGS) -o test_allocator test_allocator.cpp allocator.o
crc32c.o: crc32c.c crc32c.h
g++ $(CXXFLAGS) -c -o $@ $<
json11.o: json11/json11.cpp
g++ $(CXXFLAGS) -c -o json11.o json11/json11.cpp
# Autogenerated
allocator.o: allocator.cpp allocator.h
g++ $(CXXFLAGS) -c -o $@ $<
base64.o: base64.cpp base64.h
g++ $(CXXFLAGS) -c -o $@ $<
blockstore.o: blockstore.cpp allocator.h blockstore.h blockstore_flush.h blockstore_impl.h blockstore_init.h blockstore_journal.h cpp-btree/btree_map.h crc32c.h malloc_or_die.h object_id.h ringloop.h
g++ $(CXXFLAGS) -c -o $@ $<
blockstore_flush.o: blockstore_flush.cpp allocator.h blockstore.h blockstore_flush.h blockstore_impl.h blockstore_init.h blockstore_journal.h cpp-btree/btree_map.h crc32c.h malloc_or_die.h object_id.h ringloop.h
g++ $(CXXFLAGS) -c -o $@ $<
blockstore_impl.o: blockstore_impl.cpp allocator.h blockstore.h blockstore_flush.h blockstore_impl.h blockstore_init.h blockstore_journal.h cpp-btree/btree_map.h crc32c.h malloc_or_die.h object_id.h ringloop.h
g++ $(CXXFLAGS) -c -o $@ $<
blockstore_init.o: blockstore_init.cpp allocator.h blockstore.h blockstore_flush.h blockstore_impl.h blockstore_init.h blockstore_journal.h cpp-btree/btree_map.h crc32c.h malloc_or_die.h object_id.h ringloop.h
g++ $(CXXFLAGS) -c -o $@ $<
blockstore_journal.o: blockstore_journal.cpp allocator.h blockstore.h blockstore_flush.h blockstore_impl.h blockstore_init.h blockstore_journal.h cpp-btree/btree_map.h crc32c.h malloc_or_die.h object_id.h ringloop.h
g++ $(CXXFLAGS) -c -o $@ $<
blockstore_open.o: blockstore_open.cpp allocator.h blockstore.h blockstore_flush.h blockstore_impl.h blockstore_init.h blockstore_journal.h cpp-btree/btree_map.h crc32c.h malloc_or_die.h object_id.h ringloop.h
g++ $(CXXFLAGS) -c -o $@ $<
blockstore_read.o: blockstore_read.cpp allocator.h blockstore.h blockstore_flush.h blockstore_impl.h blockstore_init.h blockstore_journal.h cpp-btree/btree_map.h crc32c.h malloc_or_die.h object_id.h ringloop.h
g++ $(CXXFLAGS) -c -o $@ $<
blockstore_rollback.o: blockstore_rollback.cpp allocator.h blockstore.h blockstore_flush.h blockstore_impl.h blockstore_init.h blockstore_journal.h cpp-btree/btree_map.h crc32c.h malloc_or_die.h object_id.h ringloop.h
g++ $(CXXFLAGS) -c -o $@ $<
blockstore_stable.o: blockstore_stable.cpp allocator.h blockstore.h blockstore_flush.h blockstore_impl.h blockstore_init.h blockstore_journal.h cpp-btree/btree_map.h crc32c.h malloc_or_die.h object_id.h ringloop.h
g++ $(CXXFLAGS) -c -o $@ $<
blockstore_sync.o: blockstore_sync.cpp allocator.h blockstore.h blockstore_flush.h blockstore_impl.h blockstore_init.h blockstore_journal.h cpp-btree/btree_map.h crc32c.h malloc_or_die.h object_id.h ringloop.h
g++ $(CXXFLAGS) -c -o $@ $<
blockstore_write.o: blockstore_write.cpp allocator.h blockstore.h blockstore_flush.h blockstore_impl.h blockstore_init.h blockstore_journal.h cpp-btree/btree_map.h crc32c.h malloc_or_die.h object_id.h ringloop.h
g++ $(CXXFLAGS) -c -o $@ $<
cluster_client.o: cluster_client.cpp cluster_client.h etcd_state_client.h http_client.h json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd_id.h osd_ops.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
dump_journal.o: dump_journal.cpp allocator.h blockstore.h blockstore_flush.h blockstore_impl.h blockstore_init.h blockstore_journal.h cpp-btree/btree_map.h crc32c.h malloc_or_die.h object_id.h ringloop.h
g++ $(CXXFLAGS) -c -o $@ $<
epoll_manager.o: epoll_manager.cpp epoll_manager.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
etcd_state_client.o: etcd_state_client.cpp base64.h etcd_state_client.h http_client.h json11/json11.hpp object_id.h osd_id.h osd_ops.h pg_states.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
fio_cluster.o: fio_cluster.cpp cluster_client.h epoll_manager.h etcd_state_client.h fio/arch/arch.h fio/fio.h fio/optgroup.h fio_headers.h http_client.h json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd_id.h osd_ops.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
fio_engine.o: fio_engine.cpp blockstore.h fio/arch/arch.h fio/fio.h fio/optgroup.h fio_headers.h json11/json11.hpp object_id.h ringloop.h
g++ $(CXXFLAGS) -c -o $@ $<
fio_sec_osd.o: fio_sec_osd.cpp fio/arch/arch.h fio/fio.h fio/optgroup.h fio_headers.h object_id.h osd_id.h osd_ops.h rw_blocking.h
g++ $(CXXFLAGS) -c -o $@ $<
http_client.o: http_client.cpp http_client.h json11/json11.hpp timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
messenger.o: messenger.cpp json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd_id.h osd_ops.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
msgr_receive.o: msgr_receive.cpp json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd_id.h osd_ops.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
msgr_send.o: msgr_send.cpp json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd_id.h osd_ops.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
nbd_proxy.o: nbd_proxy.cpp cluster_client.h epoll_manager.h etcd_state_client.h http_client.h json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd_id.h osd_ops.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
osd.o: osd.cpp blockstore.h cpp-btree/btree_map.h epoll_manager.h etcd_state_client.h http_client.h json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd.h osd_id.h osd_ops.h osd_peering_pg.h pg_states.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
osd_cluster.o: osd_cluster.cpp base64.h blockstore.h cpp-btree/btree_map.h epoll_manager.h etcd_state_client.h http_client.h json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd.h osd_id.h osd_ops.h osd_peering_pg.h pg_states.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
osd_flush.o: osd_flush.cpp blockstore.h cpp-btree/btree_map.h epoll_manager.h etcd_state_client.h http_client.h json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd.h osd_id.h osd_ops.h osd_peering_pg.h pg_states.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
osd_main.o: osd_main.cpp blockstore.h cpp-btree/btree_map.h epoll_manager.h etcd_state_client.h http_client.h json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd.h osd_id.h osd_ops.h osd_peering_pg.h pg_states.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
osd_ops.o: osd_ops.cpp object_id.h osd_id.h osd_ops.h
g++ $(CXXFLAGS) -c -o $@ $<
osd_peering.o: osd_peering.cpp base64.h blockstore.h cpp-btree/btree_map.h epoll_manager.h etcd_state_client.h http_client.h json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd.h osd_id.h osd_ops.h osd_peering_pg.h pg_states.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
osd_peering_pg.o: osd_peering_pg.cpp cpp-btree/btree_map.h object_id.h osd_id.h osd_ops.h osd_peering_pg.h pg_states.h
g++ $(CXXFLAGS) -c -o $@ $<
osd_peering_pg_test.o: osd_peering_pg_test.cpp cpp-btree/btree_map.h object_id.h osd_id.h osd_ops.h osd_peering_pg.h pg_states.h
g++ $(CXXFLAGS) -c -o $@ $<
osd_primary.o: osd_primary.cpp blockstore.h cpp-btree/btree_map.h epoll_manager.h etcd_state_client.h http_client.h json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd.h osd_id.h osd_ops.h osd_peering_pg.h osd_primary.h osd_rmw.h pg_states.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
osd_primary_subops.o: osd_primary_subops.cpp blockstore.h cpp-btree/btree_map.h epoll_manager.h etcd_state_client.h http_client.h json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd.h osd_id.h osd_ops.h osd_peering_pg.h osd_primary.h osd_rmw.h pg_states.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
osd_rmw.o: osd_rmw.cpp malloc_or_die.h object_id.h osd_id.h osd_rmw.h xor.h
g++ $(CXXFLAGS) -c -o $@ $<
osd_rmw_test.o: osd_rmw_test.cpp malloc_or_die.h object_id.h osd_id.h osd_rmw.cpp osd_rmw.h test_pattern.h xor.h
g++ $(CXXFLAGS) -c -o $@ $<
osd_secondary.o: osd_secondary.cpp blockstore.h cpp-btree/btree_map.h epoll_manager.h etcd_state_client.h http_client.h json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd.h osd_id.h osd_ops.h osd_peering_pg.h pg_states.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
osd_test.o: osd_test.cpp object_id.h osd_id.h osd_ops.h rw_blocking.h test_pattern.h
g++ $(CXXFLAGS) -c -o $@ $<
pg_states.o: pg_states.cpp pg_states.h
g++ $(CXXFLAGS) -c -o $@ $<
qemu_proxy.o: qemu_proxy.cpp cluster_client.h etcd_state_client.h http_client.h json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd_id.h osd_ops.h qemu_proxy.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
ringloop.o: ringloop.cpp ringloop.h
g++ $(CXXFLAGS) -c -o $@ $<
rm_inode.o: rm_inode.cpp cluster_client.h etcd_state_client.h http_client.h json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd_id.h osd_ops.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
rw_blocking.o: rw_blocking.cpp rw_blocking.h
g++ $(CXXFLAGS) -c -o $@ $<
stub_bench.o: stub_bench.cpp object_id.h osd_id.h osd_ops.h rw_blocking.h
g++ $(CXXFLAGS) -c -o $@ $<
stub_osd.o: stub_osd.cpp object_id.h osd_id.h osd_ops.h rw_blocking.h
g++ $(CXXFLAGS) -c -o $@ $<
stub_uring_osd.o: stub_uring_osd.cpp epoll_manager.h json11/json11.hpp malloc_or_die.h messenger.h object_id.h osd_id.h osd_ops.h ringloop.h timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<
test_allocator.o: test_allocator.cpp allocator.h
g++ $(CXXFLAGS) -c -o $@ $<
test_blockstore.o: test_blockstore.cpp blockstore.h object_id.h ringloop.h timerfd_interval.h
g++ $(CXXFLAGS) -c -o $@ $<
test_shit.o: test_shit.cpp allocator.h blockstore.h blockstore_flush.h blockstore_impl.h blockstore_init.h blockstore_journal.h cpp-btree/btree_map.h crc32c.h malloc_or_die.h object_id.h osd_id.h osd_ops.h osd_peering_pg.h pg_states.h ringloop.h
g++ $(CXXFLAGS) -c -o $@ $<
timerfd_interval.o: timerfd_interval.cpp ringloop.h timerfd_interval.h
g++ $(CXXFLAGS) -c -o $@ $<
timerfd_manager.o: timerfd_manager.cpp timerfd_manager.h
g++ $(CXXFLAGS) -c -o $@ $<

View File

@@ -336,9 +336,8 @@ Vitastor with single-thread NBD on the same hardware:
- You can also rebuild QEMU with a patch that makes LD_PRELOAD unnecessary to load vitastor driver.
See `qemu-*.*-vitastor.patch`.
- Install fio 3.7 or later, get its source and symlink it into `<vitastor>/fio`.
- Build Vitastor with `make -j8`.
- Run `make install` (optionally with `LIBDIR=/usr/lib64 QEMU_PLUGINDIR=/usr/lib64/qemu-kvm`
if you're using an RPM-based distro).
- Build & install Vitastor with `mkdir build && cd build && cmake .. && make -j8 && make install`.
Pay attention to the `QEMU_PLUGINDIR` cmake option - it must be set to `qemu-kvm` on RHEL.
## Running
@@ -349,21 +348,16 @@ and calculate disk offsets almost by hand. This will be fixed in near future.
with lazy fsync, but prepare for inferior single-thread latency.
- Get a fast network (at least 10 Gbit/s).
- Disable CPU powersaving: `cpupower idle-set -D 0 && cpupower frequency-set -g performance`.
- Start etcd with `--max-txn-ops=100000 --auto-compaction-retention=10 --auto-compaction-mode=revision` options.
- Create global configuration in etcd: `etcdctl --endpoints=... put /vitastor/config/global '{"immediate_commit":"all"}'`
(if all your drives have capacitors).
- Create pool configuration in etcd: `etcdctl --endpoints=... put /vitastor/config/pools '{"1":{"name":"testpool","scheme":"replicated","pg_size":2,"pg_minsize":1,"pg_count":256,"failure_domain":"host"}}'`.
For jerasure pools the configuration should look like the following: `2:{"name":"ecpool","scheme":"jerasure","pg_size":4,"parity_chunks":2,"pg_minsize":2,"pg_count":256,"failure_domain":"host"}`.
- Calculate offsets for your drives with `node /usr/lib/vitastor/mon/simple-offsets.js --device /dev/sdX`.
- Make systemd units for your OSDs. Look at `/usr/lib/vitastor/mon/make-units.sh` for example.
Notable configuration variables from the example:
- Check `/usr/lib/vitastor/mon/make-units.sh` and `/usr/lib/vitastor/mon/make-osd.sh` and
put desired values into the variables at the top of these files.
- Create systemd units for the monitor and etcd: `/usr/lib/vitastor/mon/make-units.sh`
- Create systemd units for your OSDs: `/usr/lib/vitastor/mon/make-osd.sh /dev/disk/by-partuuid/XXX [/dev/disk/by-partuuid/YYY ...]`
- You can edit the units and change OSD configuration. Notable configuration variables:
- `disable_data_fsync 1` - only safe with server-grade drives with capacitors.
- `immediate_commit all` - use this if all your drives are server-grade.
- `disable_device_lock 1` - only required if you run multiple OSDs on one block device.
- `flusher_count 16` - flusher is a micro-thread that removes old data from the journal.
More flushers mean more aggressive journal flushing which allows for more throughput
but slightly hurts latency under less load. Flushing will probably be improved in the future
because currently high queue depths sometimes lead to performance degradation.
- `flusher_count 256` - flusher is a micro-thread that removes old data from the journal.
You don't have to worry about this parameter anymore, 256 is enough.
- `disk_alignment`, `journal_block_size`, `meta_block_size` should be set to the internal
block size of your SSDs which is 4096 on most drives.
- `journal_no_same_sector_overwrites true` prevents multiple overwrites of the same journal sector.
@@ -374,18 +368,22 @@ and calculate disk offsets almost by hand. This will be fixed in near future.
setting is set, it is also required to raise `journal_sector_buffer_count` setting, which is the
number of dirty journal sectors that may be written to at the same time.
- `systemctl start vitastor.target` everywhere.
- Start any number of monitors: `node /usr/lib/vitastor/mon/mon-main.js --etcd_url 'http://10.115.0.10:2379,http://10.115.0.11:2379,http://10.115.0.12:2379,http://10.115.0.13:2379' --etcd_prefix '/vitastor' --etcd_start_timeout 5`.
- Create global configuration in etcd: `etcdctl --endpoints=... put /vitastor/config/global '{"immediate_commit":"all"}'`
(if all your drives have capacitors).
- Create pool configuration in etcd: `etcdctl --endpoints=... put /vitastor/config/pools '{"1":{"name":"testpool","scheme":"replicated","pg_size":2,"pg_minsize":1,"pg_count":256,"failure_domain":"host"}}'`.
For jerasure pools the configuration should look like the following: `2:{"name":"ecpool","scheme":"jerasure","pg_size":4,"parity_chunks":2,"pg_minsize":2,"pg_count":256,"failure_domain":"host"}`.
- At this point, one of the monitors will configure PGs and OSDs will start them.
- You can check PG states with `etcdctl --endpoints=... get --prefix /vitastor/pg/state`. All PGs should become 'active'.
- Run tests with (for example): `fio -thread -ioengine=/usr/lib/x86_64-linux-gnu/vitastor/libfio_cluster.so -name=test -bs=4M -direct=1 -iodepth=16 -rw=write -etcd=10.115.0.10:2379/v3 -pool=1 -inode=1 -size=400G`.
- Run tests with (for example): `fio -thread -ioengine=libfio_vitastor.so -name=test -bs=4M -direct=1 -iodepth=16 -rw=write -etcd=10.115.0.10:2379/v3 -pool=1 -inode=1 -size=400G`.
- Upload VM disk image with qemu-img (for example):
```
LD_PRELOAD=/usr/lib/x86_64-linux-gnu/qemu/block-vitastor.so qemu-img convert -f qcow2 debian10.qcow2 -p
-O raw 'vitastor:etcd_host=10.115.0.10\:2379/v3:pool=1:inode=1:size=2147483648'
qemu-img convert -f qcow2 debian10.qcow2 -p -O raw 'vitastor:etcd_host=10.115.0.10\:2379/v3:pool=1:inode=1:size=2147483648'
```
Note that the command requires to be run with `LD_PRELOAD=/usr/lib/x86_64-linux-gnu/qemu/block-vitastor.so qemu-img ...`
if you use unmodified QEMU.
- Run QEMU with (for example):
```
LD_PRELOAD=/usr/lib/x86_64-linux-gnu/qemu/block-vitastor.so qemu-system-x86_64 -enable-kvm -m 1024
qemu-system-x86_64 -enable-kvm -m 1024
-drive 'file=vitastor:etcd_host=10.115.0.10\:2379/v3:pool=1:inode=1:size=2147483648',format=raw,if=none,id=drive-virtio-disk0,cache=none
-device virtio-blk-pci,scsi=off,bus=pci.0,addr=0x5,drive=drive-virtio-disk0,id=virtio-disk0,bootindex=1,write-cache=off,physical_block_size=4096,logical_block_size=512
-vnc 0.0.0.0:0
@@ -421,22 +419,27 @@ Copyright (c) Vitaliy Filippov (vitalif [at] yourcmc.ru), 2019+
You can also find me in the Russian Telegram Ceph chat: https://t.me/ceph_ru
All server-side code (OSD, Monitor and so on) is licensed under the terms of
Vitastor Network Public License 1.0 (VNPL 1.0), a copyleft license based on
Vitastor Network Public License 1.1 (VNPL 1.1), a copyleft license based on
GNU GPLv3.0 with the additional "Network Interaction" clause which requires
opensourcing all programs directly or indirectly interacting with Vitastor
through a computer network ("Proxy Programs"). Proxy Programs may be made public
not only under the terms of the same license, but also under the terms of any
GPL-Compatible Free Software License, as listed by the Free Software Foundation.
through a computer network and expressly designed to be used in conjunction
with it ("Proxy Programs"). Proxy Programs may be made public not only under
the terms of the same license, but also under the terms of any GPL-Compatible
Free Software License, as listed by the Free Software Foundation.
This is a stricter copyleft license than the Affero GPL.
Please note that VNPL doesn't require you to open the code of proprietary
software running inside a VM if it's not specially designed to be used with
Vitastor.
Basically, you can't use the software in a proprietary environment to provide
its functionality to users without opensourcing all intermediary components
standing between the user and Vitastor or purchasing a commercial license
from the author 😀.
Client libraries (cluster_client and so on) are dual-licensed under the same
VNPL 1.0 and also GNU GPL 2.0 or later to allow for compatibility with GPLed
VNPL 1.1 and also GNU GPL 2.0 or later to allow for compatibility with GPLed
software like QEMU and fio.
You can find the full text of VNPL-1.0 in the file [VNPL-1.0.txt](VNPL-1.0.txt).
You can find the full text of VNPL-1.1 in the file [VNPL-1.1.txt](VNPL-1.1.txt).
GPL 2.0 is also included in this repository as [GPL-2.0.txt](GPL-2.0.txt).

View File

@@ -1,7 +1,7 @@
VITASTOR NETWORK PUBLIC LICENSE
Version 1, 17 September 2020
Version 1.1, 6 February 2021
Copyright (C) 2020 Vitaliy Filippov <vitalif@yourcmc.ru>
Copyright (C) 2021 Vitaliy Filippov <vitalif@yourcmc.ru>
Everyone is permitted to copy and distribute verbatim copies
of this license document, but changing it is not allowed.
@@ -540,12 +540,15 @@ License would be to refrain entirely from conveying the Program.
13. Remote Network Interaction.
Notwithstanding any other provision of this License, if you provide
any user an opportunity to interact with the covered work directly
or indirectly through a computer network, an imitation of such network,
or an additional program (hereinafter referred to as a "Proxy Program")
that, in turn, interacts with the covered work through a computer network,
an imitation of such network, or another Proxy Program itself,
A "Proxy Program" means a separate program which is specially designed to
be used in conjunction with the covered work and interacts with it directly
or indirectly through any kind of API (application programming interfaces),
a computer network, an imitation of such network, or another Proxy Program
itself.
Notwithstanding any other provision of this License, if you provide any user
with an opportunity to interact with the covered work through a computer
network, an imitation of such network, or any number of "Proxy Programs",
you must prominently offer that user an opportunity to receive the
Corresponding Source of the covered work and all Proxy Programs from a
network server at no charge, through some standard or customary means of

View File

@@ -1,6 +1,6 @@
#!/bin/bash
gcc -E -o fio_headers.i fio_headers.h
gcc -I. -E -o fio_headers.i src/fio_headers.h
rm -rf fio-copy
for i in `grep -Po 'fio/[^"]+' fio_headers.i | sort | uniq`; do

View File

@@ -5,7 +5,7 @@
#cd b/qemu; make qapi
gcc -I qemu/b/qemu `pkg-config glib-2.0 --cflags` \
-I qemu/include -E -o qemu_driver.i qemu_driver.c
-I qemu/include -E -o qemu_driver.i src/qemu_driver.c
rm -rf qemu-copy
for i in `grep -Po 'qemu/[^"]+' qemu_driver.i | sort | uniq`; do

7
debian/build-vitastor-bullseye.sh vendored Executable file
View File

@@ -0,0 +1,7 @@
#!/bin/bash
sed 's/$REL/bullseye/' < vitastor.Dockerfile > ../Dockerfile
cd ..
mkdir -p packages
sudo podman build -v `pwd`/packages:/root/packages -f Dockerfile .
rm Dockerfile

7
debian/build-vitastor-buster.sh vendored Executable file
View File

@@ -0,0 +1,7 @@
#!/bin/bash
sed 's/$REL/buster/' < vitastor.Dockerfile > ../Dockerfile
cd ..
mkdir -p packages
sudo podman build -v `pwd`/packages:/root/packages -f Dockerfile .
rm Dockerfile

2
debian/changelog vendored
View File

@@ -1,4 +1,4 @@
vitastor (0.5.3-1) unstable; urgency=medium
vitastor (0.5.5-1) unstable; urgency=medium
* Bugfixes

13
debian/copyright vendored
View File

@@ -5,16 +5,17 @@ Source: https://vitastor.io
Files: *
Copyright: 2019+ Vitaliy Filippov <vitalif@yourcmc.ru>
License: Multiple licenses VNPL-1.0 and/or GPL-2.0+
License: Multiple licenses VNPL-1.1 and/or GPL-2.0+
All server-side code (OSD, Monitor and so on) is licensed under the terms of
Vitastor Network Public License 1.0 (VNPL 1.0), a copyleft license based on
Vitastor Network Public License 1.1 (VNPL 1.1), a copyleft license based on
GNU GPLv3.0 with the additional "Network Interaction" clause which requires
opensourcing all programs directly or indirectly interacting with Vitastor
through a computer network ("Proxy Programs"). Proxy Programs may be made public
not only under the terms of the same license, but also under the terms of any
GPL-Compatible Free Software License, as listed by the Free Software Foundation.
through a computer network and expressly designed to be used in conjunction
with it ("Proxy Programs"). Proxy Programs may be made public not only under
the terms of the same license, but also under the terms of any GPL-Compatible
Free Software License, as listed by the Free Software Foundation.
This is a stricter copyleft license than the Affero GPL.
.
Client libraries (cluster_client and so on) are dual-licensed under the same
VNPL 1.0 and also GNU GPL 2.0 or later to allow for compatibility with GPLed
VNPL 1.1 and also GNU GPL 2.0 or later to allow for compatibility with GPLed
software like QEMU and fio.

2
debian/install vendored
View File

@@ -1,3 +1,3 @@
VNPL-1.0.txt usr/share/doc/vitastor
VNPL-1.1.txt usr/share/doc/vitastor
GPL-2.0.txt usr/share/doc/vitastor
mon usr/lib/vitastor

View File

@@ -1,13 +1,8 @@
# Build patched QEMU for Debian Buster or Bullseye/Sid inside a container
# cd ..; podman build --build-arg REL=bullseye -v `pwd`/build:/root/build -f debian/patched-qemu.Dockerfile .
ARG REL=bullseye
# cd ..; podman build --build-arg REL=bullseye -v `pwd`/packages:/root/packages -f debian/patched-qemu.Dockerfile .
FROM debian:$REL
# again, it doesn't work otherwise
ARG REL=bullseye
WORKDIR /root
RUN if [ "$REL" = "buster" ]; then \
@@ -30,20 +25,20 @@ RUN apt-get --download-only source fio
ADD qemu-5.0-vitastor.patch qemu-5.1-vitastor.patch /root/vitastor/
RUN set -e; \
mkdir -p /root/build/qemu-$REL; \
rm -rf /root/build/qemu-$REL/*; \
cd /root/build/qemu-$REL; \
mkdir -p /root/packages/qemu-$REL; \
rm -rf /root/packages/qemu-$REL/*; \
cd /root/packages/qemu-$REL; \
dpkg-source -x /root/qemu*.dsc; \
if [ -d /root/build/qemu-$REL/qemu-5.0 ]; then \
cp /root/vitastor/qemu-5.0-vitastor.patch /root/build/qemu-$REL/qemu-5.0/debian/patches; \
echo qemu-5.0-vitastor.patch >> /root/build/qemu-$REL/qemu-5.0/debian/patches/series; \
if [ -d /root/packages/qemu-$REL/qemu-5.0 ]; then \
cp /root/vitastor/qemu-5.0-vitastor.patch /root/packages/qemu-$REL/qemu-5.0/debian/patches; \
echo qemu-5.0-vitastor.patch >> /root/packages/qemu-$REL/qemu-5.0/debian/patches/series; \
else \
cp /root/vitastor/qemu-5.1-vitastor.patch /root/build/qemu-$REL/qemu-*/debian/patches; \
P=`ls -d /root/build/qemu-$REL/qemu-*/debian/patches`; \
cp /root/vitastor/qemu-5.1-vitastor.patch /root/packages/qemu-$REL/qemu-*/debian/patches; \
P=`ls -d /root/packages/qemu-$REL/qemu-*/debian/patches`; \
echo qemu-5.1-vitastor.patch >> $P/series; \
fi; \
cd /root/build/qemu-$REL/qemu-*/; \
cd /root/packages/qemu-$REL/qemu-*/; \
V=$(head -n1 debian/changelog | perl -pe 's/^.*\((.*?)(~bpo[\d\+]*)?\).*$/$1/')+vitastor1; \
DEBFULLNAME="Vitaliy Filippov <vitalif@yourcmc.ru>" dch -D $REL -v $V 'Plug Vitastor block driver'; \
DEB_BUILD_OPTIONS=nocheck dpkg-buildpackage --jobs=auto -sa; \
rm -rf /root/build/qemu-$REL/qemu-*/
rm -rf /root/packages/qemu-$REL/qemu-*/

View File

@@ -1,13 +1,8 @@
# Build Vitastor packages for Debian Buster or Bullseye/Sid inside a container
# cd ..; podman build --build-arg REL=bullseye -v `pwd`/build:/root/build -f debian/vitastor.Dockerfile .
ARG REL=bullseye
# cd ..; podman build --build-arg REL=bullseye -v `pwd`/packages:/root/packages -f debian/vitastor.Dockerfile .
FROM debian:$REL
# again, it doesn't work otherwise
ARG REL=bullseye
WORKDIR /root
RUN if [ "$REL" = "buster" ]; then \
@@ -27,7 +22,7 @@ RUN apt-get -y build-dep qemu
RUN apt-get -y build-dep fio
RUN apt-get --download-only source qemu
RUN apt-get --download-only source fio
RUN apt-get -y install libjerasure-dev
RUN apt-get -y install libjerasure-dev cmake
ADD . /root/vitastor
RUN set -e -x; \
@@ -35,20 +30,20 @@ RUN set -e -x; \
cd /root/fio-build/; \
rm -rf /root/fio-build/*; \
dpkg-source -x /root/fio*.dsc; \
cd /root/build/qemu-$REL/; \
cd /root/packages/qemu-$REL/; \
rm -rf qemu*/; \
dpkg-source -x qemu*.dsc; \
cd /root/build/qemu-$REL/qemu*/; \
cd /root/packages/qemu-$REL/qemu*/; \
debian/rules b/configure-stamp; \
cd b/qemu; \
make -j8 qapi; \
mkdir -p /root/build/vitastor-$REL; \
rm -rf /root/build/vitastor-$REL/*; \
cd /root/build/vitastor-$REL; \
cp -r /root/vitastor vitastor-0.5.3; \
ln -s /root/build/qemu-$REL/qemu-*/ vitastor-0.5.3/qemu; \
ln -s /root/fio-build/fio-*/ vitastor-0.5.3/fio; \
cd vitastor-0.5.3; \
make -j8 qapi/qapi-builtin-types.h; \
mkdir -p /root/packages/vitastor-$REL; \
rm -rf /root/packages/vitastor-$REL/*; \
cd /root/packages/vitastor-$REL; \
cp -r /root/vitastor vitastor-0.5.5; \
ln -s /root/packages/qemu-$REL/qemu-*/ vitastor-0.5.5/qemu; \
ln -s /root/fio-build/fio-*/ vitastor-0.5.5/fio; \
cd vitastor-0.5.5; \
FIO=$(head -n1 fio/debian/changelog | perl -pe 's/^.*\((.*?)\).*$/$1/'); \
QEMU=$(head -n1 qemu/debian/changelog | perl -pe 's/^.*\((.*?)\).*$/$1/'); \
sh copy-qemu-includes.sh; \
@@ -60,13 +55,13 @@ RUN set -e -x; \
diff -NaurpbB a b > debian/patches/qemu-fio-headers.patch || true; \
echo qemu-fio-headers.patch >> debian/patches/series; \
rm -rf a b; \
rm -rf /root/build/qemu-$REL/qemu*/; \
rm -rf /root/packages/qemu-$REL/qemu*/; \
echo "dep:fio=$FIO" > debian/substvars; \
echo "dep:qemu=$QEMU" >> debian/substvars; \
cd /root/build/vitastor-$REL; \
tar --sort=name --mtime='2020-01-01' --owner=0 --group=0 --exclude=debian -cJf vitastor_0.5.3.orig.tar.xz vitastor-0.5.3; \
cd vitastor-0.5.3; \
cd /root/packages/vitastor-$REL; \
tar --sort=name --mtime='2020-01-01' --owner=0 --group=0 --exclude=debian -cJf vitastor_0.5.5.orig.tar.xz vitastor-0.5.5; \
cd vitastor-0.5.5; \
V=$(head -n1 debian/changelog | perl -pe 's/^.*\((.*?)\).*$/$1/'); \
DEBFULLNAME="Vitaliy Filippov <vitalif@yourcmc.ru>" dch -D $REL -v "$V""$REL" "Rebuild for $REL"; \
DEB_BUILD_OPTIONS=nocheck dpkg-buildpackage --jobs=auto -sa; \
rm -rf /root/build/vitastor-$REL/vitastor-*/
rm -rf /root/packages/vitastor-$REL/vitastor-*/

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include <iostream>
#include <functional>

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
module.exports = {
scale_pg_count,
@@ -8,7 +8,7 @@ module.exports = {
function scale_pg_count(prev_pgs, prev_pg_history, new_pg_history, new_pg_count)
{
const old_pg_count = prev_pgs.length;
// Add all possibly intersecting PGs into the history of new PGs
// Add all possibly intersecting PGs to the history of new PGs
if (!(new_pg_count % old_pg_count))
{
// New PG count is a multiple of the old PG count
@@ -16,7 +16,7 @@ function scale_pg_count(prev_pgs, prev_pg_history, new_pg_history, new_pg_count)
for (let i = 0; i < new_pg_count; i++)
{
const old_i = Math.floor(new_pg_count / mul);
new_pg_history[i] = JSON.parse(JSON.stringify(prev_pg_history[1+old_i]));
new_pg_history[i] = prev_pg_history[old_i] ? JSON.parse(JSON.stringify(prev_pg_history[old_i])) : undefined;
}
}
else if (!(old_pg_count % new_pg_count))

View File

@@ -1,31 +1,16 @@
// Functions to calculate Annualized Failure Rate of your cluster
// if you know AFR of your drives, number of drives, expected rebalance time
// and replication factor
// License: VNPL-1.0 (see README.md for details)
const { sprintf } = require('sprintf-js');
// License: VNPL-1.1 (see https://yourcmc.ru/git/vitalif/vitastor/src/branch/master/README.md for details) or AGPL-3.0
// Author: Vitaliy Filippov, 2020+
module.exports = {
cluster_afr_fullmesh,
failure_rate_fullmesh,
cluster_afr,
print_cluster_afr,
c_n_k,
};
print_cluster_afr({ n_hosts: 4, n_drives: 6, afr_drive: 0.03, afr_host: 0.05, capacity: 4000, speed: 0.1, replicas: 2 });
print_cluster_afr({ n_hosts: 4, n_drives: 3, afr_drive: 0.03, capacity: 4000, speed: 0.1, replicas: 2 });
print_cluster_afr({ n_hosts: 4, n_drives: 3, afr_drive: 0.03, afr_host: 0.05, capacity: 4000, speed: 0.1, replicas: 2 });
print_cluster_afr({ n_hosts: 4, n_drives: 3, afr_drive: 0.03, capacity: 4000, speed: 0.1, ec: [ 2, 1 ] });
print_cluster_afr({ n_hosts: 4, n_drives: 3, afr_drive: 0.03, afr_host: 0.05, capacity: 4000, speed: 0.1, ec: [ 2, 1 ] });
print_cluster_afr({ n_hosts: 10, n_drives: 10, afr_drive: 0.1, capacity: 8000, speed: 0.02, replicas: 2 });
print_cluster_afr({ n_hosts: 10, n_drives: 10, afr_drive: 0.1, afr_host: 0.05, capacity: 8000, speed: 0.02, replicas: 2 });
print_cluster_afr({ n_hosts: 10, n_drives: 10, afr_drive: 0.1, capacity: 8000, speed: 0.02, replicas: 3 });
print_cluster_afr({ n_hosts: 10, n_drives: 10, afr_drive: 0.1, afr_host: 0.05, capacity: 8000, speed: 0.02, replicas: 3 });
print_cluster_afr({ n_hosts: 10, n_drives: 10, afr_drive: 0.1, capacity: 8000, speed: 0.02, replicas: 3, pgs: 100 });
print_cluster_afr({ n_hosts: 10, n_drives: 10, afr_drive: 0.1, afr_host: 0.05, capacity: 8000, speed: 0.02, replicas: 3, pgs: 100 });
print_cluster_afr({ n_hosts: 10, n_drives: 10, afr_drive: 0.1, afr_host: 0.05, capacity: 8000, speed: 0.02, replicas: 3, pgs: 100, degraded_replacement: 1 });
/******** "FULL MESH": ASSUME EACH OSD COMMUNICATES WITH ALL OTHER OSDS ********/
// Estimate AFR of the cluster
@@ -56,93 +41,38 @@ function failure_rate_fullmesh(n, a, f)
/******** PGS: EACH OSD ONLY COMMUNICATES WITH <pgs> OTHER OSDs ********/
// <n> hosts of <m> drives of <capacity> GB, each able to backfill at <speed> GB/s,
// <k> replicas, <pgs> unique peer PGs per OSD
// <k> replicas, <pgs> unique peer PGs per OSD (~50 for 100 PG-per-OSD in a big cluster)
//
// For each of n*m drives: P(drive fails in a year) * P(any of its peers fail in <l*365> next days).
// More peers per OSD increase rebalance speed (more drives work together to resilver) if you
// let them finish rebalance BEFORE replacing the failed drive.
// let them finish rebalance BEFORE replacing the failed drive (degraded_replacement=false).
// At the same time, more peers per OSD increase probability of any of them to fail!
// osd_rm=true means that failed OSDs' data is rebalanced over all other hosts,
// not over the same host as it's in Ceph by default (dead OSDs are marked 'out').
//
// Probability of all except one drives in a replica group to fail is (AFR^(k-1)).
// So with <x> PGs it becomes ~ (x * (AFR*L/365)^(k-1)). Interesting but reasonable consequence
// is that, with k=2, total failure rate doesn't depend on number of peers per OSD,
// because it gets increased linearly by increased number of peers to fail
// and decreased linearly by reduced rebalance time.
function cluster_afr_pgs({ n_hosts, n_drives, afr_drive, capacity, speed, replicas, pgs = 1, degraded_replacement })
function cluster_afr({ n_hosts, n_drives, afr_drive, afr_host, capacity, speed, ec, ec_data, ec_parity, replicas, pgs = 1, osd_rm, degraded_replacement, down_out_interval = 600 })
{
pgs = Math.min(pgs, (n_hosts-1)*n_drives/(replicas-1));
const l = capacity/(degraded_replacement ? 1 : pgs)/speed/86400/365;
return 1 - (1 - afr_drive * (1-(1-(afr_drive*l)**(replicas-1))**pgs)) ** (n_hosts*n_drives);
}
function cluster_afr_pgs_ec({ n_hosts, n_drives, afr_drive, capacity, speed, ec: [ ec_data, ec_parity ], pgs = 1, degraded_replacement })
{
const ec_total = ec_data+ec_parity;
pgs = Math.min(pgs, (n_hosts-1)*n_drives/(ec_total-1));
const l = capacity/(degraded_replacement ? 1 : pgs)/speed/86400/365;
return 1 - (1 - afr_drive * (1-(1-failure_rate_fullmesh(ec_total-1, afr_drive*l, ec_parity))**pgs)) ** (n_hosts*n_drives);
}
// Same as above, but also take server failures into account
function cluster_afr_pgs_hosts({ n_hosts, n_drives, afr_drive, afr_host, capacity, speed, replicas, pgs = 1, degraded_replacement })
{
let otherhosts = Math.min(pgs, (n_hosts-1)/(replicas-1));
pgs = Math.min(pgs, (n_hosts-1)*n_drives/(replicas-1));
let pgh = Math.min(pgs*n_drives, (n_hosts-1)*n_drives/(replicas-1));
const ld = capacity/(degraded_replacement ? 1 : pgs)/speed/86400/365;
const lh = n_drives*capacity/pgs/speed/86400/365;
const p1 = ((afr_drive+afr_host*pgs/otherhosts)*lh);
const p2 = ((afr_drive+afr_host*pgs/otherhosts)*ld);
return 1 - ((1 - afr_host * (1-(1-p1**(replicas-1))**pgh)) ** n_hosts) *
((1 - afr_drive * (1-(1-p2**(replicas-1))**pgs)) ** (n_hosts*n_drives));
}
function cluster_afr_pgs_ec_hosts({ n_hosts, n_drives, afr_drive, afr_host, capacity, speed, ec: [ ec_data, ec_parity ], pgs = 1, degraded_replacement })
{
const ec_total = ec_data+ec_parity;
const otherhosts = Math.min(pgs, (n_hosts-1)/(ec_total-1));
pgs = Math.min(pgs, (n_hosts-1)*n_drives/(ec_total-1));
const pgh = Math.min(pgs*n_drives, (n_hosts-1)*n_drives/(ec_total-1));
const ld = capacity/(degraded_replacement ? 1 : pgs)/speed/86400/365;
const lh = n_drives*capacity/pgs/speed/86400/365;
const p1 = ((afr_drive+afr_host*pgs/otherhosts)*lh);
const p2 = ((afr_drive+afr_host*pgs/otherhosts)*ld);
return 1 - ((1 - afr_host * (1-(1-failure_rate_fullmesh(ec_total-1, p1, ec_parity))**pgh)) ** n_hosts) *
((1 - afr_drive * (1-(1-failure_rate_fullmesh(ec_total-1, p2, ec_parity))**pgs)) ** (n_hosts*n_drives));
}
// Wrapper for 4 above functions
function cluster_afr(config)
{
if (config.ec && config.afr_host)
{
return cluster_afr_pgs_ec_hosts(config);
}
else if (config.ec)
{
return cluster_afr_pgs_ec(config);
}
else if (config.afr_host)
{
return cluster_afr_pgs_hosts(config);
}
else
{
return cluster_afr_pgs(config);
}
}
function print_cluster_afr(config)
{
console.log(
`${config.n_hosts} nodes with ${config.n_drives} ${sprintf("%.1f", config.capacity/1000)}TB drives`+
`, capable to backfill at ${sprintf("%.1f", config.speed*1000)} MB/s, drive AFR ${sprintf("%.1f", config.afr_drive*100)}%`+
(config.afr_host ? `, host AFR ${sprintf("%.1f", config.afr_host*100)}%` : '')+
(config.ec ? `, EC ${config.ec[0]}+${config.ec[1]}` : `, ${config.replicas} replicas`)+
`, ${config.pgs||1} PG per OSD`+
(config.degraded_replacement ? `\n...and you don't let the rebalance finish before replacing drives` : '')
);
console.log('-> '+sprintf("%.7f%%", 100*cluster_afr(config))+'\n');
const pg_size = (ec ? ec_data+ec_parity : replicas);
pgs = Math.min(pgs, (n_hosts-1)*n_drives/(pg_size-1));
const host_pgs = Math.min(pgs*n_drives, (n_hosts-1)*n_drives/(pg_size-1));
const resilver_disk = n_drives == 1 || osd_rm ? pgs : (n_drives-1);
const disk_heal_time = (down_out_interval + capacity/(degraded_replacement ? 1 : resilver_disk)/speed)/86400/365;
const host_heal_time = (down_out_interval + n_drives*capacity/pgs/speed)/86400/365;
const disk_heal_fail = ((afr_drive+afr_host/n_drives)*disk_heal_time);
const host_heal_fail = ((afr_drive+afr_host/n_drives)*host_heal_time);
const disk_pg_fail = ec
? failure_rate_fullmesh(ec_data+ec_parity-1, disk_heal_fail, ec_parity)
: disk_heal_fail**(replicas-1);
const host_pg_fail = ec
? failure_rate_fullmesh(ec_data+ec_parity-1, host_heal_fail, ec_parity)
: host_heal_fail**(replicas-1);
return 1 - ((1 - afr_drive * (1-(1-disk_pg_fail)**pgs)) ** (n_hosts*n_drives))
* ((1 - afr_host * (1-(1-host_pg_fail)**host_pgs)) ** n_hosts);
}
/******** UTILITY ********/

28
mon/afr_test.js Normal file
View File

@@ -0,0 +1,28 @@
const { sprintf } = require('sprintf-js');
const { cluster_afr } = require('./afr.js');
print_cluster_afr({ n_hosts: 4, n_drives: 6, afr_drive: 0.03, afr_host: 0.05, capacity: 4000, speed: 0.1, replicas: 2 });
print_cluster_afr({ n_hosts: 4, n_drives: 3, afr_drive: 0.03, afr_host: 0, capacity: 4000, speed: 0.1, replicas: 2 });
print_cluster_afr({ n_hosts: 4, n_drives: 3, afr_drive: 0.03, afr_host: 0.05, capacity: 4000, speed: 0.1, replicas: 2 });
print_cluster_afr({ n_hosts: 4, n_drives: 3, afr_drive: 0.03, afr_host: 0, capacity: 4000, speed: 0.1, ec: true, ec_data: 2, ec_parity: 1 });
print_cluster_afr({ n_hosts: 4, n_drives: 3, afr_drive: 0.03, afr_host: 0.05, capacity: 4000, speed: 0.1, ec: true, ec_data: 2, ec_parity: 1 });
print_cluster_afr({ n_hosts: 10, n_drives: 10, afr_drive: 0.1, afr_host: 0, capacity: 8000, speed: 0.02, replicas: 2 });
print_cluster_afr({ n_hosts: 10, n_drives: 10, afr_drive: 0.1, afr_host: 0.05, capacity: 8000, speed: 0.02, replicas: 2 });
print_cluster_afr({ n_hosts: 10, n_drives: 10, afr_drive: 0.1, afr_host: 0, capacity: 8000, speed: 0.02, replicas: 3 });
print_cluster_afr({ n_hosts: 10, n_drives: 10, afr_drive: 0.1, afr_host: 0.05, capacity: 8000, speed: 0.02, replicas: 3 });
print_cluster_afr({ n_hosts: 10, n_drives: 10, afr_drive: 0.1, afr_host: 0, capacity: 8000, speed: 0.02, replicas: 3, pgs: 100 });
print_cluster_afr({ n_hosts: 10, n_drives: 10, afr_drive: 0.1, afr_host: 0.05, capacity: 8000, speed: 0.02, replicas: 3, pgs: 100 });
print_cluster_afr({ n_hosts: 10, n_drives: 10, afr_drive: 0.1, afr_host: 0.05, capacity: 8000, speed: 0.02, replicas: 3, pgs: 100, degraded_replacement: 1 });
function print_cluster_afr(config)
{
console.log(
`${config.n_hosts} nodes with ${config.n_drives} ${sprintf("%.1f", config.capacity/1000)}TB drives`+
`, capable to backfill at ${sprintf("%.1f", config.speed*1000)} MB/s, drive AFR ${sprintf("%.1f", config.afr_drive*100)}%`+
(config.afr_host ? `, host AFR ${sprintf("%.1f", config.afr_host*100)}%` : '')+
(config.ec ? `, EC ${config.ec_data}+${config.ec_parity}` : `, ${config.replicas} replicas`)+
`, ${config.pgs||1} PG per OSD`+
(config.degraded_replacement ? `\n...and you don't let the rebalance finish before replacing drives` : '')
);
console.log('-> '+sprintf("%.7f%%", 100*cluster_afr(config))+'\n');
}

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
// Data distribution optimizer using linear programming (lp_solve)
@@ -58,7 +58,7 @@ async function optimize_initial({ osd_tree, pg_count, pg_size = 3, pg_minsize =
}
const all_weights = Object.assign({}, ...Object.values(osd_tree));
const total_weight = Object.values(all_weights).reduce((a, c) => Number(a) + Number(c), 0);
const all_pgs = Object.values(random_combinations(osd_tree, pg_size, max_combinations));
const all_pgs = Object.values(random_combinations(osd_tree, pg_size, max_combinations, parity_space > 1));
const pg_per_osd = {};
for (const pg of all_pgs)
{
@@ -249,7 +249,7 @@ async function optimize_change({ prev_pgs: prev_int_pgs, osd_tree, pg_size = 3,
}
}
// Get all combinations
let all_pgs = random_combinations(osd_tree, pg_size, max_combinations);
let all_pgs = random_combinations(osd_tree, pg_size, max_combinations, parity_space > 1);
add_valid_previous(osd_tree, prev_weights, all_pgs);
all_pgs = Object.values(all_pgs);
const pg_per_osd = {};
@@ -275,6 +275,11 @@ async function optimize_change({ prev_pgs: prev_int_pgs, osd_tree, pg_size = 3,
lp += 'max: '+all_pg_names.map(pg_name => (
prev_weights[pg_name] ? `${pg_size+1}*add_${pg_name} - ${pg_size+1}*del_${pg_name}` : `${pg_size+1-move_weights[pg_name]}*${pg_name}`
)).join(' + ')+';\n';
lp += all_pg_names
.map(pg_name => (prev_weights[pg_name] ? `add_${pg_name} - del_${pg_name}` : `${pg_name}`))
.join(' + ')+' = '+(pg_count
- Object.keys(prev_weights).reduce((a, old_pg_name) => (a + (all_pgs_hash[old_pg_name] ? prev_weights[old_pg_name] : 0)), 0)
)+';\n';
for (const osd in pg_per_osd)
{
if (osd !== NO_OSD)
@@ -488,7 +493,8 @@ function extract_osds(osd_tree, levels, osd_level, osds = {})
return osds;
}
function random_combinations(osd_tree, pg_size, count)
// ordered = don't treat (x,y) and (y,x) as equal
function random_combinations(osd_tree, pg_size, count, ordered)
{
let seed = 0x5f020e43;
let rng = () =>
@@ -516,25 +522,47 @@ function random_combinations(osd_tree, pg_size, count)
pg.push(osds[cur_hosts[next_host]][next_osd]);
cur_hosts.splice(next_host, 1);
}
while (pg.length < pg_size)
const cyclic_pgs = [ pg ];
if (ordered)
{
pg.push(NO_OSD);
for (let i = 1; i < pg.size; i++)
{
cyclic_pgs.push([ ...pg.slice(i), ...pg.slice(0, i) ]);
}
}
for (const pg of cyclic_pgs)
{
while (pg.length < pg_size)
{
pg.push(NO_OSD);
}
r['pg_'+pg.join('_')] = pg;
}
r['pg_'+pg.join('_')] = pg;
}
}
// Generate purely random combinations
restart: while (count > 0)
while (count > 0)
{
let host_idx = [];
for (let i = 0; i < pg_size && i < hosts.length; i++)
const cur_hosts = [ ...hosts.map((h, i) => i) ];
const max_hosts = pg_size < hosts.length ? pg_size : hosts.length;
if (ordered)
{
let start = i > 0 ? host_idx[i-1]+1 : 0;
if (start >= hosts.length)
for (let i = 0; i < max_hosts; i++)
{
continue restart;
const r = rng() % cur_hosts.length;
host_idx[i] = cur_hosts[r];
cur_hosts.splice(r, 1);
}
}
else
{
for (let i = 0; i < max_hosts; i++)
{
const r = rng() % (cur_hosts.length - (max_hosts - i - 1));
host_idx[i] = cur_hosts[r];
cur_hosts.splice(0, r+1);
}
host_idx[i] = start + rng() % (hosts.length-start);
}
let pg = host_idx.map(h => osds[hosts[h]][rng() % osds[hosts[h]].length]);
while (pg.length < pg_size)

76
mon/make-osd.sh Executable file
View File

@@ -0,0 +1,76 @@
#!/bin/bash
# Very simple systemd unit generator for vitastor-osd services
# Not the final solution yet, mostly for tests
# Copyright (c) Vitaliy Filippov, 2019+
# License: MIT
# USAGE: ./make-osd.sh /dev/disk/by-partuuid/xxx [ /dev/disk/by-partuuid/yyy]...
IP_SUBSTR="10.200.1."
ETCD_HOSTS="etcd0=http://10.200.1.10:2380,etcd1=http://10.200.1.11:2380,etcd2=http://10.200.1.12:2380"
set -e -x
IP=`ip -json a s | jq -r '.[].addr_info[] | select(.local | startswith("'$IP_SUBSTR'")) | .local'`
[ "$IP" != "" ] || exit 1
ETCD_MON=$(echo $ETCD_HOSTS | perl -pe 's/:2380/:2379/g; s/etcd\d*=//g;')
D=`dirname $0`
# Create OSDs on all passed devices
OSD_NUM=1
for DEV in $*; do
# Ugly :) -> node.js rework pending
while true; do
ST=$(etcdctl --endpoints="$ETCD_MON" get --print-value-only /vitastor/osd/stats/$OSD_NUM)
if [ "$ST" = "" ]; then
break
fi
OSD_NUM=$((OSD_NUM+1))
done
etcdctl --endpoints="$ETCD_MON" put /vitastor/osd/stats/$OSD_NUM '{}'
echo Creating OSD $OSD_NUM on $DEV
OPT=`node $D/simple-offsets.js --device $DEV --format options | tr '\n' ' '`
META=`echo $OPT | grep -Po '(?<=data_offset )\d+'`
dd if=/dev/zero of=$DEV bs=1048576 count=$(((META+1048575)/1048576)) oflag=direct
cat >/etc/systemd/system/vitastor-osd$OSD_NUM.service <<EOF
[Unit]
Description=Vitastor object storage daemon osd.$OSD_NUM
After=network-online.target local-fs.target time-sync.target
Wants=network-online.target local-fs.target time-sync.target
PartOf=vitastor.target
[Service]
LimitNOFILE=1048576
LimitNPROC=1048576
LimitMEMLOCK=infinity
ExecStart=/usr/bin/vitastor-osd \\
--etcd_address $IP:2379/v3 \\
--bind_address $IP \\
--osd_num $OSD_NUM \\
--disable_data_fsync 1 \\
--immediate_commit all \\
--flusher_count 256 \\
--disk_alignment 4096 --journal_block_size 4096 --meta_block_size 4096 \\
--journal_no_same_sector_overwrites true \\
--journal_sector_buffer_count 1024 \\
$OPT
WorkingDirectory=/
ExecStartPre=+chown vitastor:vitastor $DEV
User=vitastor
PrivateTmp=false
TasksMax=infinity
Restart=always
StartLimitInterval=0
RestartSec=10
[Install]
WantedBy=vitastor.target
EOF
systemctl enable vitastor-osd$OSD_NUM
done

138
mon/make-units.sh Normal file → Executable file
View File

@@ -1,19 +1,25 @@
#!/bin/bash
# Example startup script generator
# Of course this isn't a production solution yet, this is just for tests
# Very simple systemd unit generator for etcd & vitastor-mon services
# Not the final solution yet, mostly for tests
# Copyright (c) Vitaliy Filippov, 2019+
# License: MIT
IP=`ip -json a s | jq -r '.[].addr_info[] | select(.broadcast == "10.115.0.255") | .local'`
# USAGE: ./make-units.sh
IP_SUBSTR="10.200.1."
ETCD_HOSTS="etcd0=http://10.200.1.10:2380,etcd1=http://10.200.1.11:2380,etcd2=http://10.200.1.12:2380"
# determine IP
IP=`ip -json a s | jq -r '.[].addr_info[] | select(.local | startswith("'$IP_SUBSTR'")) | .local'`
[ "$IP" != "" ] || exit 1
ETCD_NUM=${ETCD_HOSTS/$IP*/}
[ "$ETCD_NUM" != "$ETCD_HOSTS" ] || exit 1
ETCD_NUM=$(echo $ETCD_NUM | tr -d -c , | wc -c)
BASE=${IP/*./}
BASE=$((BASE-10))
# etcd
useradd etcd
mkdir -p /var/lib/etcd$BASE.etcd
mkdir -p /var/lib/etcd$ETCD_NUM.etcd
cat >/etc/systemd/system/etcd.service <<EOF
[Unit]
Description=etcd for vitastor
@@ -22,19 +28,18 @@ Wants=network-online.target local-fs.target time-sync.target
[Service]
Restart=always
ExecStart=/usr/local/bin/etcd -name etcd$BASE --data-dir /var/lib/etcd$BASE.etcd \\
ExecStart=/usr/local/bin/etcd -name etcd$ETCD_NUM --data-dir /var/lib/etcd$ETCD_NUM.etcd \\
--advertise-client-urls http://$IP:2379 --listen-client-urls http://$IP:2379 \\
--initial-advertise-peer-urls http://$IP:2380 --listen-peer-urls http://$IP:2380 \\
--initial-cluster-token vitastor-etcd-1 --initial-cluster etcd0=http://10.115.0.10:2380,etcd1=http://10.115.0.11:2380,etcd2=http://10.115.0.12:2380,etcd3=http://10.115.0.13:2380 \\
--initial-cluster-token vitastor-etcd-1 --initial-cluster $ETCD_HOSTS \\
--initial-cluster-state new --max-txn-ops=100000 --auto-compaction-retention=10 --auto-compaction-mode=revision
WorkingDirectory=/var/lib/etcd$BASE.etcd
ExecStartPre=+chown -R etcd /var/lib/etcd$BASE.etcd
WorkingDirectory=/var/lib/etcd$ETCD_NUM.etcd
ExecStartPre=+chown -R etcd /var/lib/etcd$ETCD_NUM.etcd
User=etcd
PrivateTmp=false
TasksMax=infinity
Restart=always
StartLimitInterval=0
StartLimitIntervalSec=0
RestartSec=10
[Install]
@@ -48,9 +53,7 @@ systemctl start etcd
useradd vitastor
chmod 755 /root
BASE=${IP/*./}
BASE=$(((BASE-10)*12))
# Vitastor target
cat >/etc/systemd/system/vitastor.target <<EOF
[Unit]
Description=vitastor target
@@ -58,116 +61,25 @@ Description=vitastor target
WantedBy=multi-user.target
EOF
i=1
for DEV in `ls /dev/disk/by-id/ | grep ata-INTEL_SSDSC2KB`; do
dd if=/dev/zero of=/dev/disk/by-id/$DEV bs=1048576 count=$(((427814912+1048575)/1048576+2))
dd if=/dev/zero of=/dev/disk/by-id/$DEV bs=1048576 count=$(((427814912+1048575)/1048576+2)) seek=$((1920377991168/1048576))
cat >/etc/systemd/system/vitastor-osd$((BASE+i)).service <<EOF
# Monitor unit
ETCD_MON=$(echo $ETCD_HOSTS | perl -pe 's/:2380/:2379/g; s/etcd\d*=//g;')
cat >/etc/systemd/system/vitastor-mon.service <<EOF
[Unit]
Description=Vitastor object storage daemon osd.$((BASE+i))
Description=Vitastor monitor
After=network-online.target local-fs.target time-sync.target
Wants=network-online.target local-fs.target time-sync.target
PartOf=vitastor.target
[Service]
LimitNOFILE=1048576
LimitNPROC=1048576
LimitMEMLOCK=infinity
ExecStart=/root/vitastor/osd \\
--etcd_address $IP:2379/v3 \\
--bind_address $IP \\
--osd_num $((BASE+i)) \\
--disable_data_fsync 1 \\
--disable_device_lock 1 \\
--immediate_commit all \\
--flusher_count 8 \\
--disk_alignment 4096 --journal_block_size 4096 --meta_block_size 4096 \\
--journal_no_same_sector_overwrites true \\
--journal_sector_buffer_count 1024 \\
--journal_offset 0 \\
--meta_offset 16777216 \\
--data_offset 427814912 \\
--data_size $((1920377991168-427814912)) \\
--data_device /dev/disk/by-id/$DEV
WorkingDirectory=/root/vitastor
ExecStartPre=+chown vitastor:vitastor /dev/disk/by-id/$DEV
Restart=always
ExecStart=node /usr/lib/vitastor/mon/mon-main.js --etcd_url '$ETCD_MON' --etcd_prefix '/vitastor' --etcd_start_timeout 5
WorkingDirectory=/
User=vitastor
PrivateTmp=false
TasksMax=infinity
Restart=always
StartLimitInterval=0
StartLimitIntervalSec=0
RestartSec=10
[Install]
WantedBy=vitastor.target
EOF
systemctl enable vitastor-osd$((BASE+i))
i=$((i+1))
cat >/etc/systemd/system/vitastor-osd$((BASE+i)).service <<EOF
[Unit]
Description=Vitastor object storage daemon osd.$((BASE+i))
After=network-online.target local-fs.target time-sync.target
Wants=network-online.target local-fs.target time-sync.target
PartOf=vitastor.target
[Service]
LimitNOFILE=1048576
LimitNPROC=1048576
LimitMEMLOCK=infinity
ExecStart=/root/vitastor/osd \\
--etcd_address $IP:2379/v3 \\
--bind_address $IP \\
--osd_num $((BASE+i)) \\
--disable_data_fsync 1 \\
--immediate_commit all \\
--flusher_count 8 \\
--disk_alignment 4096 --journal_block_size 4096 --meta_block_size 4096 \\
--journal_no_same_sector_overwrites true \\
--journal_sector_buffer_count 1024 \\
--journal_offset 1920377991168 \\
--meta_offset $((1920377991168+16777216)) \\
--data_offset $((1920377991168+427814912)) \\
--data_size $((1920377991168-427814912)) \\
--data_device /dev/disk/by-id/$DEV
WorkingDirectory=/root/vitastor
ExecStartPre=+chown vitastor:vitastor /dev/disk/by-id/$DEV
User=vitastor
PrivateTmp=false
TasksMax=infinity
Restart=always
StartLimitInterval=0
StartLimitIntervalSec=0
RestartSec=10
[Install]
WantedBy=vitastor.target
EOF
systemctl enable vitastor-osd$((BASE+i))
i=$((i+1))
done
exit
node mon-main.js --etcd_url 'http://10.115.0.10:2379,http://10.115.0.11:2379,http://10.115.0.12:2379,http://10.115.0.13:2379' --etcd_prefix '/vitastor' --etcd_start_timeout 5
podman run -d --network host --restart always -v /var/lib/etcd0.etcd:/etcd0.etcd --name etcd quay.io/coreos/etcd:v3.4.13 etcd -name etcd0 \
-advertise-client-urls http://10.115.0.10:2379 -listen-client-urls http://10.115.0.10:2379 \
-initial-advertise-peer-urls http://10.115.0.10:2380 -listen-peer-urls http://10.115.0.10:2380 \
-initial-cluster-token vitastor-etcd-1 -initial-cluster etcd0=http://10.115.0.10:2380,etcd1=http://10.115.0.11:2380,etcd2=http://10.115.0.12:2380,etcd3=http://10.115.0.13:2380 \
-initial-cluster-state new --max-txn-ops=100000 --auto-compaction-retention=10 --auto-compaction-mode=revision
etcdctl --endpoints http://10.115.0.10:2379 put /vitastor/config/global '{"immediate_commit":"all"}'
etcdctl --endpoints http://10.115.0.10:2379 put /vitastor/config/pools '{"1":{"name":"testpool","scheme":"replicated","pg_size":2,"pg_minsize":1,"pg_count":48,"failure_domain":"host"}}'
#let pgs = {};
#for (let n = 0; n < 48; n++) { let i = n/2 | 0; pgs[1+n] = { osd_set: [ (1+i%12+(i/12 | 0)*24), (1+12+i%12+(i/12 | 0)*24) ], primary: (1+(n%2)*12+i%12+(i/12 | 0)*24) }; };
#console.log(JSON.stringify({ items: { 1: pgs } }));
#etcdctl --endpoints http://10.115.0.10:2379 put /vitastor/config/pgs ...
# --disk_alignment 4096 --journal_block_size 4096 --meta_block_size 4096 \\
# --data_offset 427814912 \\
# --disk_alignment 4096 --journal_block_size 512 --meta_block_size 512 \\
# --data_offset 433434624 \\

View File

@@ -1,7 +1,7 @@
#!/usr/bin/node
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
const Mon = require('./mon.js');

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
const http = require('http');
const crypto = require('crypto');
@@ -811,6 +811,17 @@ class Mon
}
PGUtil.scale_pg_count(prev_pgs, this.state.pg.history[pool_id]||{}, pg_history, pool_cfg.pg_count);
}
for (const pg of prev_pgs)
{
while (pg.length < pool_cfg.pg_size)
{
pg.push(0);
}
while (pg.length > pool_cfg.pg_size)
{
pg.pop();
}
}
optimize_result = await LPOptimizer.optimize_change({
prev_pgs,
osd_tree: pool_tree,
@@ -840,7 +851,7 @@ class Mon
this.save_new_pgs_txn(etcd_request, pool_id, up_osds, prev_pgs, optimize_result.int_pgs, pg_history);
}
this.state.config.pgs.hash = tree_hash;
await this.save_pg_config();
await this.save_pg_config(etcd_request);
}
else
{

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
// Interesting real-world example coming from Ceph with EC and compression enabled.
// EC parity chunks can't be compressed as efficiently as data chunks,

View File

@@ -0,0 +1,25 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.1 (see README.md for details)
const LPOptimizer = require('./lp-optimizer.js');
async function run()
{
const osd_tree = { a: { 1: 1 }, b: { 2: 1 }, c: { 3: 1 } };
let res;
console.log('16 PGs, size=3');
res = await LPOptimizer.optimize_initial({ osd_tree, pg_size: 3, pg_count: 16 });
LPOptimizer.print_change_stats(res, false);
console.log('\nReduce PG size to 2');
res = await LPOptimizer.optimize_change({ prev_pgs: res.int_pgs.map(pg => pg.slice(0, 2)), osd_tree, pg_size: 2 });
LPOptimizer.print_change_stats(res, false);
console.log('\nRemove OSD 3');
delete osd_tree['c'];
res = await LPOptimizer.optimize_change({ prev_pgs: res.int_pgs, osd_tree, pg_size: 2 });
LPOptimizer.print_change_stats(res, false);
}
run().catch(console.error);

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
const LPOptimizer = require('./lp-optimizer.js');

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
const LPOptimizer = require('./lp-optimizer.js');

View File

@@ -48,4 +48,4 @@ FIO=`rpm -qi fio | perl -e 'while(<>) { /^Epoch[\s:]+(\S+)/ && print "$1:"; /^Ve
QEMU=`rpm -qi qemu qemu-kvm | perl -e 'while(<>) { /^Epoch[\s:]+(\S+)/ && print "$1:"; /^Version[\s:]+(\S+)/ && print $1; /^Release[\s:]+(\S+)/ && print "-$1"; }'`
perl -i -pe 's/(Requires:\s*fio)([^\n]+)?/$1 = '$FIO'/' $VITASTOR/rpm/vitastor-el$EL.spec
perl -i -pe 's/(Requires:\s*qemu(?:-kvm)?)([^\n]+)?/$1 = '$QEMU'/' $VITASTOR/rpm/vitastor-el$EL.spec
tar --transform 's#^#vitastor-0.5.3/#' --exclude 'rpm/*.rpm' -czf $VITASTOR/../vitastor-0.5.3$(rpm --eval '%dist').tar.gz *
tar --transform 's#^#vitastor-0.5.5/#' --exclude 'rpm/*.rpm' -czf $VITASTOR/../vitastor-0.5.5$(rpm --eval '%dist').tar.gz *

View File

@@ -1,5 +1,5 @@
# Build packages for CentOS 8 inside a container
# cd ..; podman build -t qemu-el8 -v `pwd`/build:/root/build -f rpm/qemu-el8.Dockerfile .
# cd ..; podman build -t qemu-el8 -v `pwd`/packages:/root/packages -f rpm/qemu-el8.Dockerfile .
FROM centos:8
@@ -14,8 +14,8 @@ RUN cd ~/rpmbuild/SPECS && dnf builddep -y --enablerepo=PowerTools --spec qemu-k
ADD qemu-*-vitastor.patch /root/vitastor/
RUN set -e; \
mkdir -p /root/build/qemu-el8; \
rm -rf /root/build/qemu-el8/*; \
mkdir -p /root/packages/qemu-el8; \
rm -rf /root/packages/qemu-el8/*; \
rpm --nomd5 -i /root/qemu*.src.rpm; \
cd ~/rpmbuild/SPECS; \
PN=$(grep ^Patch qemu-kvm.spec | tail -n1 | perl -pe 's/Patch(\d+).*/$1/'); \
@@ -27,5 +27,5 @@ RUN set -e; \
perl -i -pe 's/(^Release:\s*\d+)/$1.vitastor/' qemu-kvm.spec; \
cp /root/vitastor/qemu-4.2-vitastor.patch ~/rpmbuild/SOURCES; \
rpmbuild --nocheck -ba qemu-kvm.spec; \
cp ~/rpmbuild/RPMS/*/*qemu* /root/build/qemu-el8/; \
cp ~/rpmbuild/SRPMS/*qemu* /root/build/qemu-el8/
cp ~/rpmbuild/RPMS/*/*qemu* /root/packages/qemu-el8/; \
cp ~/rpmbuild/SRPMS/*qemu* /root/packages/qemu-el8/

View File

@@ -1,5 +1,5 @@
# Build packages for CentOS 7 inside a container
# cd ..; podman build -t vitastor-el7 -v `pwd`/build:/root/build -f rpm/vitastor-el7.Dockerfile .
# cd ..; podman build -t vitastor-el7 -v `pwd`/packages:/root/packages -f rpm/vitastor-el7.Dockerfile .
# localedef -i ru_RU -f UTF-8 ru_RU.UTF-8
FROM centos:7
@@ -25,23 +25,23 @@ RUN set -e; \
cd ~/rpmbuild/SPECS/; \
. /opt/rh/devtoolset-9/enable; \
rpmbuild -ba liburing.spec; \
mkdir -p /root/build/liburing-el7; \
rm -rf /root/build/liburing-el7/*; \
cp ~/rpmbuild/RPMS/*/liburing* /root/build/liburing-el7/; \
cp ~/rpmbuild/SRPMS/liburing* /root/build/liburing-el7/
mkdir -p /root/packages/liburing-el7; \
rm -rf /root/packages/liburing-el7/*; \
cp ~/rpmbuild/RPMS/*/liburing* /root/packages/liburing-el7/; \
cp ~/rpmbuild/SRPMS/liburing* /root/packages/liburing-el7/
RUN rpm -i `ls /root/build/liburing-el7/liburing-*.x86_64.rpm | grep -v debug`
RUN rpm -i `ls /root/packages/liburing-el7/liburing-*.x86_64.rpm | grep -v debug`
ADD . /root/vitastor
RUN set -e; \
cd /root/vitastor/rpm; \
sh build-tarball.sh; \
cp /root/vitastor-0.5.3.el7.tar.gz ~/rpmbuild/SOURCES; \
cp /root/vitastor-0.5.5.el7.tar.gz ~/rpmbuild/SOURCES; \
cp vitastor-el7.spec ~/rpmbuild/SPECS/vitastor.spec; \
cd ~/rpmbuild/SPECS/; \
rpmbuild -ba vitastor.spec; \
mkdir -p /root/build/vitastor-el7; \
rm -rf /root/build/vitastor-el7/*; \
cp ~/rpmbuild/RPMS/*/vitastor* /root/build/vitastor-el7/; \
cp ~/rpmbuild/SRPMS/vitastor* /root/build/vitastor-el7/
mkdir -p /root/packages/vitastor-el7; \
rm -rf /root/packages/vitastor-el7/*; \
cp ~/rpmbuild/RPMS/*/vitastor* /root/packages/vitastor-el7/; \
cp ~/rpmbuild/SRPMS/vitastor* /root/packages/vitastor-el7/

View File

@@ -1,11 +1,11 @@
Name: vitastor
Version: 0.5.3
Release: 2%{?dist}
Version: 0.5.5
Release: 1%{?dist}
Summary: Vitastor, a fast software-defined clustered block storage
License: Vitastor Network Public License 1.0
License: Vitastor Network Public License 1.1
URL: https://vitastor.io/
Source0: vitastor-0.5.3.el7.tar.gz
Source0: vitastor-0.5.5.el7.tar.gz
BuildRequires: liburing-devel >= 0.6
BuildRequires: gperftools-devel
@@ -14,6 +14,7 @@ BuildRequires: rh-nodejs12
BuildRequires: rh-nodejs12-npm
BuildRequires: jerasure-devel
BuildRequires: gf-complete-devel
BuildRequires: cmake
Requires: fio = 3.7-1.el7
Requires: qemu-kvm = 2.0.0-1.el7.6
Requires: rh-nodejs12
@@ -35,12 +36,13 @@ size with configurable redundancy (replication or erasure codes/XOR).
%build
. /opt/rh/devtoolset-9/enable
make %{?_smp_mflags} BINDIR=%_bindir LIBDIR=%_libdir QEMU_PLUGINDIR=%_libdir/qemu-kvm
%cmake . -DQEMU_PLUGINDIR=qemu-kvm
%make_build
%install
rm -rf $RPM_BUILD_ROOT
%make_install BINDIR=%_bindir LIBDIR=%_libdir QEMU_PLUGINDIR=%_libdir/qemu-kvm
%make_install
. /opt/rh/rh-nodejs12/enable
cd mon
npm install
@@ -56,7 +58,11 @@ cp -r mon %buildroot/usr/lib/vitastor/mon
%_bindir/vitastor-osd
%_bindir/vitastor-rm
%_libdir/qemu-kvm/block-vitastor.so
%_libdir/vitastor
%_libdir/libfio_vitastor.so
%_libdir/libfio_vitastor_blk.so
%_libdir/libfio_vitastor_sec.so
%_libdir/libvitastor_blk.so
%_libdir/libvitastor_client.so
/usr/lib/vitastor

View File

@@ -1,5 +1,5 @@
# Build packages for CentOS 8 inside a container
# cd ..; podman build -t vitastor-el8 -v `pwd`/build:/root/build -f rpm/vitastor-el8.Dockerfile .
# cd ..; podman build -t vitastor-el8 -v `pwd`/packages:/root/packages -f rpm/vitastor-el8.Dockerfile .
FROM centos:8
@@ -13,8 +13,8 @@ RUN rm -rf /var/lib/dnf/*; dnf download --disablerepo='*' --enablerepo='vitastor
RUN dnf download --source fio
RUN rpm --nomd5 -i qemu*.src.rpm
RUN rpm --nomd5 -i fio*.src.rpm
RUN cd ~/rpmbuild/SPECS && dnf builddep -y --enablerepo=PowerTools --spec qemu-kvm.spec
RUN cd ~/rpmbuild/SPECS && dnf builddep -y --enablerepo=PowerTools --spec fio.spec
RUN cd ~/rpmbuild/SPECS && dnf builddep -y --enablerepo=powertools --spec qemu-kvm.spec
RUN cd ~/rpmbuild/SPECS && dnf builddep -y --enablerepo=powertools --spec fio.spec && dnf install -y cmake
ADD https://vitastor.io/rpms/liburing-el7/liburing-0.7-2.el7.src.rpm /root
@@ -23,23 +23,23 @@ RUN set -e; \
cd ~/rpmbuild/SPECS/; \
. /opt/rh/gcc-toolset-9/enable; \
rpmbuild -ba liburing.spec; \
mkdir -p /root/build/liburing-el8; \
rm -rf /root/build/liburing-el8/*; \
cp ~/rpmbuild/RPMS/*/liburing* /root/build/liburing-el8/; \
cp ~/rpmbuild/SRPMS/liburing* /root/build/liburing-el8/
mkdir -p /root/packages/liburing-el8; \
rm -rf /root/packages/liburing-el8/*; \
cp ~/rpmbuild/RPMS/*/liburing* /root/packages/liburing-el8/; \
cp ~/rpmbuild/SRPMS/liburing* /root/packages/liburing-el8/
RUN rpm -i `ls /root/build/liburing-el7/liburing-*.x86_64.rpm | grep -v debug`
RUN rpm -i `ls /root/packages/liburing-el7/liburing-*.x86_64.rpm | grep -v debug`
ADD . /root/vitastor
RUN set -e; \
cd /root/vitastor/rpm; \
sh build-tarball.sh; \
cp /root/vitastor-0.5.3.el8.tar.gz ~/rpmbuild/SOURCES; \
cp /root/vitastor-0.5.5.el8.tar.gz ~/rpmbuild/SOURCES; \
cp vitastor-el8.spec ~/rpmbuild/SPECS/vitastor.spec; \
cd ~/rpmbuild/SPECS/; \
rpmbuild -ba vitastor.spec; \
mkdir -p /root/build/vitastor-el8; \
rm -rf /root/build/vitastor-el8/*; \
cp ~/rpmbuild/RPMS/*/vitastor* /root/build/vitastor-el8/; \
cp ~/rpmbuild/SRPMS/vitastor* /root/build/vitastor-el8/
mkdir -p /root/packages/vitastor-el8; \
rm -rf /root/packages/vitastor-el8/*; \
cp ~/rpmbuild/RPMS/*/vitastor* /root/packages/vitastor-el8/; \
cp ~/rpmbuild/SRPMS/vitastor* /root/packages/vitastor-el8/

View File

@@ -1,11 +1,11 @@
Name: vitastor
Version: 0.5.3
Release: 2%{?dist}
Version: 0.5.5
Release: 1%{?dist}
Summary: Vitastor, a fast software-defined clustered block storage
License: Vitastor Network Public License 1.0
License: Vitastor Network Public License 1.1
URL: https://vitastor.io/
Source0: vitastor-0.5.3.el8.tar.gz
Source0: vitastor-0.5.5.el8.tar.gz
BuildRequires: liburing-devel >= 0.6
BuildRequires: gperftools-devel
@@ -13,6 +13,7 @@ BuildRequires: gcc-toolset-9-gcc-c++
BuildRequires: nodejs >= 10
BuildRequires: jerasure-devel
BuildRequires: gf-complete-devel
BuildRequires: cmake
Requires: fio = 3.7-3.el8
Requires: qemu-kvm = 4.2.0-29.el8.6
Requires: nodejs >= 10
@@ -33,12 +34,13 @@ size with configurable redundancy (replication or erasure codes/XOR).
%build
. /opt/rh/gcc-toolset-9/enable
make %{?_smp_mflags} BINDIR=%_bindir LIBDIR=%_libdir QEMU_PLUGINDIR=%_libdir/qemu-kvm
%cmake . -DQEMU_PLUGINDIR=qemu-kvm
%make_build
%install
rm -rf $RPM_BUILD_ROOT
%make_install BINDIR=%_bindir LIBDIR=%_libdir QEMU_PLUGINDIR=%_libdir/qemu-kvm
%make_install
cd mon
npm install
cd ..
@@ -53,7 +55,11 @@ cp -r mon %buildroot/usr/lib/vitastor
%_bindir/vitastor-osd
%_bindir/vitastor-rm
%_libdir/qemu-kvm/block-vitastor.so
%_libdir/vitastor
%_libdir/libfio_vitastor.so
%_libdir/libfio_vitastor_blk.so
%_libdir/libfio_vitastor_sec.so
%_libdir/libvitastor_blk.so
%_libdir/libvitastor_client.so
/usr/lib/vitastor

View File

@@ -1,77 +0,0 @@
#!/bin/bash
if [ ! "$BASH_VERSION" ] ; then
echo "Use bash to run this script ($0)" 1>&2
exit 1
fi
format_error()
{
echo $(echo -n -e "\033[1;31m")$1$(echo -n -e "\033[m")
$ETCDCTL get --prefix /vitastor > ./testdata/etcd-dump.txt
exit 1
}
format_green()
{
echo $(echo -n -e "\033[1;32m")$1$(echo -n -e "\033[m")
}
set -e -x
trap 'kill -9 $(jobs -p)' EXIT
ETCD=${ETCD:-etcd}
ETCD_PORT=${ETCD_PORT:-12379}
rm -rf ./testdata
mkdir -p ./testdata
dd if=/dev/zero of=./testdata/test_osd1.bin bs=1024 count=1 seek=$((1024*1024-1))
dd if=/dev/zero of=./testdata/test_osd2.bin bs=1024 count=1 seek=$((1024*1024-1))
dd if=/dev/zero of=./testdata/test_osd3.bin bs=1024 count=1 seek=$((1024*1024-1))
$ETCD -name etcd_test --data-dir ./testdata/etcd \
--advertise-client-urls http://127.0.0.1:$ETCD_PORT --listen-client-urls http://127.0.0.1:$ETCD_PORT \
--initial-advertise-peer-urls http://127.0.0.1:$((ETCD_PORT+1)) --listen-peer-urls http://127.0.0.1:$((ETCD_PORT+1)) \
--max-txn-ops=100000 --auto-compaction-retention=10 --auto-compaction-mode=revision &>./testdata/etcd.log &
ETCD_PID=$!
ETCD_URL=127.0.0.1:$ETCD_PORT/v3
ETCDCTL="${ETCD}ctl --endpoints=http://$ETCD_URL"
./osd --osd_num 1 --bind_address 127.0.0.1 --etcd_address $ETCD_URL $(node mon/simple-offsets.js --format options --device ./testdata/test_osd1.bin 2>/dev/null) &>./testdata/osd1.log &
OSD1_PID=$!
./osd --osd_num 2 --bind_address 127.0.0.1 --etcd_address $ETCD_URL $(node mon/simple-offsets.js --format options --device ./testdata/test_osd2.bin 2>/dev/null) &>./testdata/osd2.log &
OSD2_PID=$!
./osd --osd_num 3 --bind_address 127.0.0.1 --etcd_address $ETCD_URL $(node mon/simple-offsets.js --format options --device ./testdata/test_osd3.bin 2>/dev/null) &>./testdata/osd3.log &
OSD3_PID=$!
cd mon
npm install
cd ..
node mon/mon-main.js --etcd_url http://$ETCD_URL --etcd_prefix "/vitastor" &>./testdata/mon.log &
MON_PID=$!
$ETCDCTL put /vitastor/config/pools '{"1":{"name":"testpool","scheme":"xor","pg_size":3,"pg_minsize":2,"parity_chunks":1,"pg_count":1,"failure_domain":"osd"}}'
sleep 2
if ! ($ETCDCTL get /vitastor/config/pgs --print-value-only | jq -s -e '(. | length) != 0 and (.[0].items["1"]["1"].osd_set | sort) == ["1","2","3"]'); then
format_error "FAILED: 1 PG NOT CONFIGURED"
fi
if ! ($ETCDCTL get /vitastor/pg/state/1/1 --print-value-only | jq -s -e '(. | length) != 0 and .[0].state == ["active"]'); then
format_error "FAILED: 1 PG NOT UP"
fi
echo leak:fio >> testdata/lsan-suppress.txt
echo leak:tcmalloc >> testdata/lsan-suppress.txt
echo leak:ceph >> testdata/lsan-suppress.txt
echo leak:librbd >> testdata/lsan-suppress.txt
echo leak:_M_mutate >> testdata/lsan-suppress.txt
echo leak:_M_assign >> testdata/lsan-suppress.txt
#LSAN_OPTIONS=suppressions=`pwd`/testdata/lsan-suppress.txt LD_PRELOAD=libasan.so.5 \
# fio -thread -name=test -ioengine=./libfio_sec_osd.so -bs=4k -fsync=128 `$ETCDCTL get /vitastor/osd/state/1 --print-value-only | jq -r '"-host="+.addresses[0]+" -port="+(.port|tostring)'` -rw=write -size=32M
LSAN_OPTIONS=suppressions=`pwd`/testdata/lsan-suppress.txt LD_PRELOAD=libasan.so.5 \
fio -thread -name=test -ioengine=./libfio_cluster.so -bs=4M -direct=1 -iodepth=1 -fsync=1 -rw=write -etcd=$ETCD_URL -pool=1 -inode=1 -size=1G -cluster_log_level=10
format_green OK

188
src/CMakeLists.txt Normal file
View File

@@ -0,0 +1,188 @@
cmake_minimum_required(VERSION 2.8)
project(vitastor)
include(GNUInstallDirs)
set(QEMU_PLUGINDIR qemu CACHE STRING "QEMU plugin directory suffix (qemu-kvm on RHEL)")
set(WITH_ASAN false CACHE BOOL "Build with AddressSanitizer")
if("${CMAKE_INSTALL_PREFIX}" MATCHES "^/usr/local/?$")
if(EXISTS "/etc/debian_version")
set(CMAKE_INSTALL_LIBDIR "lib/${CMAKE_LIBRARY_ARCHITECTURE}")
endif()
set(CMAKE_INSTALL_RPATH "${CMAKE_INSTALL_PREFIX}/${CMAKE_INSTALL_LIBDIR}")
endif()
add_definitions(-DVERSION="0.6-dev")
add_definitions(-Wall -Wno-sign-compare -Wno-comment -Wno-parentheses -Wno-pointer-arith)
if (${WITH_ASAN})
add_definitions(-fsanitize=address -fno-omit-frame-pointer)
add_link_options(-fsanitize=address -fno-omit-frame-pointer)
endif (${WITH_ASAN})
set(CMAKE_BUILD_TYPE RelWithDebInfo)
string(REGEX REPLACE "([\\/\\-]O)[12]?" "\\13" CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE}")
string(REGEX REPLACE "([\\/\\-]O)[12]?" "\\13" CMAKE_CXX_FLAGS_MINSIZEREL "${CMAKE_CXX_FLAGS_MINSIZEREL}")
string(REGEX REPLACE "([\\/\\-]O)[12]?" "\\13" CMAKE_CXX_FLAGS_RELWITHDEBINFO "${CMAKE_CXX_FLAGS_RELWITHDEBINFO}")
string(REGEX REPLACE "([\\/\\-]D) *NDEBUG" "" CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE}")
string(REGEX REPLACE "([\\/\\-]D) *NDEBUG" "" CMAKE_CXX_FLAGS_MINSIZEREL "${CMAKE_CXX_FLAGS_MINSIZEREL}")
string(REGEX REPLACE "([\\/\\-]D) *NDEBUG" "" CMAKE_CXX_FLAGS_RELWITHDEBINFO "${CMAKE_CXX_FLAGS_RELWITHDEBINFO}")
string(REGEX REPLACE "([\\/\\-]O)[12]?" "\\13" CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE}")
string(REGEX REPLACE "([\\/\\-]O)[12]?" "\\13" CMAKE_C_FLAGS_MINSIZEREL "${CMAKE_C_FLAGS_MINSIZEREL}")
string(REGEX REPLACE "([\\/\\-]O)[12]?" "\\13" CMAKE_C_FLAGS_RELWITHDEBINFO "${CMAKE_C_FLAGS_RELWITHDEBINFO}")
string(REGEX REPLACE "([\\/\\-]D) *NDEBUG" "" CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE}")
string(REGEX REPLACE "([\\/\\-]D) *NDEBUG" "" CMAKE_C_FLAGS_MINSIZEREL "${CMAKE_C_FLAGS_MINSIZEREL}")
string(REGEX REPLACE "([\\/\\-]D) *NDEBUG" "" CMAKE_C_FLAGS_RELWITHDEBINFO "${CMAKE_C_FLAGS_RELWITHDEBINFO}")
find_package(PkgConfig)
pkg_check_modules(LIBURING REQUIRED liburing)
pkg_check_modules(GLIB REQUIRED glib-2.0)
include_directories(
../
/usr/include/jerasure
${LIBURING_INCLUDE_DIRS}
)
# libvitastor_blk.so
add_library(vitastor_blk SHARED
allocator.cpp blockstore.cpp blockstore_impl.cpp blockstore_init.cpp blockstore_open.cpp blockstore_journal.cpp blockstore_read.cpp
blockstore_write.cpp blockstore_sync.cpp blockstore_stable.cpp blockstore_rollback.cpp blockstore_flush.cpp crc32c.c ringloop.cpp
)
target_link_libraries(vitastor_blk
${LIBURING_LIBRARIES}
tcmalloc_minimal
)
# libfio_vitastor_blk.so
add_library(fio_vitastor_blk SHARED
fio_engine.cpp
../json11/json11.cpp
)
target_link_libraries(fio_vitastor_blk
vitastor_blk
)
# vitastor-osd
add_executable(vitastor-osd
osd_main.cpp osd.cpp osd_secondary.cpp msgr_receive.cpp msgr_send.cpp osd_peering.cpp osd_flush.cpp osd_peering_pg.cpp
osd_primary.cpp osd_primary_subops.cpp etcd_state_client.cpp messenger.cpp osd_cluster.cpp http_client.cpp osd_ops.cpp pg_states.cpp
osd_rmw.cpp base64.cpp timerfd_manager.cpp epoll_manager.cpp ../json11/json11.cpp
)
target_link_libraries(vitastor-osd
vitastor_blk
Jerasure
)
# libfio_vitastor_sec.so
add_library(fio_vitastor_sec SHARED
fio_sec_osd.cpp
rw_blocking.cpp
)
target_link_libraries(fio_vitastor_sec
tcmalloc_minimal
)
# libvitastor_client.so
add_library(vitastor_client SHARED
cluster_client.cpp epoll_manager.cpp etcd_state_client.cpp
messenger.cpp msgr_send.cpp msgr_receive.cpp ringloop.cpp ../json11/json11.cpp
http_client.cpp osd_ops.cpp pg_states.cpp timerfd_manager.cpp base64.cpp
)
target_link_libraries(vitastor_client
tcmalloc_minimal
${LIBURING_LIBRARIES}
)
# libfio_vitastor.so
add_library(fio_vitastor SHARED
fio_cluster.cpp
)
target_link_libraries(fio_vitastor
vitastor_client
)
# vitastor-nbd
add_executable(vitastor-nbd
nbd_proxy.cpp
)
target_link_libraries(vitastor-nbd
vitastor_client
)
# vitastor-rm
add_executable(vitastor-rm
rm_inode.cpp
)
target_link_libraries(vitastor-rm
vitastor_client
)
# vitastor-dump-journal
add_executable(vitastor-dump-journal
dump_journal.cpp crc32c.c
)
# qemu_driver.so
add_library(qemu_proxy STATIC qemu_proxy.cpp)
target_compile_options(qemu_proxy PUBLIC -fPIC)
target_include_directories(qemu_proxy PUBLIC
../qemu/b/qemu
../qemu/include
${GLIB_INCLUDE_DIRS}
)
target_link_libraries(qemu_proxy
vitastor_client
)
add_library(qemu_vitastor SHARED
qemu_driver.c
)
target_link_libraries(qemu_vitastor
qemu_proxy
)
set_target_properties(qemu_vitastor PROPERTIES
PREFIX ""
OUTPUT_NAME "block-vitastor"
)
### Test stubs
# stub_osd, stub_bench, osd_test
add_executable(stub_osd stub_osd.cpp rw_blocking.cpp)
target_link_libraries(stub_osd tcmalloc_minimal)
add_executable(stub_bench stub_bench.cpp rw_blocking.cpp)
target_link_libraries(stub_bench tcmalloc_minimal)
add_executable(osd_test osd_test.cpp rw_blocking.cpp)
target_link_libraries(osd_test tcmalloc_minimal)
# osd_rmw_test
add_executable(osd_rmw_test osd_rmw_test.cpp allocator.cpp)
target_link_libraries(osd_rmw_test Jerasure tcmalloc_minimal)
# stub_uring_osd
add_executable(stub_uring_osd
stub_uring_osd.cpp epoll_manager.cpp messenger.cpp msgr_send.cpp msgr_receive.cpp ringloop.cpp timerfd_manager.cpp ../json11/json11.cpp
)
target_link_libraries(stub_uring_osd
${LIBURING_LIBRARIES}
tcmalloc_minimal
)
# osd_peering_pg_test
add_executable(osd_peering_pg_test osd_peering_pg_test.cpp osd_peering_pg.cpp)
target_link_libraries(osd_peering_pg_test tcmalloc_minimal)
# test_allocator
add_executable(test_allocator test_allocator.cpp allocator.cpp)
## test_blockstore, test_shit
#add_executable(test_blockstore test_blockstore.cpp timerfd_interval.cpp)
#target_link_libraries(test_blockstore blockstore)
#add_executable(test_shit test_shit.cpp osd_peering_pg.cpp)
#target_link_libraries(test_shit ${LIBURING_LIBRARIES} m)
### Install
install(TARGETS vitastor-osd vitastor-dump-journal vitastor-nbd vitastor-rm RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})
install(TARGETS fio_vitastor fio_vitastor_blk fio_vitastor_sec vitastor_blk vitastor_client LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR})
install(TARGETS qemu_vitastor LIBRARY DESTINATION /usr/${CMAKE_INSTALL_LIBDIR}/${QEMU_PLUGINDIR})

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include <stdexcept>
#include "allocator.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#pragma once

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "base64.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#pragma once
#include <string>

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "blockstore_impl.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#pragma once

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "blockstore_impl.h"
@@ -7,6 +7,8 @@ journal_flusher_t::journal_flusher_t(int flusher_count, blockstore_impl_t *bs)
{
this->bs = bs;
this->flusher_count = flusher_count;
this->cur_flusher_count = 1;
this->target_flusher_count = 1;
dequeuing = false;
trimming = false;
active_flushers = 0;
@@ -68,10 +70,24 @@ bool journal_flusher_t::is_active()
void journal_flusher_t::loop()
{
for (int i = 0; (active_flushers > 0 || dequeuing) && i < flusher_count; i++)
target_flusher_count = bs->write_iodepth*2;
if (target_flusher_count <= 0)
target_flusher_count = 1;
else if (target_flusher_count > flusher_count)
target_flusher_count = flusher_count;
if (target_flusher_count > cur_flusher_count)
cur_flusher_count = target_flusher_count;
else if (target_flusher_count < cur_flusher_count)
{
co[i].loop();
while (target_flusher_count < cur_flusher_count)
{
if (co[cur_flusher_count-1].wait_state)
break;
cur_flusher_count--;
}
}
for (int i = 0; (active_flushers > 0 || dequeuing) && i < cur_flusher_count; i++)
co[i].loop();
}
void journal_flusher_t::enqueue_flush(obj_ver_id ov)
@@ -223,6 +239,7 @@ bool journal_flusher_co::loop()
resume_0:
if (!flusher->flush_queue.size() || !flusher->dequeuing)
{
stop_flusher:
if (flusher->trim_wanted > 0 && flusher->journal_trim_counter > 0)
{
// Attempt forced trim
@@ -327,9 +344,7 @@ resume_0:
#ifdef BLOCKSTORE_DEBUG
printf("No older flushes, stopping\n");
#endif
flusher->dequeuing = false;
wait_state = 0;
return true;
goto stop_flusher;
}
}
}

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
struct copy_buffer_t
{
@@ -80,7 +80,7 @@ class journal_flusher_t
{
int trim_wanted = 0;
bool dequeuing;
int flusher_count;
int flusher_count, cur_flusher_count, target_flusher_count;
int flusher_start_threshold;
journal_flusher_co *co;
blockstore_impl_t *bs;

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "blockstore_impl.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#pragma once
@@ -199,7 +199,10 @@ class blockstore_impl_t
// Suitable only for server SSDs with capacitors, requires disabled data and journal fsyncs
int immediate_commit = IMMEDIATE_NONE;
bool inmemory_meta = false;
int flusher_count;
// Maximum flusher count
unsigned flusher_count;
// Maximum queue depth
unsigned max_write_iodepth = 128;
/******* END OF OPTIONS *******/
struct ring_consumer_t ring_consumer;
@@ -226,6 +229,7 @@ class blockstore_impl_t
struct journal_t journal;
journal_flusher_t *flusher;
int write_iodepth = 0;
bool live = false, queue_stall = false;
ring_loop_t *ringloop;

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "blockstore_impl.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#pragma once

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "blockstore_impl.h"
@@ -20,7 +20,7 @@ int blockstore_journal_check_t::check_available(blockstore_op_t *op, int entries
int required = entries_required;
while (1)
{
int fits = bs->journal.no_same_sector_overwrites && bs->journal.sector_info[next_sector].written
int fits = bs->journal.no_same_sector_overwrites && next_pos == bs->journal.next_free && bs->journal.sector_info[next_sector].written
? 0
: (bs->journal.block_size - next_in_pos) / size;
if (fits > 0)
@@ -81,7 +81,8 @@ int blockstore_journal_check_t::check_available(blockstore_op_t *op, int entries
}
// In fact, it's even more rare than "ran out of journal space", so print a warning
printf(
"Ran out of journal sector buffers: %d/%lu buffers used (%d dirty), next buffer (%ld) is %s and flushed %lu times\n",
"Ran out of journal sector buffers: %d/%lu buffers used (%d dirty), next buffer (%ld)"
" is %s and flushed %lu times. Consider increasing \'journal_sector_buffer_count\'\n",
used, bs->journal.sector_count, dirty, next_sector,
bs->journal.sector_info[next_sector].dirty ? "dirty" : "not dirty",
bs->journal.sector_info[next_sector].flush_count
@@ -103,11 +104,8 @@ int blockstore_journal_check_t::check_available(blockstore_op_t *op, int entries
{
// No space in the journal. Wait until used_start changes.
printf(
"Ran out of journal space (free space: %lu bytes, sectors to write: %d)\n",
(bs->journal.next_free >= bs->journal.used_start
? bs->journal.len-bs->journal.block_size - (bs->journal.next_free-bs->journal.used_start)
: bs->journal.used_start - bs->journal.next_free),
sectors_to_write
"Ran out of journal space (used_start=%08lx, next_free=%08lx, dirty_start=%08lx)\n",
bs->journal.used_start, bs->journal.next_free, bs->journal.dirty_start
);
PRIV(op)->wait_for = WAIT_JOURNAL;
bs->flusher->request_trim();

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#pragma once

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include <sys/file.h>
#include "blockstore_impl.h"
@@ -70,6 +70,7 @@ void blockstore_impl_t::parse_config(blockstore_config_t & config)
meta_block_size = strtoull(config["meta_block_size"].c_str(), NULL, 10);
bitmap_granularity = strtoull(config["bitmap_granularity"].c_str(), NULL, 10);
flusher_count = strtoull(config["flusher_count"].c_str(), NULL, 10);
max_write_iodepth = strtoull(config["max_write_iodepth"].c_str(), NULL, 10);
// Validate
if (!block_size)
{
@@ -83,6 +84,10 @@ void blockstore_impl_t::parse_config(blockstore_config_t & config)
{
flusher_count = 32;
}
if (!max_write_iodepth)
{
max_write_iodepth = 128;
}
if (!disk_alignment)
{
disk_alignment = 4096;

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "blockstore_impl.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "blockstore_impl.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "blockstore_impl.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "blockstore_impl.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "blockstore_impl.h"
@@ -167,6 +167,10 @@ int blockstore_impl_t::dequeue_write(blockstore_op_t *op)
.version = op->version,
}, e).first;
}
if (write_iodepth >= max_write_iodepth)
{
return 0;
}
if ((dirty_it->second.state & BS_ST_TYPE_MASK) == BS_ST_BIG_WRITE)
{
blockstore_journal_check_t space_check(this);
@@ -191,6 +195,7 @@ int blockstore_impl_t::dequeue_write(blockstore_op_t *op)
FINISH_OP(op);
return 1;
}
write_iodepth++;
BS_SUBMIT_GET_SQE(sqe, data);
dirty_it->second.location = loc << block_order;
dirty_it->second.state = (dirty_it->second.state & ~BS_ST_WORKFLOW_MASK) | BS_ST_SUBMITTED;
@@ -243,6 +248,7 @@ int blockstore_impl_t::dequeue_write(blockstore_op_t *op)
{
return 0;
}
write_iodepth++;
// There is sufficient space. Get SQE(s)
struct io_uring_sqe *sqe1 = NULL;
if (immediate_commit != IMMEDIATE_NONE ||
@@ -432,6 +438,7 @@ resume_4:
}
// Acknowledge write
op->retval = op->len;
write_iodepth--;
FINISH_OP(op);
return 1;
}

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#include <stdexcept>
#include "cluster_client.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#pragma once

View File

@@ -8,4 +8,10 @@
// unsigned __int64 _mm_crc32_u64 (unsigned __int64 crc, unsigned __int64 v)
// unsigned int _mm_crc32_u8 (unsigned int crc, unsigned char v)
#ifdef __cplusplus
extern "C" {
#endif
uint32_t crc32c(uint32_t crc, const void *buf, size_t len);
#ifdef __cplusplus
};
#endif

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#define _LARGEFILE64_SOURCE
#include <sys/types.h>

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#include <sys/epoll.h>
#include <sys/poll.h>

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#pragma once

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#include "osd_ops.h"
#include "pg_states.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#pragma once

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
// FIO engine to test cluster I/O
//

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
// FIO engine to test Blockstore
//

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
// FIO engine to test Blockstore through Secondary OSD interface
//

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#include <netinet/tcp.h>
#include <sys/epoll.h>

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#pragma once
#include <string>

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#pragma once

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#include <unistd.h>
#include <fcntl.h>

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#pragma once

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#include "messenger.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#define _XOPEN_SOURCE
#include <limits.h>

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
// Similar to qemu-nbd, but sets timeout and uses io_uring
#include <linux/nbd.h>
@@ -111,7 +111,7 @@ public:
{
printf(
"Vitastor NBD proxy\n"
"(c) Vitaliy Filippov, 2020 (VNPL-1.0)\n\n"
"(c) Vitaliy Filippov, 2020 (VNPL-1.1)\n\n"
"USAGE:\n"
" %s map --etcd_address <etcd_address> --pool <pool> --inode <inode> --size <size in bytes>\n"
" %s unmap /dev/nbd0\n"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#pragma once

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include <sys/socket.h>
#include <sys/poll.h>

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#pragma once

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "osd.h"
#include "base64.h"
@@ -802,11 +802,11 @@ void osd_t::report_pg_states()
if (pg_it->second.state == PG_OFFLINE)
{
// Remove offline PGs after reporting their state
this->pgs.erase(pg_it);
if (pg_it->second.scheme == POOL_SCHEME_JERASURE)
{
use_jerasure(pg_it->second.pg_size, pg_it->second.pg_size-pg_it->second.parity_chunks, false);
}
this->pgs.erase(pg_it);
}
}
}

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "osd.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#pragma once

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "osd.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#include "osd_ops.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#pragma once

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include <netinet/tcp.h>
#include <sys/epoll.h>

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include <unordered_map>
#include "osd_peering_pg.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include <map>
#include <vector>

View File

@@ -1,8 +1,9 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#define _LARGEFILE64_SOURCE
#include "malloc_or_die.h"
#include "osd_peering_pg.h"
#define STRIPE_SHIFT 12

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "osd_primary.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#pragma once

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "osd_primary.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include <stdexcept>
#include <string.h>
@@ -11,7 +11,7 @@
#include "osd_rmw.h"
#include "malloc_or_die.h"
#define OSD_JERASURE_W 32
#define OSD_JERASURE_W 8
static inline void extend_read(uint32_t start, uint32_t end, osd_rmw_stripe_t & stripe)
{

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#pragma once

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#define RMW_DEBUG

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include "osd.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 (see README.md for details)
// License: VNPL-1.1 (see README.md for details)
#include <sys/types.h>
#include <sys/socket.h>

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#include "pg_states.h"

View File

@@ -1,5 +1,5 @@
// Copyright (c) Vitaliy Filippov, 2019+
// License: VNPL-1.0 or GNU GPL-2.0+ (see README.md for details)
// License: VNPL-1.1 or GNU GPL-2.0+ (see README.md for details)
#pragma once

Some files were not shown because too many files have changed in this diff Show More