Debugging T90384 showed that after reboots there can be a (short) period where user name look-ups are failing. That task suggests that this may cause bigbrother to not honour .bigbrotherrcs, i. e. do nothing.
It needs to be tested what effects such failing look-ups have on bigbrother's data structures, and if necessary, fix it.
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | scfc | T90384 Restart webservice for /magnustools/ | |||
| Declined | None | T91414 make bigbrother or its replacement reliable | |||
| Invalid | scfc | T90410 Test how bigbrother reacts to user names not resolving and, if necessary, fix it |
I had to restart bigbrother again just now after the reboot of tools-submit so this suggests that this wasn't a one-time fluke.
I tested this on toolsbeta-exec-01 by blocking ldap and ldaps in iptables and running nscd -i passwd to remove the cache:
scfc@toolsbeta-exec-01:~$ perl -wMData::Dumper -e 'print Dumper getpwnam "scfc";' scfc@toolsbeta-exec-01:~$ perl -wMData::Dumper -e 'print Dumper undef;' $VAR1 = undef; scfc@toolsbeta-exec-01:~$ perl -wMData::Dumper -e 'print Dumper ();' scfc@toolsbeta-exec-01:~$
So getpwnam indeed does what the manual says :-). So now to debug what effect this has on bigbrother's code paths.
I still think / hope to have completely gotten rid of bigbrother by
end of the month :D
I hope that, too, but until then … :-) Especially, glancing over the code it isn't obvious to me why a failed look-up would block those users weeks after.