Vitaliy Filippov
8f64fc61e7
Ignore empty events in mon
1 year ago
Vitaliy Filippov
4a9f001d9e
Make mon also ping etcd websockets regularly
1 year ago
Vitaliy Filippov
68b6763ebe
Add asserts for lp-optimizer tests, pass `ordered` from the monitor
1 year ago
Vitaliy Filippov
08e467270a
Fix pg_size changing from 3 to 2
1 year ago
Vitaliy Filippov
5473d5b4a2
Rework HTTP client to use keepalive, move getifaddr_list to addr_util
1 year ago
Vitaliy Filippov
fa687d3878
Allow to configure OSD placement in node_placement
1 year ago
Vitaliy Filippov
32b1312abb
Remove stale deleted inode statistics in monitor
2 years ago
Vitaliy Filippov
d5c8fde5de
Remove kludgy $IP and $ETCD_MON parsing from make-osd.sh, suggest to use vitastor.conf
2 years ago
Vitaliy Filippov
7a0b5212fe
Exit if unable to restart watches
...
FIXME: It's probably not OK for the client to exit in this case
2 years ago
Vitaliy Filippov
a8f5c71ae8
Use the same etcd address selection algorithm in the monitor
2 years ago
Vitaliy Filippov
6e12aca53b
Remove the total PG count restriction in optimize_change which was leading to unfeasible problems sometimes
2 years ago
Vitaliy Filippov
6e0e172e15
Implement OSD address selection from a specified subnet
2 years ago
Vitaliy Filippov
f0ebfae3b8
Fix vitastor-cli alloc-osd, use vitastor-cli in make-osd.sh
2 years ago
Vitaliy Filippov
66fe1a469b
Additionally balance parity chunks over OSDs using round-robin when generating initial distribution
2 years ago
Vitaliy Filippov
aa436027c8
Report pg/history from OSD on every degraded activation
...
Required to prevent data loss due to activation of an OSD with older data
when PG OSD set change doesn't occur. I.e. fixes the simplest case:
- Run 2 OSDs with 1 PG
- Start writing into the PG
- Stop OSD 2
- Stop OSD 1
- Start OSD 2
After this change the PG will refuse to start after the last step.
2 years ago
Vitaliy Filippov
0f3f0a9d29
Calculate average statistics in mon, remove buggy "fix_stat_overflows"
2 years ago
Vitaliy Filippov
6e6f407df3
Simplify & fix monitor stats aggregation
2 years ago
Vitaliy Filippov
4d43774cbb
Use 5s etcd_report_interval by default
2 years ago
Vitaliy Filippov
ffb06536ff
Revoke lease in mon on SIGINT & SIGTERM, fix raw_to_usable calculation
2 years ago
Vitaliy Filippov
cfe8de9b84
Autosync based on number of unstable ops to prevent journal stalls
2 years ago
Vitaliy Filippov
da99686a15
Correctly aggregate pool statistics for unknown pools
2 years ago
Vitaliy Filippov
b66160a7ad
Aggregate per-pool statistics in mon
2 years ago
Vitaliy Filippov
dfdf5c1f9c
Fix comments in mon.js
2 years ago
Vitaliy Filippov
6810e93c3f
Add RDMA options to mon.js list
2 years ago
Vitaliy Filippov
2a02f3c4c7
Add metadata superblock and check it on start
...
Refuse to start if the superblock is missing or bad version;
zero out the metadata area when initializing superblock.
2 years ago
Vitaliy Filippov
82c1a7ec67
Fix statistics reporting, split inode number into pool & inode
2 years ago
Vitaliy Filippov
2612d3198a
Introduce image names and metadata storage in etcd
...
Each inode has: image name, parent inode number & pool, size and readonly flag
Snapshots are created by switching image name to a different inode number
while using the older inode as parent.
2 years ago
Vitaliy Filippov
d0c2e31312
Add a test for snapshots, fix bugs. Now the test passes
2 years ago
Vitaliy Filippov
691f066055
Actual snapshot support (untested)
2 years ago
Vitaliy Filippov
ffe1cd4c79
Report inode I/O statistics, aggregate it in the monitor
2 years ago
Vitaliy Filippov
4ae1b84c67
Report inode space usage statistics to etcd, aggregate it in the monitor
2 years ago
Vitaliy Filippov
97efb9e299
Do not crash on PG re-peering events when operations are in progress
2 years ago
Vitaliy Filippov
75a6a556b5
Shuffle PGs for better data device utilisation
2 years ago
Vitaliy Filippov
8f8b90be7a
Add min_flusher_count configuration
2 years ago
Vitaliy Filippov
df99e232ee
Deduplicate osd_sets in pg history + raise request size limit for etcd
2 years ago
Vitaliy Filippov
3a40fa4127
Fix monitor errors in case of OSD removal
2 years ago
Vitaliy Filippov
435045751d
Delete objects only after a SYNC during rebalance in the non-immediate_commit mode
...
Previously OSDs could commit deletes before writes during recovery or rebalance
in the "lazy fsync" (immediate_commit=off) mode which could result in lost objects
2 years ago
Vitaliy Filippov
9f59381bea
Re-distribute PG primaries over OSDs that come up after a short downtime
2 years ago
Vitaliy Filippov
87dbd8fa57
Use empty hash as the default value for some etcd keys in the monitor
2 years ago
Vitaliy Filippov
b44f49aab2
Ignore zero OSDs in history osd_sets
2 years ago
Vitaliy Filippov
af5155fcd9
Implement "no_recovery" and "no_rebalance" flags
2 years ago
Vitaliy Filippov
0d2efbecc9
Preserve previous PG history when changing PG distribution
...
Fixes incorrect PG history in case when a new rebalance is started
before the finish of the previous one which could make primary OSDs unable
to locate some objects on some secondaries.
2 years ago
Vitaliy Filippov
e62e8b6bae
Use real pg configuration instead of the "last clean" one for generating PG history
...
Basically fixes the bug introduced in 0.5.7 where an rebalance interrupted
by the monitor could result in forgetting objects moved to the new place
2 years ago
Vitaliy Filippov
7006875a24
Make monitor stick to one etcd until the restart
2 years ago
Vitaliy Filippov
836635c518
Use osd_out_time = 10 minutes by default
2 years ago
Vitaliy Filippov
2a5036669d
Fix PG count change procedure
...
In previous versions PG histories were calculated incorrectly during
PG count change which led to objects being lost on OSDs not in PG's osd set.
2 years ago
Vitaliy Filippov
07912fd670
Use history/last_clean_pgs to avoid extra data move when observing a series of changes in the cluster
2 years ago
Vitaliy Filippov
24e7075f08
Fix monitor's statistics aggregation
2 years ago
Vitaliy Filippov
e899ed2c25
Make OSDs with 256 flushers (as they are now dynamic)
2 years ago
Vitaliy Filippov
e16b87ecc8
Rename random_combinations() parameter from "unordered" to "ordered" as it's more correct
2 years ago
