Mike Kupfer <mike.kupfer at sun.com> writes: >>>>>> "stevel" == Stephen Lau <stevel at opensolaris.org> writes: > >>> remote: *** failed to import extension scmhooks.hg: No section: 'db' > > stevel> IIRC, that's from it not being able to read the [db] section of > stevel> /etc/tonic/common.ini to figure out the db-host to contact. > > Hmm. I got the same error again the following morning, but not since > then. The mctime and ctime for /etc/tonic/common.ini are both from last > April, so I'd expect the problem to be elsewhere. > > The two mornings when my cron job got an error, there were also several > messages about service problems from the automated service check > framework, culminating in an email from one of the admins saying that > she had restarted tomcat and Apache. > > I haven't looked into this completely, but it seems like maybe the init > code for one of the hooks is trying to contact the database?
I'd expect a better error message, but I guess it could be. > And if the hook is ignored when it fails to load, do we have a setup > where the gate would be wide open (no checking) every time there's a > problem with the database? ... only if the hooks involved needed the database, or otherwise failed to load if they couldn't get it. The hooks that do real checking should have no need of the database, or should fail safe if they can't reach the database. -- Rich
