[14:55:17] volans: have you seen dbstore1001's disk space alert? [14:56:22] paravoid: sure, it's kinda expected, I didn't want to ack blindly though (non-sticky of course) [14:57:47] and still didn't had time today to look at it properly, but it should be because of the backups of yesterday night (weekly) [14:59:36] nod :) [15:00:40] paravoid: I just checked that it was stable today and not growing, so I decided was not more urgent than the other stuff ;) [15:00:52] but if it's sending email/alert to someone I'll ack-it [15:01:17] nah, it's fine [21:13:41] volans: Do we have stats about database writes (or more specifically updates)? [21:14:04] Number of would be enough already, I presume [21:14:05] yes we do [21:15:47] hoo: what do you need? [21:16:21] volans: I'm going to deploy a patch taht should significantly reduce the number of UPDATEs within the cluster [21:16:29] some stats about that would be nice [21:16:37] not going to the Wikipedias today, but tomorrow [21:16:45] we should see a significant drop [21:16:47] on some particular shard? [21:16:56] probably mostly s3 today [21:18:19] ok, we have metrics for updates/deletes/insert/replace on tendril at 5m [21:21:00] and for some hosts on ganglia too, s3 master is one of them [21:34:35] hoo: I think I'm going to head off to bed soon, do you need anything from me now? [21:35:14] volans: No, should be fine [21:35:19] good night [21:35:52] ok, thanks, in particular on behalf of the databases if you reduce the updates :-) [21:36:47] feel free to ping me when you'll deploy the update if you want me to take a look at the metrics or where to find them [21:37:47] I found them in the mean time on tendril [21:38:08] look for com_update [21:38:23] above the graphs there is the link to ganglia [21:38:37] for db1038 it seems to have proper data [21:39:00] and has a more detailed time-frame