xtdb 2020-08-31 | Slack Archive

Akshay C. Gollapalli19:08:58

I’m running into issues where I get db sync’s timing out on startup after a long stop. Using confluent cloud for Kafka. I understand that starting nodes after long gaps is probably not an expected use case, but it’s been one I find rather helpful for development. Is it likely there’s some sort of user error here on my part, or is it a bug? Thanks EDIT: for anyone who sees this, it was an issue with cached data. For anyone else who runs into it, try clearing out your .cpcache and any cached data from local dependencies.

refset21:08:07

I can't think of a reason why you might be having an issue, although I have found Confluent's endpoints can take a few seconds to "warm up" after a reasonable period of inactivity. Is the sync timeout message along the lines of "Timed out waiting for: ..."? (i.e. from https://github.com/juxt/crux/blob/f83f0ad775078777a7f421279b9be638c32ee4a6/crux-core/src/crux/node.clj#L60 )

Akshay C. Gollapalli21:08:58

Yeah, that’s the one.

👍 3

Akshay C. Gollapalli21:08:29

It only occurs on new nodes though. Or nodes that were shut off for a few days (say on the dev machine) while the cloud nodes were still up and running and sync just fine)

refset22:08:26

hmm, does it work if you pass a sufficiently large timeout as an arg to sync (not that you should need to)?

Akshay C. Gollapalli13:09:48

Afraid not. I gave it a 10 minute timeout. No dice.

Akshay C. Gollapalli14:09:09

You’re not gonna believe it. It was a .cpcache issue (or an issue with cached data in a local repo, not sure, as I cleared out both). Sorry to bother. Thanks for the help.

🙂 3

refset14:09:10

Ah - frustrating! I wonder if there's something we could add/change to help you know which (cached) code version you're using :thinking_face:

2020-08-31

Channels