Opened 5 days ago

Last modified 5 days ago

#32968 new defect

Stem Jenkins CI failing with segfaults

Reported by: atagar Owned by: weasel
Priority: Medium Milestone:
Component: Internal Services/Service - jenkins Version:
Severity: Normal Keywords:
Cc: Actual Points:
Parent ID: Points:
Reviewer: Sponsor:

Description

Hi lovely sysadmins. Stem's Jenkins CI is frequently failing with segmentation faults...

10:35:36 Running tests...
10:35:36 
10:35:36   util.conf...                                         success (0.00s)
10:35:36   util.connection...                                   success (0.00s)
10:35:36   util.proc...                                         success (0.00s)
10:35:37   util.system...                                       success (7.65s)
10:35:44   interpreter...                                       success (0.00s)
10:35:44   version...                                           success (0.07s)
10:35:44   manual...                                            success (0.00s)
10:35:44   directory.authority...                               success (0.00s)
10:35:44   directory.fallback...                                success (0.00s)
10:35:44   client.connection...                                Segmentation fault
10:35:44 + rc=139
10:35:44 + echo '=== [build-wrapper] end ==='
10:35:44 === [build-wrapper] end ===

Nick and I both attempted to reproduce this locally with the latest repos (stem commit 86d5578, tor commit 5e70c27) and they pass.

Per chance could this be related to our recent CI changes for #32879? It's unclear to me if python or tor is segfaulting. Maybe installing a different python 3.x version will help? Unfortunately without a local repro we're a bit stuck on how to resolve this.

On a side note, if Jenkins installs git we will be able to confirm what Jenkins is testing. Its test output has...

======================================================================
10:35:30                              INITIALISING                             
10:35:30 ======================================================================
10:35:30 
10:35:30   stem version...                            1.8.0-dev
10:35:30   tor version...                             0.4.3.0-alpha-dev
10:35:30   python version...                          3.7.3
10:35:30   operating system...                        Linux (debian 10.2)

... whereas if git is included the git versions will be added. Here's what I get locally...

======================================================================
                             INITIALISING                             
======================================================================

  stem version...                            1.8.0-dev (commit 86d55785)
  tor version...                             0.4.3.0-alpha-dev (commit 5e70c27e)
  python version...                          3.7.0
  operating system...                        Linux (debian stretch/sid)

Thanks!

Child Tickets

Change History (1)

comment:1 Changed 5 days ago by weasel

I wasn't really able to reproduce this yet. Will keep looking.

I came across at least one other issue, though: The test assumes it can listen on ports 1111-1113. If that doesn't work, it fails.

Note: See TracTickets for help on using tickets.