osd/OSDMapMapping: fix spurious threadpool timeout errors

We were passing a grace of zero seconds to our temporary work queue, which led to the HeartbeatMap issuing cpu_tp timeout errors to the log. By using a non-zero grace period we can avoid these. Use the same default grace we use for the workqueue itself when it goes to sleep. Fixes: https://tracker.ceph.com/issues/53506 Signed-off-by: Sage Weil <[email protected]>
thomasgoirand · Dec 6, 2021 · 30ac5e7 · 30ac5e7
1 parent cf829d0
commit 30ac5e7
Show file tree

Hide file tree

Showing 2 changed files with 8 additions and 5 deletions.
diff --git a/src/common/WorkQueue.cc b/src/common/WorkQueue.cc
@@ -98,7 +98,9 @@ void ThreadPool::worker(WorkThread *wt)
       break;
     }
 
-    if (!_pause && !work_queues.empty()) {
+    if (work_queues.empty()) {
+      ldout(cct, 10) << "worker no work queues" << dendl;
+    } else if (!_pause) {
       WorkQueue_* wq;
       int tries = 2 * work_queues.size();
       bool did = false;

diff --git a/src/osd/OSDMapMapping.h b/src/osd/OSDMapMapping.h
@@ -115,10 +115,11 @@ class ParallelPGMapper {
     ParallelPGMapper *m;
 
     WQ(ParallelPGMapper *m_, ThreadPool *tp)
-      : ThreadPool::WorkQueue<Item>("ParallelPGMapper::WQ",
-				    ceph::timespan::zero(),
-				    ceph::timespan::zero(),
-				    tp),
+      : ThreadPool::WorkQueue<Item>(
+	"ParallelPGMapper::WQ",
+	ceph::make_timespan(m_->cct->_conf->threadpool_default_timeout),
+	ceph::timespan::zero(),
+	tp),
         m(m_) {}
 
     bool _enqueue(Item *i) override {