WIP Implement resizing partitions created with vitastor-disk

src/disk_tool/CMakeLists.txt

@@ -5,7 +5,8 @@ project(vitastor)
 # vitastor-disk
 add_executable(vitastor-disk
 	disk_tool.cpp disk_simple_offsets.cpp
-	disk_tool_journal.cpp disk_tool_meta.cpp disk_tool_prepare.cpp disk_tool_resize.cpp disk_tool_udev.cpp disk_tool_utils.cpp disk_tool_upgrade.cpp
+	disk_tool_journal.cpp disk_tool_meta.cpp disk_tool_prepare.cpp disk_tool_resize.cpp
+	disk_tool_resize_auto.cpp disk_tool_udev.cpp disk_tool_utils.cpp disk_tool_upgrade.cpp
 	../util/crc32c.c ../util/str_util.cpp ../util/json_util.cpp ../../json11/json11.cpp ../util/rw_blocking.cpp ../util/allocator.cpp ../util/ringloop.cpp ../blockstore/blockstore_disk.cpp
 )
 target_link_libraries(vitastor-disk

src/disk_tool/disk_tool.cpp

@@ -92,8 +92,22 @@ static const char *help_text =
     "  \n"
     "  Requires the `sfdisk` utility.\n"
     "\n"
-    "vitastor-disk resize <ALL_OSD_PARAMETERS> <NEW_LAYOUT> [--iodepth 32]\n"
-    "  Resize data area and/or rewrite/move journal and metadata\n"
+    "vitastor-disk resize <osd_num>|<osd_device> [OPTIONS]\n"
+    "  Resize data area and/or move journal and metadata:\n"
+    "    --move-journal TARGET    move journal to TARGET\n"
+    "    --move-meta TARGET       move metadata to TARGET\n"
+    "    --journal-size NEW_SIZE  resize journal to NEW_SIZE\n"
+    "    --data-size NEW_SIZE     resize data device to NEW_SIZE\n"
+    "    --dry-run                only show new layout, do not apply it\n"
+    "  \n"
+    "  NEW_SIZE may include k/m/g/t suffixes.\n"
+    "  TARGET may be one of:\n"
+    "    <partition>    move journal/metadata to an existing GPT partition\n"
+    "    <raw_device>   create a GPT partition on <raw_device> and move journal/metadata to it\n"
+    "    \"\"           (empty string) move journal/metadata back to the data device\n"
+    "\n"
+    "vitastor-disk raw-resize <ALL_OSD_PARAMETERS> <NEW_LAYOUT> [--iodepth 32]\n"
+    "  Resize data area and/or rewrite/move journal and metadata (manual format).\n"
     "  ALL_OSD_PARAMETERS must include all (at least all disk-related)\n"
     "  parameters from OSD command line (i.e. from systemd unit or superblock).\n"
     "  NEW_LAYOUT may include new disk layout parameters:\n"
@@ -236,6 +250,10 @@ int main(int argc, char *argv[])
         {
             self.options["force"] = "1";
         }
+        else if (!strcmp(argv[i], "--dry-run") || !strcmp(argv[i], "--dry_run"))
+        {
+            self.options["dry_run"] = "1";
+        }
         else if (!strcmp(argv[i], "--allow-data-loss"))
         {
             self.options["allow_data_loss"] = "1";
@@ -243,7 +261,7 @@ int main(int argc, char *argv[])
         else if (argv[i][0] == '-' && argv[i][1] == '-' && i < argc-1)
         {
             char *key = argv[i]+2;
-            self.options[key] = argv[++i];
+            self.options[str_replace(key, "-", "_")] = argv[++i];
         }
         else
         {
@@ -375,7 +393,16 @@ int main(int argc, char *argv[])
     }
     else if (!strcmp(cmd[0], "resize"))
     {
-        return self.resize_data();
+        if (cmd.size() != 2)
+        {
+            fprintf(stderr, "Exactly 1 OSD number or OSD device path argument is required\n");
+            return 1;
+        }
+        return self.resize_data(cmd[1]);
+    }
+    else if (!strcmp(cmd[0], "raw-resize"))
+    {
+        return self.raw_resize();
     }
     else if (!strcmp(cmd[0], "simple-offsets"))
     {

src/disk_tool/disk_tool.h

@@ -98,7 +98,11 @@ struct disk_tool_t
     int write_json_journal(json11::Json entries);
     int write_json_meta(json11::Json meta);
 
-    int resize_data();
+    int resize_data(std::string device);
+    int resize_parse_move_journal(std::map<std::string, std::string> & move_options);
+    int resize_parse_move_meta(std::map<std::string, std::string> & move_options);
+
+    int raw_resize();
     int resize_parse_params();
     void resize_init(blockstore_meta_header_v2_t *hdr);
     int resize_remap_blocks();
@@ -116,6 +120,7 @@ struct disk_tool_t
     int systemd_start_stop_osds(const std::vector<std::string> & cmd, const std::vector<std::string> & devices);
     int pre_exec_osd(std::string device);
     int purge_devices(const std::vector<std::string> & devices);
+    int clear_osd_superblock(const std::string & dev);
 
     json11::Json read_osd_superblock(std::string device, bool expect_exist = true, bool ignore_nonref = false);
     uint32_t write_osd_superblock(std::string device, json11::Json params);

src/disk_tool/disk_tool_resize.cpp

@@ -18,7 +18,7 @@ struct resizer_data_moving_t
     uint64_t old_loc, new_loc;
 };
 
-int disk_tool_t::resize_data()
+int disk_tool_t::raw_resize()
 {
     int r;
     // Parse parameters

src/disk_tool/disk_tool_resize_auto.cpp

@@ -0,0 +1,282 @@
+// Copyright (c) Vitaliy Filippov, 2019+
+// License: VNPL-1.1 (see README.md for details)
+
+#include "disk_tool.h"
+#include "rw_blocking.h"
+#include "str_util.h"
+#include "json_util.h"
+
+int disk_tool_t::resize_data(std::string device)
+{
+    if (options.find("move_journal") == options.end() &&
+        options.find("move_data") == options.end() &&
+        options.find("journal_size") == options.end() &&
+        options.find("data_size") == options.end())
+    {
+        fprintf(stderr, "None of --move-journal, --move-data, --journal-size, --data-size options are specified - nothing to do!\n");
+        return 1;
+    }
+    if (stoull_full(device))
+        device = "/dev/vitastor/osd"+device+"-data";
+    json11::Json sb = read_osd_superblock(device, true, false);
+    if (sb.is_null())
+        return 1;
+    auto sb_params = json_to_string_map(sb["params"].object_items());
+    blockstore_disk_t dsk;
+    try
+    {
+        dsk.parse_config(sb_params);
+        dsk.data_io = dsk.meta_io = dsk.journal_io = "cached";
+        dsk.open_data();
+        dsk.open_meta();
+        dsk.open_journal();
+        dsk.calc_lengths(true);
+    }
+    catch (std::exception & e)
+    {
+        dsk.close_all();
+        fprintf(stderr, "%s\n", e.what());
+        return 1;
+    }
+    dsk.close_all();
+    auto old_journal_device = dsk.journal_device;
+    auto old_meta_device = dsk.meta_device;
+    new_data_offset = dsk.data_offset;
+    new_meta_offset = dsk.meta_offset;
+    new_journal_len = dsk.journal_len;
+    if (options.find("journal_size") != options.end())
+    {
+        new_journal_len = parse_size(options["journal_size"]);
+        if (options.find("move_journal") == options.end())
+            options["move_journal"] = dsk.journal_device == dsk.data_device ? "" : dsk.journal_device;
+    }
+    std::map<std::string, std::string> move_options;
+    if (options.find("move_journal") != options.end())
+    {
+        if (resize_parse_move_journal(move_options) != 0)
+            return 1;
+    }
+    if (options.find("move_meta") != options.end())
+    {
+        if (resize_parse_move_meta(move_options) != 0)
+            return 1;
+    }
+    if (options.find("data_size") != options.end())
+    {
+        auto new_data_dev_size = options["data_size"] == "max"
+            ? dsk.data_device_size : parse_size(options["data_size"]);
+        if (new_data_dev_size-dsk.data_offset != dsk.data_len)
+            move_options["new_data_len"] = std::to_string(new_data_dev_size-new_data_offset);
+    }
+    if (new_data_offset != dsk.data_offset)
+        move_options["new_data_offset"] = std::to_string(new_data_offset);
+    if (new_meta_offset != dsk.meta_offset)
+        move_options["new_meta_offset"] = std::to_string(new_meta_offset);
+    auto orig_options = std::move(options);
+    options = sb_params;
+    for (auto & kv: move_options)
+        options[kv.first] = kv.second;
+    bool dry_run = orig_options.find("dry_run") != orig_options.end();
+    if (!json)
+    {
+        std::string cmd;
+        for (auto & kv: move_options)
+            cmd += " --"+kv.first+" "+kv.second;
+        fprintf(stderr, "Running resize%s\n", cmd.c_str());
+    }
+    if (!dry_run && raw_resize() != 0)
+        return 1;
+    // Write new superblocks
+    json11::Json::object new_sb_params = sb["params"].object_items();
+    if (move_options.find("new_journal_device") != move_options.end())
+        new_sb_params["journal_device"] = move_options["new_journal_device"];
+    if (move_options.find("new_meta_device") != move_options.end())
+        new_sb_params["meta_device"] = move_options["new_meta_device"];
+    new_sb_params["data_offset"] = new_data_offset;
+    new_sb_params["meta_offset"] = new_meta_offset;
+    if (move_options.find("new_data_len") != move_options.end())
+        new_sb_params["data_size"] = stoull_full(move_options["new_data_len"]);
+    std::set<std::string> clear_superblocks, write_superblocks;
+    auto new_journal_device = move_options.find("new_journal_device") != move_options.end()
+        ? move_options["new_journal_device"] : dsk.journal_device;
+    auto new_meta_device = move_options.find("new_meta_device") != move_options.end()
+        ? move_options["new_meta_device"] : dsk.meta_device;
+    write_superblocks.insert(dsk.data_device);
+    write_superblocks.insert(new_journal_device);
+    write_superblocks.insert(new_meta_device);
+    if (write_superblocks.find(old_journal_device) == write_superblocks.end())
+        clear_superblocks.insert(old_journal_device);
+    if (write_superblocks.find(old_meta_device) == write_superblocks.end())
+        clear_superblocks.insert(old_meta_device);
+    for (auto & dev: clear_superblocks)
+    {
+        if (!json)
+            fprintf(stderr, "Clearing OSD superblock on %s\n", dev.c_str());
+        if (!dry_run && !clear_osd_superblock(dev))
+            return 1;
+    }
+    for (auto & dev: write_superblocks)
+    {
+        if (!json)
+            fprintf(stderr, "Writing new OSD superblock to %s\n", dev.c_str());
+        if (!dry_run && !write_osd_superblock(dev, new_sb_params))
+            return 1;
+    }
+    if (json)
+    {
+        printf("%s\n", json11::Json(json11::Json::object {
+            { "new_sb_params", new_sb_params },
+        }).dump().c_str());
+    }
+    return 0;
+}
+
+int disk_tool_t::resize_parse_move_journal(std::map<std::string, std::string> & move_options)
+{
+    if (options["move_journal"] == "")
+    {
+        // move back to the data device
+        // but first check if not already there :)
+        if (dsk.journal_device == dsk.data_device && new_journal_len == dsk.journal_len)
+        {
+            // already there
+            fprintf(stderr, "journal is already on data device and has the same size\n");
+            return 0;
+        }
+        move_options["new_journal_device"] = dsk.data_device;
+        move_options["new_journal_offset"] = "4096";
+        move_options["new_journal_len"] = std::to_string(new_journal_len);
+        new_data_offset += new_journal_len;
+        if (dsk.meta_device == dsk.data_device)
+            new_meta_offset += new_journal_len;
+    }
+    else
+    {
+        std::string real_dev = realpath_str(options["move_journal"], false);
+        if (real_dev == "")
+            return 1;
+        std::string parent_dev = get_parent_device(real_dev);
+        if (parent_dev == "")
+            return 1;
+        if (parent_dev == real_dev)
+        {
+            // whole disk - create partition
+            std::string old_real_dev = realpath_str(dsk.journal_device);
+            if (old_real_dev == "")
+                return 1;
+            if (options.find("force") == options.end() &&
+                get_parent_device(old_real_dev) == parent_dev)
+            {
+                // already there
+                fprintf(stderr, "journal is already on a partition of %s, add --force to create a new partition\n", options["move_journal"].c_str());
+                return 0;
+            }
+            auto devinfos = collect_devices({ real_dev });
+            if (devinfos.size() == 0)
+                return 1;
+            std::vector<std::string> sizes;
+            sizes.push_back(std::to_string((new_journal_len+1024*1024-1)/1024/1024)+"MiB");
+            auto new_parts = add_partitions(devinfos[0], sizes);
+            if (!new_parts.array_items().size())
+                return 1;
+            options["move_journal"] = "/dev/disk/by-partuuid/"+new_parts[0]["uuid"].string_value();
+        }
+        else if (options["move_journal"].substr(0, 22) != "/dev/disk/by-partuuid/")
+        {
+            // Partitions should be identified by GPT partition UUID
+            fprintf(stderr, "%s does not start with /dev/disk/by-partuuid/. Partitions should be identified by GPT partition UUIDs\n", options["move_journal"].c_str());
+            return 1;
+        }
+        else
+        {
+            // already a partition - check that it's a GPT partition with correct type
+            if (options.find("force") == options.end() &&
+                check_existing_partition(real_dev) != 0)
+            {
+                return 1;
+            }
+        }
+        new_journal_len = get_device_size(options["move_journal"]) - 4096;
+        move_options["new_journal_device"] = options["move_journal"];
+        move_options["new_journal_offset"] = "4096";
+        move_options["new_journal_len"] = std::to_string(new_journal_len);
+        if (dsk.journal_device == dsk.data_device)
+            new_data_offset -= dsk.journal_len;
+    }
+    return 0;
+}
+
+int disk_tool_t::resize_parse_move_meta(std::map<std::string, std::string> & move_options)
+{
+    if (options["move_meta"] == "")
+    {
+        // move back to the data device
+        // but first check if not already there :)
+        if (dsk.meta_device == dsk.data_device)
+        {
+            // already there
+            fprintf(stderr, "metadata is already on data device\n");
+            return 0;
+        }
+        auto new_journal_device = move_options.find("new_journal_device") != move_options.end()
+            ? move_options["new_journal_device"] : dsk.journal_device;
+        move_options["new_meta_device"] = dsk.data_device;
+        move_options["new_meta_offset"] = std::to_string(new_journal_device == dsk.data_device
+            ? 4096+new_journal_len : 4096);
+        move_options["new_meta_len"] = std::to_string(dsk.meta_len);
+        new_data_offset += dsk.meta_len;
+    }
+    else
+    {
+        std::string real_dev = realpath_str(options["move_meta"], false);
+        if (real_dev == "")
+            return 1;
+        std::string parent_dev = get_parent_device(real_dev);
+        if (parent_dev == "")
+            return 1;
+        if (parent_dev == real_dev)
+        {
+            // whole disk - create partition
+            std::string old_real_dev = realpath_str(dsk.meta_device);
+            if (old_real_dev == "")
+                return 1;
+            if (options.find("force") == options.end() &&
+                get_parent_device(old_real_dev) == parent_dev)
+            {
+                // already there
+                fprintf(stderr, "metadata is already on a partition of %s\n", options["move_meta"].c_str());
+                return 0;
+            }
+            auto devinfos = collect_devices({ real_dev });
+            if (devinfos.size() == 0)
+                return 1;
+            std::vector<std::string> sizes;
+            sizes.push_back(std::to_string((dsk.meta_len+1024*1024-1)/1024/1024)+"MiB");
+            auto new_parts = add_partitions(devinfos[0], sizes);
+            if (!new_parts.array_items().size())
+                return 1;
+            options["move_meta"] = "/dev/disk/by-partuuid/"+new_parts[0]["uuid"].string_value();
+        }
+        else if (options["move_meta"].substr(0, 22) != "/dev/disk/by-partuuid/")
+        {
+            // Partitions should be identified by GPT partition UUID
+            fprintf(stderr, "%s does not start with /dev/disk/by-partuuid/. Partitions should be identified by GPT partition UUIDs\n", options["move_meta"].c_str());
+            return 1;
+        }
+        else
+        {
+            // already a partition - check that it's a GPT partition with correct type
+            if (options.find("force") == options.end() &&
+                check_existing_partition(real_dev) != 0)
+            {
+                return 1;
+            }
+        }
+        move_options["new_meta_len"] = std::to_string(get_device_size(options["move_meta"]) - 4096);
+        move_options["new_meta_device"] = options["move_meta"];
+        move_options["new_meta_offset"] = "4096";
+        if (dsk.meta_device == dsk.data_device)
+            new_data_offset -= dsk.meta_len;
+    }
+    return 0;
+}

src/disk_tool/disk_tool_udev.cpp

@@ -382,6 +382,34 @@ int disk_tool_t::pre_exec_osd(std::string device)
     return 0;
 }
 
+int disk_tool_t::clear_osd_superblock(const std::string & dev)
+{
+    uint8_t *buf = (uint8_t*)memalign_or_die(MEM_ALIGNMENT, 4096);
+    int fd = -1, r = open(dev.c_str(), O_DIRECT|O_RDWR);
+    if (r >= 0)
+    {
+        fd = r;
+        r = read_blocking(fd, buf, 4096);
+        if (r == 4096)
+        {
+            // Clear magic and CRC
+            memset(buf, 0, 12);
+            r = lseek64(fd, 0, 0);
+            if (r == 0)
+            {
+                r = write_blocking(fd, buf, 4096);
+                if (r == 4096)
+                    r = 0;
+            }
+        }
+    }
+    if (fd >= 0)
+        close(fd);
+    free(buf);
+    buf = NULL;
+    return r;
+}
+
 int disk_tool_t::purge_devices(const std::vector<std::string> & devices)
 {
     std::set<uint64_t> osd_numbers;
@@ -440,7 +468,6 @@ int disk_tool_t::purge_devices(const std::vector<std::string> & devices)
         return 1;
     }
     // Destroy OSD superblocks
-    uint8_t *buf = (uint8_t*)memalign_or_die(MEM_ALIGNMENT, 4096);
     for (auto & sb: superblocks)
     {
         for (auto dev_type: std::vector<std::string>{ "data", "meta", "journal" })
@@ -448,26 +475,7 @@ int disk_tool_t::purge_devices(const std::vector<std::string> & devices)
             auto dev = sb["real_"+dev_type+"_device"].string_value();
             if (dev != "")
             {
-                int fd = -1, r = open(dev.c_str(), O_DIRECT|O_RDWR);
-                if (r >= 0)
-                {
-                    fd = r;
-                    r = read_blocking(fd, buf, 4096);
-                    if (r == 4096)
-                    {
-                        // Clear magic and CRC
-                        memset(buf, 0, 12);
-                        r = lseek64(fd, 0, 0);
-                        if (r == 0)
-                        {
-                            r = write_blocking(fd, buf, 4096);
-                            if (r == 4096)
-                                r = 0;
-                        }
-                    }
-                }
-                if (fd >= 0)
-                    close(fd);
+                int r = clear_osd_superblock(dev);
                 if (r != 0)
                 {
                     fprintf(stderr, "Failed to clear OSD %ju %s device %s superblock: %s\n",
@@ -517,7 +525,5 @@ int disk_tool_t::purge_devices(const std::vector<std::string> & devices)
             }
         }
     }
-    free(buf);
-    buf = NULL;
     return 0;
 }

src/disk_tool/disk_tool_upgrade.cpp

@@ -101,7 +101,7 @@ int disk_tool_t::upgrade_simple_unit(std::string unit)
         resizer.options = options;
         for (auto & kv: resize)
             resizer.options[kv.first] = std::to_string(kv.second);
-        if (resizer.resize_data() != 0)
+        if (resizer.raw_resize() != 0)
         {
             // FIXME: Resize with backup or journal
             fprintf(