Skip to content
This repository has been archived by the owner on Nov 7, 2020. It is now read-only.

Update redis on the server #76

Open
drewww opened this issue Aug 6, 2013 · 3 comments
Open

Update redis on the server #76

drewww opened this issue Aug 6, 2013 · 3 comments

Comments

@drewww
Copy link
Collaborator

drewww commented Aug 6, 2013

We're running an ancient version of redis on the unhangout server (2.2, latest is 2.6.7)

The problem is that unbuntu doesn't have more recent versions of redis in its repository, so installing a more recent version is a bit of a chore.

Our options here are:

  • update the server to 13.04 (from 12.04), which gives us access to a package that is in the 2.6.* branch
  • hand-install the server

Neither of these is that attractive. This issue isn't a burning one, but it seems like the older version might be at the root of #35. That issue has been somewhat resolved, but redis is still timing out our connection every 5 minutes which is annoying. I don't think it does it to connections that have active traffic, but it would still be nice to fix.

Plus, I'm sure there are features/fixes in newer redis that would be nice to have.

@drewww
Copy link
Collaborator Author

drewww commented Aug 6, 2013

Actually, this isn't that crucial. I've changed the timeout parameter in redis.conf, and that seems to have resolved the issue for now. Still would be good to fix this at some point.

@yourcelf
Copy link
Collaborator

Not sure if this is related -- but we had a crash today where the last logged event was a redis "end". Normally in development, redis seems to end / restart about every 5 minutes or so; presumably this happens in production too, but doesn't ordinarily cause a crash. https://www.exratione.com/2013/01/nodejs-connections-will-end-close-and-otherwise-blow-up/ discusses redis connections just randomly terminating, with possible badness depending on whether something is trying to write to redis at the same time.

@drewww
Copy link
Collaborator Author

drewww commented Mar 26, 2014

That dev restart is definitely the idle timeout issue - I've seen that a bunch of times before. But a while-loaded end is definitely new territory to me.

In theory it shouldn't be fatal; it would cause all the various logging mechanisms to fail but I wouldn't expect that to hard crash the server without some extra traces about saves failing. Hard to say, though - would definitely have an impact all over the codebase in bad ways.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Projects
None yet
Development

No branches or pull requests

2 participants