fapifta commented on a change in pull request #719: HDDS-3270. Allow safemode
listeners to be notified when some precheck rules pass
URL: https://github.com/apache/hadoop-ozone/pull/719#discussion_r400819244
##########
File path:
hadoop-hdds/server-scm/src/main/java/org/apache/hadoop/hdds/scm/pipeline/SCMPipelineManager.java
##########
@@ -641,13 +663,26 @@ public boolean getSafeModeStatus() {
}
@Override
- public void handleSafeModeTransition(
+ public synchronized void handleSafeModeTransition(
SCMSafeModeManager.SafeModeStatus status) {
- this.isInSafeMode.set(status.getSafeModeStatus());
- if (!status.getSafeModeStatus()) {
- // TODO: #CLUTIL if we reenter safe mode the fixed interval pipeline
- // creation job needs to stop
+ // TODO: #CLUTIL - handle safemode getting re-enabled
+ boolean currentAllowPipelines =
+ allowPipelineCreation.getAndSet(status.isPreCheckComplete());
+ boolean currentlyInSafeMode =
+ isInSafeMode.getAndSet(status.isInSafeMode());
+ boolean triggerPipelines = false;
+
+ // Trigger pipeline creation only if the preCheck status has changed to
+ // complete.
+ if (allowPipelineCreation.get() && !currentAllowPipelines) {
+ triggerPipelines = true;
+ }
+ // Start the pipeline creation thread only when safemode switches off
+ if (!getSafeModeStatus() && currentlyInSafeMode) {
startPipelineCreator();
+ triggerPipelines = true;
+ }
+ if (triggerPipelines) {
Review comment:
Oh, I missed one small detail in the shouldSchedulePipelineCreator method...
and misinterpreted it sorry about that... So triggerPipelineCreation will
create pipelines if the safemode precheck is finished and allowPipelineCreation
is already true in SCMPipelineManager.
With that it makes perfect sense, that we want to trigger the pipeline
creation once when precheck is complete, and start the scheduled runs after the
safe mode is exited.
But in this case, it would be enough to call triggerPipelineCreation when we
arrived to the end of the precheck stage, and start the background pipeline
creator schedule when we left safe mode.
Now what I understand happens at a startup when this is interesting is the
following:
for (0..nodecount) {
triggerPipelineCreation() // noop as allowPipelineCreation is false
}
precheck finished
triggerPipelineCreation // runs and creates pipelines
safe mode exit
startFixedIntervalPipelineCreator // calls createPipelines right away once
at start
triggerPipelineCreation // tries to call createPipelines if it is not
running from the prev. call
I think we don't need this last call, which makes the triggerPipelines
boolean unnecessary and the code a bit simpler. Is it correct, or I am still
missing something? :)
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: ozone-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: ozone-issues-h...@hadoop.apache.org
