mon: Integrate discard queue overflow into pg health warnings

Added a health warning mechanism to monitor the discard queue for potential overload
Emits a warning if the accumulated discarded bytes in the queue exceed the configured threshold
Introduced a debugging tool to simulate slow discard operations by adding a configurable delay

Fixes : https://tracker.ceph.com/issues/69082

Signed-off-by: Jaya Prakash <jayaprakash@ibm.com>

diff --git a/src/mon/PGMap.cc b/src/mon/PGMap.cc
index 940b7ef9348939aac0e5cc9195b4ce73f42ef32e..486e38de7f6c3b9ce8ee2b3db11476b9536982c7 100644 (file)
--- a/src/mon/PGMap.cc
+++ b/src/mon/PGMap.cc
@@ -3313,6 +3313,10 @@ void PGMap::get_health_checks(
         summary += " experiencing stalled read in wal device of BlueFS";
       } else if (asum.first == "DB_DEVICE_STALLED_READ_ALERT") {
         summary += " experiencing stalled read in db device of BlueFS";
+      } else if (asum.first.find("_DISCARD_QUEUE") != std::string::npos) {
+       for (auto str : asum.second.second) {
+         summary += str;
+       }
       }
 
       auto& d = checks->add(asum.first, HEALTH_WARN, summary, asum.second.first);