core: skip crush rule cleanup if no pools exist yet
During the initial cluster setup, CleanUnusedCrushRules() runs after MGR startup but before any pools are created (if user does not create the .mgr pool along with the cephCluster manifest). It deletes the default replicated_pool since no pool references it. This leaves the cluster in a state if Zero crush rules. When the MGR tries to start the device_health_metris during this window, thne it fails to find the crush rule and crashes. Skip the cleanup when no pools exists yet, since there is nothing meaninful to cleanup. Signed-off-by: Santosh <sapillai@redhat.com>
This commit is contained in:
1 parent
eb9f5fe00a
commit
1c1d6cf261
2 files changed
+37
-1
No files matched your search
@@ -607,6 +607,15 @@ func CleanupUnusedCrushRules(context *clusterd.Context, clusterInfo *ClusterInfo
|
||||
return err
|
||||
}
|
||||
|
||||
// Skip cleanup when no pools exist yet. During initial cluster setup, this function
|
||||
// runs after MGR startup but before pools are created. Deleting the default
|
||||
// "replicated_rule" in this window leaves the cluster with zero CRUSH rules, causing
|
||||
// Ceph internal modules (e.g. devicehealth) to crash when they try to create pools.
|
||||
if len(usedRules) == 0 {
|
||||
logger.Infof("no crush rules are in use yet, skipping cleanup")
|
||||
return nil
|
||||
}
|
||||
|
||||
crushMap, err := GetCrushMap(context, clusterInfo)
|
||||
if err != nil {
|
||||
return err
|
||||
@@ -618,6 +627,10 @@ func CleanupUnusedCrushRules(context *clusterd.Context, clusterInfo *ClusterInfo
|
||||
logger.Debugf("crush rule %q is still in use", rule.Name)
|
||||
continue
|
||||
}
|
||||
if rule.Name == "replicated_rule" {
|
||||
logger.Debugf("skipping deletion of default crush rule %q", rule.Name)
|
||||
continue
|
||||
}
|
||||
logger.Infof("crush rule %q is unused and will be deleted", rule.Name)
|
||||
if err := deleteCrushRule(context, clusterInfo, rule.Name); err != nil {
|
||||
logger.Warningf("failed to delete unused crush rule %q, continuing with remaining rules. %v", rule.Name, err)
|
||||
|
||||
@@ -397,7 +397,8 @@ func TestCleanupUnusedCrushRules(t *testing.T) {
|
||||
{"rule_name":"mypool_zone","steps":[{},{"type":"zone"}]},
|
||||
{"rule_name":"mypool_zone_ssd","steps":[{},{"type":"zone","item_name":"default~ssd"}]},
|
||||
{"rule_name":"shared_rule","steps":[{"item_name":"default"},{"type":"host"}]},
|
||||
{"rule_name":"custom_rule","steps":[{"item_name":"default"},{"type":"zone"}]}
|
||||
{"rule_name":"custom_rule","steps":[{"item_name":"default"},{"type":"zone"}]},
|
||||
{"rule_name":"replicated_rule","steps":[{"item_name":"default"},{"type":"host"}]}
|
||||
]}`, nil
|
||||
}
|
||||
if command == "ceph" && args[1] == "crush" && args[2] == "rule" && args[3] == "rm" {
|
||||
@@ -413,6 +414,28 @@ func TestCleanupUnusedCrushRules(t *testing.T) {
|
||||
slices.Sort(removedRules)
|
||||
expected := []string{"custom_rule", "mypool", "mypool_zone"}
|
||||
assert.Equal(t, expected, removedRules)
|
||||
assert.NotContains(t, removedRules, "replicated_rule")
|
||||
}
|
||||
|
||||
func TestCleanupUnusedCrushRulesNoPools(t *testing.T) {
|
||||
removedRules := []string{}
|
||||
executor := &exectest.MockExecutor{}
|
||||
context := &clusterd.Context{Executor: executor}
|
||||
executor.MockExecuteCommandWithOutput = func(command string, args ...string) (string, error) {
|
||||
logger.Infof("Command: %s %v", command, args)
|
||||
if command == "ceph" && args[1] == "lspools" {
|
||||
return `[]`, nil
|
||||
}
|
||||
if command == "ceph" && args[1] == "crush" && args[2] == "rule" && args[3] == "rm" {
|
||||
removedRules = append(removedRules, args[4])
|
||||
return "", nil
|
||||
}
|
||||
return "", errors.Errorf("unexpected ceph command %q", args)
|
||||
}
|
||||
|
||||
err := CleanupUnusedCrushRules(context, AdminTestClusterInfo("mycluster"))
|
||||
assert.NoError(t, err)
|
||||
assert.Empty(t, removedRules)
|
||||
}
|
||||
|
||||
func TestExtractPoolDetails(t *testing.T) {
|
||||
|
||||
Reference in new issue
Block a user