All The Tropes Wiki Project, Part XXV

[robkelk]

I am really starting to worry about Miraheze.  I know the situation has stabilized for now, but after a truly lackluster incident report, I'm starting to lose confidence.  I raised my concerns about it over a week ago, including the lack of regular feedback and no finding on whether the outage was preventable, and there hasn't been any response.  If there really was nothing to be learned and no actions to be taken after a week-long outage, I think we need to make sure we're getting regular backups, in case the worst happens.

Just read that Incident Report early draft. If I submitted something like this to my boss, I'd be asked by her boss' boss when the real report will be finished.

I am extremely tempted to enter this as a Phabricator task

The Incident Report at https://meta.miraheze.org/wiki/Special:I...Reports/51 regarding the recent 11388-minute outage (Down to Up) says "Gluster has not been really been fined tuned for our setup so probably made things worse."

This is the only thing identified in the incident report as being a possible contributing factor (although for some reason it was not labelled as such).

Please fine-tune Gluster, so that the next such outage (Down to Up) does not last for 7.9+ days.

But perhaps that should come from somebody with a good working relationship with the techs.

A week and a half later, and I'm wondering whether doing this would cut down on the chronic 50* errors we're still getting. (I am not a DBA.)