Page MenuHomePhabricator

start testing mariadb 10.1.23 for fundraising
Closed, ResolvedPublic

Description

It's possible the replication lag issues we've been seeing in fundraising are due to a bug in 10.0.2x. Let's add a test box running 10.1.23, and see if we still see the problem.

Event Timeline

Do you have the bug report description you think you might be hitting?

Do you have the bug report description you think you might be hitting?

No--I don't have a specific reported bug in mind, and I'm not by any means convinced it's a mariadb bug.

Do you have the bug report description you think you might be hitting?

No--I don't have a specific reported bug in mind, and I'm not by any means convinced it's a mariadb bug.

Ah - sorry! I misunderstood the task description and I thought you had found a bug or a potential bug that might cause this issue.
I was curious/worried to see if it could be something that we could hit in production.

I would be surprise if it is a mariadb (or innodb) bug and we haven't seen it in production given the amount of hosts we have.

frdb1003 is up and running stretch with stock mariadb 10.1.23, we'll see how it does

frdb1003 is up and running stretch with stock mariadb 10.1.23, we'll see how it does

So far it behaves exactly as 10.0.x does in terms of our replication lag issue. We'll keep looking at it, but closing this ticket because we've completed its original intent.