From: DanWritesCode <github@dann.me>
Date: Mon, 18 Dec 2023 21:09:07 +0000 (-0500)
Subject: osd: add clear_shards_repaired command
X-Git-Tag: v20.0.0~732^2~1
X-Git-Url: http://git-server-git.apps.pok.os.sepia.ceph.com/?a=commitdiff_plain;h=78d6bfe54c3b9b60fab36a640b1ce77c8f022fa9;p=ceph.git

osd: add clear_shards_repaired command

This command will allow us to clear the OSD_TOO_MANY_REPAIRS alert
by setting the shard repair count to 0. This will help in cases where
the alert was a false positive, or a condition that has since cleared
at the disk level. Often, zeroing out the repair count is
better than muting the alert or restarting the OSD.

Fixes: https://tracker.ceph.com/issues/54182
Co-authored-by: David Zafman <dzafman@redhat.com>
Signed-off-by: Daniel Radjenovic <dradjenovic@digitalocean.com>
---

diff --git a/doc/rados/operations/health-checks.rst b/doc/rados/operations/health-checks.rst
index 54bfd427967..4059911654a 100644
--- a/doc/rados/operations/health-checks.rst
+++ b/doc/rados/operations/health-checks.rst
@@ -983,6 +983,13 @@ the object data, there might exist failing disks that are not registering any
 scrub errors. This repair count is maintained as a way of identifying any such
 failing disks.
 
+In order to allow clearing of the warning, a new command
+``ceph tell osd.# clear_shards_repaired [count]`` has been added.
+By default it will set the repair count to 0. A `count` value can be passed 
+to the command. Thus, the administrator has the option to re-enable the warning
+by passing the value of ``mon_osd_warn_num_repaired`` (or above) to the command.
+An alternative to using `clear_shards_repaired` is to mute the
+`OSD_TOO_MANY_REPAIRS` alert with `ceph health mute`.
 
 LARGE_OMAP_OBJECTS
 __________________
diff --git a/qa/standalone/osd/osd-rep-recov-eio.sh b/qa/standalone/osd/osd-rep-recov-eio.sh
index 6fea441b3a9..a34f4a47189 100755
--- a/qa/standalone/osd/osd-rep-recov-eio.sh
+++ b/qa/standalone/osd/osd-rep-recov-eio.sh
@@ -219,6 +219,18 @@ function TEST_rados_repair_warning() {
     ceph health | $(! grep -q "Too many repaired reads on 1 OSDs") || return 1
     set +o pipefail
 
+    ceph health unmute OSD_TOO_MANY_REPAIRS
+    ceph tell osd.$primary clear_shards_repaired
+    sleep 10
+
+    set -o pipefail
+    # Should clear this
+    ceph health | $(! grep -q "Too many repaired reads on 1 OSDs") || return 1
+    set +o pipefail
+
+    ceph tell osd.$primary clear_shards_repaired $OBJS
+    sleep 10
+
     for i in $(seq 1 $OBJS)
      do
        inject_$inject rep data $poolname ${objbase}-$i $dir 0 || return 1
@@ -235,7 +247,7 @@ function TEST_rados_repair_warning() {
     COUNT=$(ceph pg dump --format=json-pretty | jq ".pg_map.osd_stats_sum.num_shards_repaired")
     test "$COUNT" = "$(expr $OBJS \* 3)" || return 1
 
-    # Give mon a chance to notice additional OSD and unmute
+    # Give mon a chance to notice additional OSD and reset num_shards_repaired
     # The default tick time is 5 seconds
     CHECKTIME=10
     LOOPS=0
diff --git a/src/osd/OSD.cc b/src/osd/OSD.cc
index a14d70605f8..a3f887edaa0 100644
--- a/src/osd/OSD.cc
+++ b/src/osd/OSD.cc
@@ -1051,6 +1051,13 @@ void OSDService::inc_osd_stat_repaired()
   return;
 }
 
+void OSDService::set_osd_stat_repaired(int64_t count)
+{
+  std::lock_guard l(stat_lock);
+  osd_stat.num_shards_repaired = count;
+  return;
+}
+
 float OSDService::compute_adjusted_ratio(osd_stat_t new_stat, float *pratio,
 				         uint64_t adjust_used)
 {
@@ -3128,6 +3135,11 @@ will start to track new ops received afterwards.";
     scrub_purged_snaps();
   }
 
+  else if (prefix == "clear_shards_repaired") {
+    int64_t count = cmd_getval_or<int64_t>(cmdmap, "count", 0);
+    service.set_osd_stat_repaired(count);
+  }
+
   else if (prefix == "reset_purged_snaps_last") {
     lock_guard l(osd_lock);
     superblock.purged_snaps_last = 0;
@@ -4350,6 +4362,12 @@ void OSD::final_init()
     asok_hook,
     "debug the scrubber");
   ceph_assert(r == 0);
+  r = admin_socket->register_command(
+    "clear_shards_repaired "
+    "name=count,type=CephInt,req=false,range=0",
+    asok_hook,
+    "clear num_shards_repaired to clear health warning");
+  ceph_assert(r == 0);
 
   // -- pg commands --
   // old form: ceph pg <pgid> command ...
diff --git a/src/osd/OSD.h b/src/osd/OSD.h
index 83aed8a410b..f59d3ab7882 100644
--- a/src/osd/OSD.h
+++ b/src/osd/OSD.h
@@ -716,6 +716,7 @@ public:
     osd_alert_list_t& alerts);
   osd_stat_t set_osd_stat(std::vector<int>& hb_peers, int num_pgs);
   void inc_osd_stat_repaired(void);
+  void set_osd_stat_repaired(int64_t count);
   float compute_adjusted_ratio(osd_stat_t new_stat, float *pratio, uint64_t adjust_used = 0);
   osd_stat_t get_osd_stat() {
     std::lock_guard l(stat_lock);