kudu git commit: KUDU-2083. Decrement running maintenance ops on failed prepare

[ad...@apache.org]

Repository: kudu
Updated Branches:
  refs/heads/branch-1.2.x 62b96231b -> 7b6755d3a


KUDU-2083. Decrement running maintenance ops on failed prepare

There is currently a bug where we don't decrement the number of
running ops when an op->Prepare() fails. Although rare, when this
bug is hit, it will decrease the number of simultaneous mm ops
that can run until none can, causing the tserver to run OOM.

Change-Id: I8022bcd4c6470dfef2dece0cbefede916a752291
Reviewed-on: http://gerrit.cloudera.org:8080/7610
Reviewed-by: Todd Lipcon <to...@apache.org>
Tested-by: Kudu Jenkins
(cherry picked from commit b365ad0bd6372d459f547da0f1cb82e36148c541)
Reviewed-on: http://gerrit.cloudera.org:8080/7970


Project: http://git-wip-us.apache.org/repos/asf/kudu/repo
Commit: http://git-wip-us.apache.org/repos/asf/kudu/commit/7b6755d3
Tree: http://git-wip-us.apache.org/repos/asf/kudu/tree/7b6755d3
Diff: http://git-wip-us.apache.org/repos/asf/kudu/diff/7b6755d3

Branch: refs/heads/branch-1.2.x
Commit: 7b6755d3a2ceae01e19f163d155304a33b2b9368
Parents: 62b9623
Author: David Alves <dr...@apache.org>
Authored: Mon Aug 7 15:42:34 2017 -0700
Committer: Todd Lipcon <to...@apache.org>
Committed: Wed Sep 6 17:37:55 2017 +0000

----------------------------------------------------------------------
 src/kudu/util/maintenance_manager.cc | 1 +
 1 file changed, 1 insertion(+)
----------------------------------------------------------------------


http://git-wip-us.apache.org/repos/asf/kudu/blob/7b6755d3/src/kudu/util/maintenance_manager.cc
----------------------------------------------------------------------
diff --git a/src/kudu/util/maintenance_manager.cc b/src/kudu/util/maintenance_manager.cc
index 6a50f6f..e56d10b 100644
--- a/src/kudu/util/maintenance_manager.cc
+++ b/src/kudu/util/maintenance_manager.cc
@@ -227,6 +227,7 @@ void MaintenanceManager::RunSchedulerThread() {
       LOG(INFO) << "Prepare failed for " << op->name()
                 << ".  Re-running scheduler.";
       op->running_--;
+      running_ops_--;
       op->cond_->Signal();
       continue;
     }