Timeout Issue with bb8-redis 0.14.0 on GCP Memorystore Redis

[Hen.Lee]

Hi,

I've encountered an issue after updating to bb8-redis 0.14.0. Local tests showed no problems, but on GCP with Memorystore Redis, timeouts occur every 10-20 minutes for 2-3 users.

[Dirkjan Ochtman]

What is your config setup? In particular, do you customize the connection timeout?

Given the timing of your issue, I think this might be related to the recent bb8 0.8.2 update instead of the bb8-redis changes themselves (which only bumped the version of the upstream redis crate). bb8 0.8.2 included some fairly invasive changes from #186 in order to try and fix a potential deadlock issue. As part of this, the crate is now a little more strict about timing out for getting a connection (see in particular 016c164).

[Dirkjan Ochtman]

#187 identified an issue in 0.8.2 which might somehow be related to your issue, so maybe have a look at #189.

[Hen.Lee]

I'll take a look at the bb8 0.8.2 update and the related issues you mentioned, particularly #187 and #189. Since I haven't customized the connection timeout and I'm experiencing timeouts on GCP with Memorystore Redis, it does seem plausible that the recent bb8 changes could be the culprit.

I'll check out your suggestions and update you on any progress.

Thanks again for your help!

[Dirkjan Ochtman]

Please consider having your organization sponsor me on GitHub.

[Hen.Lee]

Please consider having your organization sponsor me on GitHub.

I need to check with my company first, so please wait a moment.