osd/scheduler: Reset ephemeral changes to mClock built-in profile

This is a follow-up to PR: https://github.com/ceph/ceph/pull/48703.
This commit also considers changes made ephemerally using either the
'daemon' or the 'tell' interfaces to override the built-in mClock
QoS parameters. In such a scenario, the ephemeral changes are removed
using the rm_val() method exposed by the config subsytem and logging
this information.

Other changes:

1. Add a standalone test to exercise the fix.
2. Add documentation note on the outcome of the attempt to modify
   built-in profile defaults.

Fixes: https://tracker.ceph.com/issues/61155
Signed-off-by: Sridhar Seshasayee <sseshasa@redhat.com>
(cherry picked from commit 414ac7dd2ce48d2a31f798e8df09bd3c975e95b1)

diff --git a/doc/rados/configuration/mclock-config-ref.rst b/doc/rados/configuration/mclock-config-ref.rst
index cb607a326f19ec713aebf978776d3b92a1917e7d..e49391698507dcbe2f93d3045d8e25876ccf5028 100644 (file)
--- a/doc/rados/configuration/mclock-config-ref.rst
+++ b/doc/rados/configuration/mclock-config-ref.rst
@@ -177,6 +177,11 @@ in order to ensure mClock scheduler is able to provide predictable QoS.
 
 mClock Config Options
 ---------------------
+.. important:: These defaults cannot be changed using any of the config
+   subsytem commands like *config set* or via the *config daemon* or *config
+   tell* interfaces. Although the above command(s) report success, the mclock
+   QoS parameters are reverted to their respective built-in profile defaults.
+
 When a built-in profile is enabled, the mClock scheduler calculates the low
 level mclock parameters [*reservation*, *weight*, *limit*] based on the profile
 enabled for each client type. The mclock parameters are calculated based on

diff --git a/qa/standalone/misc/mclock-config.sh b/qa/standalone/misc/mclock-config.sh
index bc5473114a95959c9532b18c11ad54c9f7ce9f63..59f002584d9da2da41bd828c51522485e42b4fe4 100755 (executable)
--- a/qa/standalone/misc/mclock-config.sh
+++ b/qa/standalone/misc/mclock-config.sh
@@ -380,6 +380,85 @@ function TEST_profile_disallow_builtin_params_modify() {
     teardown $dir || return 1
 }
 
+function TEST_profile_disallow_builtin_params_override() {
+    local dir=$1
+
+    setup $dir || return 1
+    run_mon $dir a || return 1
+    run_mgr $dir x || return 1
+
+    run_osd $dir 0 --osd_op_queue=mclock_scheduler || return 1
+
+    # Verify that the default mclock profile is set on the OSD
+    local def_mclock_profile=$(ceph config get osd.0 osd_mclock_profile)
+    test "$def_mclock_profile" = "balanced" || return 1
+
+    # Verify the running mClock profile
+    local cur_mclock_profile=$(CEPH_ARGS='' ceph --format=json daemon \
+      $(get_asok_path osd.0) config get osd_mclock_profile |\
+      jq .osd_mclock_profile)
+    cur_mclock_profile=$(eval echo $cur_mclock_profile)
+    test $cur_mclock_profile = "high_recovery_ops" || return 1
+
+    declare -a options=("osd_mclock_scheduler_background_recovery_res"
+      "osd_mclock_scheduler_client_res")
+
+    local retries=10
+    local errors=0
+    for opt in "${options[@]}"
+    do
+      # Override a mclock config param and confirm that no change occurred
+      local opt_val_orig=$(CEPH_ARGS='' ceph --format=json daemon \
+        $(get_asok_path osd.0) config get $opt | jq .$opt | bc)
+      local opt_val_new=$(echo "$opt_val_orig + 0.1" | bc -l)
+      ceph tell osd.0 config set $opt $opt_val_new || return 1
+
+      # Check configuration values
+      for count in $(seq 0 $(expr $retries - 1))
+      do
+        errors=0
+        sleep 2 # Allow time for changes to take effect
+
+        echo "Check configuration values - Attempt#: $count"
+        # Check configuration value on Mon store (or the default) for the osd
+        local res=$(ceph config get osd.0 $opt) || return 1
+        echo "Mon db (or default): osd.0 $opt = $res"
+        if (( $(echo "$res == $opt_val_new" | bc -l) )); then
+          errors=$(expr $errors + 1)
+        fi
+
+        # Check running configuration value using "config show" cmd
+        res=$(ceph config show osd.0 | grep $opt |\
+          awk '{ print $2 }' | bc ) || return 1
+        echo "Running config: osd.0 $opt = $res"
+        if (( $(echo "$res == $opt_val_new" | bc -l) || \
+              $(echo "$res != $opt_val_orig" | bc -l)  )); then
+          errors=$(expr $errors + 1)
+        fi
+
+        # Check value in the in-memory 'values' map is unmodified
+        res=$(CEPH_ARGS='' ceph --format=json daemon $(get_asok_path \
+          osd.0) config get $opt | jq .$opt | bc)
+        echo "Values map: osd.0 $opt = $res"
+        if (( $(echo "$res == $opt_val_new" | bc -l) || \
+              $(echo "$res != $opt_val_orig" | bc -l) )); then
+          errors=$(expr $errors + 1)
+        fi
+
+        # Check if we succeeded or exhausted retry count
+        if [ $errors -eq 0 ]
+        then
+          break
+        elif [ $count -eq $(expr $retries - 1) ]
+        then
+          return 1
+        fi
+      done
+    done
+
+    teardown $dir || return 1
+}
+
 main mclock-config "$@"
 
 # Local Variables:

diff --git a/src/osd/scheduler/mClockScheduler.cc b/src/osd/scheduler/mClockScheduler.cc
index 4f56a8e48177a7de0ea55298678c8c519095a7ae..4b30766796c6d7897197c1884ada6572f720211c 100644 (file)
--- a/src/osd/scheduler/mClockScheduler.cc
+++ b/src/osd/scheduler/mClockScheduler.cc
@@ -539,6 +539,12 @@ void mClockScheduler::handle_conf_change(
         }
       }
     }
+    // Alternatively, the QoS parameter, if set ephemerally for this OSD via
+    // the 'daemon' or 'tell' interfaces must be removed.
+    if (!cct->_conf.rm_val(*key)) {
+      dout(10) << __func__ << " Restored " << *key << " to default" << dendl;
+      cct->_conf.apply_changes(nullptr);
+    }
   }
 }