perfSONAR User Q&A and Other Discussion

[Mark Feit <>]

George Uhl writes:

 

I’m getting major apache errors since last night’s auto upgrade to pS 4.3:

 

SERVERS A and B: 

Nov 03 14:37:44 nasatest2.asf.alaska.edu httpd[28528]: pscheduler-api ERROR    Internal error xxx.xxx.xxx.xxx POST https://<FQDN>/pscheduler/t...liseconds

…

[Tue Nov 03 17:30:22.100392 2020] [wsgi:error] [pid 16748] [remote 137.229.86.189:42812] ImportError: No module named crontab

 

That looks like an upgrade failure.  If you’re on CentOS, you should have a RPM named python3-parse-crontab installed.  That’s a new package that wasn’t present in 4.2, and I’m not seeing it missing in our testbed or a handful of production hosts.  If you’re on Debian, I’ll have to pass you over to Antoine, who runs that corner of the show.

 

 SERVER C (after systemctl restart httpd):

Nov 03 15:17:44 enpl httpd[9185]: AH00558: httpd: Could not reliably determine the server's fully qualified domain name, using fe80::20a:f7ff:fe21:a62c. Set the 'ServerName' dir... this message

Nov 03 15:17:44 enpl httpd[9185]: httpd (pid 40510) already running

Nov 03 15:17:44 enpl systemd[1]: Failed to start The Apache HTTP Server.

Nov 03 15:17:44 enpl systemd[1]: Unit httpd.service entered failed state.

Nov 03 15:17:44 enpl systemd[1]: httpd.service failed.

 

 

We don’t do anything unusual with Apache other than starting and stopping it (and, of course, adding config).  Did the system get itself into a state where the Apache PID file didn’t get erased and the PID got re-used for something else?

 

The AH00558 error is harmless and is standard fare.  We don’t configure ServerName, and there isn’t a need to.

 

--Mark