ceph-objectstore-tool: add mark-complete operation

It is supposed to be used as a last resort to fix a cluster that has
PGs in 'incomplete' state, using the following procedure:

1) stop the osd that is primary for the incomplete PG;
2) run:
  ceph-objectstore-tool --data-path ... --journal-path ... --pgid $PGID --op mark-complete
3) start the osd.

Fixes: #10098
Signed-off-by: Mykola Golub <mgolub@mirantis.com>
(cherry picked from commit 6907778d767ba08bb80c495785056ed122b023fe)

Conflicts:
	src/test/ceph_objectstore_tool.py (trivial)
	src/tools/ceph_objectstore_tool.cc (trivial)

diff --git a/src/test/ceph_objectstore_tool.py b/src/test/ceph_objectstore_tool.py
index c875c0bcfdc30c8aab6014b82e212dc3996456cf..b18d5f94d865ea24cf329cd53aad539ffc2acb60 100755 (executable)
--- a/src/test/ceph_objectstore_tool.py
+++ b/src/test/ceph_objectstore_tool.py
@@ -791,7 +791,7 @@ def main(argv):
 
     # Specify a bad --op command
     cmd = (CFSD_PREFIX + "--op oops").format(osd=ONEOSD)
-    ERRORS += test_failure(cmd, "Must provide --op (info, log, remove, export, import, list, fix-lost, list-pgs, rm-past-intervals, set-allow-sharded-objects, dump-journal, dump-super)")
+    ERRORS += test_failure(cmd, "Must provide --op (info, log, remove, export, import, list, fix-lost, list-pgs, rm-past-intervals, set-allow-sharded-objects, dump-journal, dump-super, get-osdmap, set-osdmap, get-inc-osdmap, set-inc-osdmap, mark-complete)")
 
     # Provide just the object param not a command
     cmd = (CFSD_PREFIX + "object").format(osd=ONEOSD)

diff --git a/src/tools/ceph_objectstore_tool.cc b/src/tools/ceph_objectstore_tool.cc
index f98dfd80d31fdf30e03eda95a3075b0cd96915d1..e4ce3a19b041c8ecd305755e36898fb9b2ee2631 100644 (file)
--- a/src/tools/ceph_objectstore_tool.cc
+++ b/src/tools/ceph_objectstore_tool.cc
@@ -2737,10 +2737,10 @@ int main(int argc, char **argv)
     ("journal-path", po::value<string>(&jpath),
      "path to journal, mandatory for filestore type")
     ("pgid", po::value<string>(&pgidstr),
-     "PG id, mandatory for info, log, remove, export, rm-past-intervals")
+     "PG id, mandatory for info, log, remove, export, rm-past-intervals, mark-complete")
     ("op", po::value<string>(&op),
      "Arg is one of [info, log, remove, export, import, list, fix-lost, list-pgs, rm-past-intervals, set-allow-sharded-objects, dump-journal, dump-super, "
-        "get-osdmap, set-osdmap, get-inc-osdmap, set-inc-osdmap]")
+        "get-osdmap, set-osdmap, get-inc-osdmap, set-inc-osdmap, mark-complete]")
     ("epoch", po::value<unsigned>(&epoch),
      "epoch# for get-osdmap and get-inc-osdmap, the current epoch in use if not specified")
     ("file", po::value<string>(&file),
@@ -3144,7 +3144,8 @@ int main(int argc, char **argv)
   // The ops which require --pgid option are checked here and
   // mentioned in the usage for --pgid.
   if ((op == "info" || op == "log" || op == "remove" || op == "export"
-      || op == "rm-past-intervals") && pgidstr.length() == 0) {
+      || op == "rm-past-intervals" || op == "mark-complete") &&
+      pgidstr.length() == 0) {
     cerr << "Must provide pgid" << std::endl;
     usage(desc);
     ret = 1;
@@ -3414,9 +3415,9 @@ int main(int argc, char **argv)
 
   // If not an object command nor any of the ops handled below, then output this usage
   // before complaining about a bad pgid
-  if (!vm.count("objcmd") && op != "export" && op != "info" && op != "log" && op != "rm-past-intervals") {
+  if (!vm.count("objcmd") && op != "export" && op != "info" && op != "log" && op != "rm-past-intervals" && op != "mark-complete") {
     cerr << "Must provide --op (info, log, remove, export, import, list, fix-lost, list-pgs, rm-past-intervals, set-allow-sharded-objects, dump-journal, dump-super, "
-      "get-osdmap, set-osdmap, get-inc-osdmap, set-inc-osdmap)"
+      "get-osdmap, set-osdmap, get-inc-osdmap, set-inc-osdmap, mark-complete)"
         << std::endl;
     usage(desc);
     ret = 1;
@@ -3677,6 +3678,32 @@ int main(int argc, char **argv)
         fs->apply_transaction(*t);
         cout << "Removal succeeded" << std::endl;
       }
+    } else if (op == "mark-complete") {
+      ObjectStore::Transaction tran;
+      ObjectStore::Transaction *t = &tran;
+
+      if (struct_ver != PG::cur_struct_v) {
+       cerr << "Can't mark-complete, version mismatch " << (int)struct_ver
+            << " (pg)  != " << (int)PG::cur_struct_v << " (tool)"
+            << std::endl;
+       ret = 1;
+       goto out;
+      }
+
+      cout << "Marking complete " << std::endl;
+
+      info.last_update = eversion_t(superblock.current_epoch, info.last_update.version + 1);
+      info.last_backfill = hobject_t::get_max();
+      info.last_epoch_started = superblock.current_epoch;
+      info.history.last_epoch_started = superblock.current_epoch;
+      info.history.last_epoch_clean = superblock.current_epoch;
+      past_intervals.clear();
+
+      ret = write_info(*t, map_epoch, info, past_intervals);
+      if (ret == 0) {
+       fs->apply_transaction(*t);
+       cout << "Marking complete succeeded" << std::endl;
+      }
     } else {
       assert(!"Should have already checked for valid --op");
     }