[MCOL-3776] Follow up on vTpch22.sql hanging Created: 2020-02-10  Updated: 2022-02-14  Resolved: 2020-02-18

Status: Closed
Project: MariaDB ColumnStore
Component/s: ExeMgr
Affects Version/s: 1.4.2
Fix Version/s: 1.4.3

Type: Bug Priority: Major
Reporter: Patrick LeBlanc (Inactive) Assignee: Daniel Lee (Inactive)
Resolution: Fixed Votes: 0
Labels: None
Environment:

Centos 7

Attachments: File aCreateViews.sql     Text File mcol-3776-backtraces.txt     File vTpch22.sql    
Issue Links:
Issue split
split to MCOL-3792 All queries share a timezone Closed
Relates
relates to MCOL-4957 Functions on TIMESTAMP columns in pro... Closed
Sprint: 2020-2

 Description   

Validating the merge for the next 1.4 release. vTpch22.sql keeps hanging for me. Looks to me like there is a joblist abort problem. I think the joblist threadpool in exemgr happens to run out of threads at this query. The queries themselves run successfully after a restart.

 Comments   
Comment by Patrick LeBlanc (Inactive) [ 2020-02-11 ]

the link with mcol-3716 was a series of coincidences it seems. I've reproduced it in dev-1.4 @ commit 4b4198608, which doesn't include 3716.

Reproduced by running vTpch22.sql in a loop on a clean installation.

Comment by Patrick LeBlanc (Inactive) [ 2020-02-14 ]

Found & fixed the problem. The timezone var in the Functor class is read & written concurrently, but which had no sync. Added sync to prevent the crashes. Created MCOL-3792 to followup on other repercussions of that being a var shared between queries.

Comment by Patrick LeBlanc (Inactive) [ 2020-02-18 ]

i attached the files i was using to reproduce it. run aCreateViews.sql, then run vTpch22.sql in a loop 100x.

For me, I was able to reproduce the crash in < 10 iterations. Will be different for everybody.

Comment by Daniel Lee (Inactive) [ 2020-02-18 ]

Build verified: 1.4.3-1 BB nightly

engine commit:
1a65d34

1.4.2-1
executed vTpch.sql in a loop of 100. primproc failed after 9 runs.

1.4.3-1
ran to completion of 100 iterations.

Generated at Thu Feb 08 02:45:21 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.