On 07/12/13 05:04, roger peppe wrote: > I think that there are a few things that could help here, > most important points first: > > - We should limit agent restarting in some way (exponential backoff or > retry limits or both)
I think this is a key one. If the agent is repeatedly failing, we should back-off and ask for help :-) > - We should rotate log files and compress old ones. Yes. We have a pending proposal from Nate that adds in log rotation. > - We should have kind of policy for expiring and deleting old log files. The standard log rotation kind of handles this. You say things like: * Max of 100 MB * gzip all but the first * keep 5 backups This is fairly normal. > - We should have some way of garbage collecting the transaction log. This is our internal issue, but yes, we should consider this as a fairly high priority item. As I understand it, our transaction log is primarily for notifications and transactional-ability. What else do we use it for? How long is the log useful for? I know that full logs are useful for diagnosis, but probably not worthwhile keeping forever. Is it feasible for us to be able to export the transaction log to an external file, and then truncate the internal log? > We *could* consider disabling logging when the disk is tending > towards full, but I suspect that could make a bad problem worse > by losing any possibility of seeing what has actually been going > on. I think that if we manage the first several items here (backing off and log rotate), we shouldn't get into this problem. I'm -1 on attempting to disable logging on low disk. Tim -- Juju mailing list Juju@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/juju