Thread

1. Odd openbsd CI failure

   Andres Freund <andres@anarazel.de> — 2025-10-08T13:21:55Z

   Hi,
   
   A recent run of CI on master had this failure:
     https://cirrus-ci.com/task/5786871604707328
   on master, for openbsd.
   
   regress log:
     https://api.cirrus-ci.com/v1/artifact/task/5786871604707328/testrun/build/testrun/subscription/022_twophase_cascade/log/regress_log_022_twophase_cascade
   server log:
     https://api.cirrus-ci.com/v1/artifact/task/5786871604707328/testrun/build/testrun/subscription/022_twophase_cascade/log/022_twophase_cascade_node_C.log
   
   The relevant excerpt from the log seems to be:
   
   2025-10-06 07:36:47.556 UTC postmaster[14742] LOG:  database system is ready to accept connections
   2025-10-06 07:36:47.594 UTC unrecognized[36715] FATAL:  setsid() failed: Operation not permitted
   2025-10-06 07:36:47.602 UTC postmaster[14742] LOG:  WAL writer process (PID 36715) exited with exit code 1
   
   Severel interesting bits:
   a) why is setsid() failing
   b) why did we write "unrecognized" as the backend type?
   
   Greetings,
   
   Andres Freund
   
   
   
   

2. Re: Odd openbsd CI failure

   Melanie Plageman <melanieplageman@gmail.com> — 2025-11-20T17:16:37Z

   On Wed, Oct 8, 2025 at 9:22 AM Andres Freund <andres@anarazel.de> wrote:
   >
   > Severel interesting bits:
   > a) why is setsid() failing
   > b) why did we write "unrecognized" as the backend type?
   
   I just saw this again (causing a different test to fail).
   https://cirrus-ci.com/task/6409088281608192
   
   log from primary with same error:
   https://api.cirrus-ci.com/v1/artifact/task/6409088281608192/testrun/build/testrun/bloom/001_wal/log/001_wal_primary.log
   
   - Melanie