moderated Event: Database upgrade (try #2) #downtime - Wednesday, October 5, 2022 #downtime #cal-invite


Hi All,

The site was down for about 35 minutes of the one hour planned downtime on Wednesday. Unfortunately, I ran into another snag and was unable to complete the switchover to the new database cluster. I've spent the last two days trying to isolate the problem. I think I have figured it out, but I need to do a bunch of testing.

Right now I'm targeting Saturday, October 22nd at 9am Pacific Time for the (hopefully final attempt at the) switchover. I will post again as that date gets closer.

I don't think I shared the motivation for why I'm spending so much time working to upgrade the database cluster. Right now, we're running Postgresql 9.6, which at this point is a very old version of Postgres and is no longer being maintained. It's important that the database software we use is up to date, so that we can benefit from any bug or security fixes. Version 9.6 is no longer receiving those.

Ensuring that the switchover happens with the absolute least amount of risk and downtime is a large project. Fortunately, the work I've done towards this means that future database upgrades will require much less work.


Group Notification <noreply@...>

Database upgrade (try #2) #downtime

Wednesday, October 5, 2022
9:00am to 10:00am
(UTC-07:00) America/Los Angeles

View Event


The final step of the database cluster upgrade to the latest version of Postgres will require some downtime as we switch over from the old master database to the new master. I am scheduling one hour for the downtime, although I anticipate that it will be much shorter than that. This is the second try at switching over to the new database cluster; the first time we tried to switch over a problem was discovered. That has since been fixed.

I will put up a banner on the website on Monday, October 3rd, announcing the upcoming downtime.

Email sent during this time will be queued and sent when the site returns.