[SERVER-56731] Consider running server tests on newer SLES 12 service pack Created: 06/May/21  Updated: 29/Oct/23  Resolved: 19/Oct/21

Status: Closed
Project: Core Server
Component/s: None
Affects Version/s: None
Fix Version/s: 5.0.8, 5.2.0, 5.1.0-rc2

Type: Task Priority: Major - P3
Reporter: Max Hirschhorn Assignee: Zakhar Kleyman
Resolution: Fixed Votes: 0
Labels: dp-qp-stakeholder-request-2021-07, dp-qp-stakeholder-request-2021-10
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Backports
Depends
Documented
is documented by DOCS-14878 [SERVER] Investigate changes in SERVE... Closed
Related
related to SERVER-72294 Upgrade all uses of sus12 to suse12-sp5 Closed
Backwards Compatibility: Minor Change
Backport Requested:
v5.1, v5.0, v4.2
Sprint: Build 2021-05-17, Build 2021-05-31, Build 2021-10-18
Participants:
Linked BF Score: 155
Story Points: 2

 Description   

There have been several hangs observed in Evergreen when running on the suse12-large distro related to pthread_mutex_lock waiting to acquire a mutex which the core dump reports as not being owned by any thread. It is strongly suspected to be an issue with the platform itself given that we haven't observed similar hangs on any other Linux distro. This has led to duplicative effort across multiple subteams to diagnose what are ultimately believed to be false positives and has led to an increase in distrust of server hangs reported on SLSE 12 in general.

We appear to be running tests on SLES 12 (SP0) which has been unsupported since 01 July 2019. SLES 12 SP3 appears to be the oldest currently supported release of SLES 12. https://www.suse.com/lifecycle

> cat /etc/SuSE-release
SUSE Linux Enterprise Server 12 (x86_64)
VERSION = 12
PATCHLEVEL = 0

My understanding is that we intentionally compile on unsupported versions to avoid breaking our minimum supported version. This doesn't mean we must also test on those same unsupported versions, especially when there is evidence to suggest that an issue exists with the platform itself. My request is that we run server tests on a newer SLES 12 service pack in hopes it addresses the pthread_mutex_lock issues we've seen.



 Comments   
Comment by Githook User [ 07/Apr/22 ]

Author:

{'name': 'Zakhar Kleyman', 'email': 'zakhar.kleyman@mongodb.com', 'username': 'zakhark'}

Message: SERVER-56731 use suse12-sp5 for suse12 builds
Branch: v5.0
https://github.com/mongodb/mongo/commit/509f187473262cd3b0e87f66eb96ea6176be60ba

Comment by Githook User [ 19/Oct/21 ]

Author:

{'name': 'Zakhar Kleyman', 'email': 'zakhar.kleyman@mongodb.com', 'username': 'zakhark'}

Message: SERVER-56731 use suse12-sp5 for suse12 builds
Branch: v5.1
https://github.com/mongodb/mongo/commit/66e2c63be90ad90c3a490ff713580aa8f8aaf8e0

Comment by Githook User [ 18/Oct/21 ]

Author:

{'name': 'Zakhar Kleyman', 'email': 'zakhar.kleyman@mongodb.com', 'username': 'zakhark'}

Message: SERVER-56731 use suse12-sp5 for suse12 builds
Branch: master
https://github.com/mongodb/mongo/commit/2c07044c75164eebcfbfafff608cc6230a76db60

Comment by John Chen (Inactive) [ 13/May/21 ]

Hey max.hirschhorn, did you have any thoughts about this?

Comment by John Chen (Inactive) [ 11/May/21 ]

We have a suse12-sp5 variant available in Evergreen already. This is the variant that the Build team supports as the original suse12 variant is no longer able to be build (original AMI is gone).
I created a patch with SUSE 12 and Enterprise SLES 12 with the suse12-sp5 variant: https://spruce.mongodb.com/version/60994dcc3627e02d622fdb10/tasks?sorts=STATUS%3AASC%3BBASE_STATUS%3ADESC
but it looks like it also has issues with timing out. Not sure if using a newer service pack will fix your issue.

Generated at Thu Feb 08 05:40:03 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.