>From Michael Blow <[email protected]>:

Attention is currently required from: Ali Alsuliman, Ali Alsuliman, Ian Maxon.

Michael Blow has posted comments on this change by Ian Maxon. ( 
https://asterix-gerrit.ics.uci.edu/c/asterixdb/+/20270?usp=email )

Change subject: [ASTERIXDB-3631][RT] Profile nested groupby clauses
......................................................................


Patch Set 8:

(1 comment)

File 
hyracks-fullstack/hyracks/hyracks-control/hyracks-control-common/src/main/java/org/apache/hyracks/control/common/job/profiling/StatsCollector.java:

https://asterix-gerrit.ics.uci.edu/c/asterixdb/+/20270/comment/b869a6d9_dcb9036e?usp=email
 :
PS8, Line 38:     private final Map<String, IOperatorStats> operatorStatsMap = 
new ConcurrentHashMap<>();
> i don't believe so. part of the reason the issue with missing stats was so 
> hard to find was because this structure is guarded by the locks around the 
> task, which establish the correct sequence of events chronologically. 
> however, that doesn't guarantee visibility of updates between threads, so 
> even though the operators had long ago added the stats to the map, the thread 
> managing the tasks can get a stale read.

I don't think this is right if I understand what you're describing as current 
behavior. If threadA holds lockX and updates the linked hash map, when threadB 
obtains lockX it is guaranteed to see all updates to the map made by threadA. 
The only way there would be missing stats, is if a thread not holding lockX 
attempts to read the map (and that in fact would be very unsafe, and could 
cause unpredictable results (like infinite loops).



--
To view, visit https://asterix-gerrit.ics.uci.edu/c/asterixdb/+/20270?usp=email
To unsubscribe, or for help writing mail filters, visit 
https://asterix-gerrit.ics.uci.edu/settings?usp=email

Gerrit-MessageType: comment
Gerrit-Project: asterixdb
Gerrit-Branch: master
Gerrit-Change-Id: I516dc90b8da3a7086dc80b67946ac14f6ade0973
Gerrit-Change-Number: 20270
Gerrit-PatchSet: 8
Gerrit-Owner: Ian Maxon <[email protected]>
Gerrit-Reviewer: Ali Alsuliman <[email protected]>
Gerrit-Reviewer: Ali Alsuliman <[email protected]>
Gerrit-Reviewer: Anon. E. Moose #1000171
Gerrit-Reviewer: Ian Maxon <[email protected]>
Gerrit-Reviewer: Jenkins <[email protected]>
Gerrit-CC: Michael Blow <[email protected]>
Gerrit-Attention: Ian Maxon <[email protected]>
Gerrit-Attention: Ali Alsuliman <[email protected]>
Gerrit-Attention: Ali Alsuliman <[email protected]>
Gerrit-Comment-Date: Wed, 17 Dec 2025 21:54:44 +0000
Gerrit-HasComments: Yes
Gerrit-Has-Labels: No
Comment-In-Reply-To: Ian Maxon <[email protected]>
Comment-In-Reply-To: Ali Alsuliman <[email protected]>

Reply via email to