#038

More fixing of Elasticsearch issues. The biggest was replacing the now-removed indices query, but now that that’s done we only have one thing to do: changing the text similarity metric. Replacing the indices query was a bit tricky, because the recommended migration has had a known problem since February 2017. It’s a bit disappointing that a feature was removed when the suggested alternative had a significant flaw, and I implemented a fairly nasty work-around to preserve the behaviour we wanted.

We had a call with AWS about Elasticsearch, and talked through what sort of things we currently do and want to do. We’re going to have another call with some of their “solutions architects” to make sure our cluster architecture is the best choice for our needs, and also to talk through our upgrade to Elasticsearch 6.

We had an incident with access control, where the GitHub teams which we use for authenticating into Jenkins and the like got accidentally deleted. So there was a fun hour of manually adding people back to the teams and quickly hacking together a script to sort out the hundreds of repositories we have. This has apparently happened twice before, done by two different (now three different) people, so it seems the UI for removing someone from a team on GitHub isn’t great.

A long time ago now I applied to be a senior developer, I had an interview and didn’t hear anything. Finally on Friday the results came out, and I’m now a senior.