Has the issue resolved after you bumped up to 8gb? I’m considering doing the same as you since we’re still get these sidekiq errors on a 4gb machine.
Fortunately, after increasing to 8GB, the problem has not recurred.
2 Likes
Thanks! I’ll try that approach then.
1 Like
It suddenly started appearing to me a couple of days ago.
Before I never saw Sidekiq is using too much memory errors in the logs. Now I see them a few times a day.
And today it happened exactly as @pelcami has mentioned: the VM just plainly became unresponsive. SSH didn’t work. No access on HTTP or HTTPS.
The VM is on Azure, so I pulled up the charts, and it showed network activity suddenly dropping down completely to zero (like a cliff) and CPU stuck at around 40% (probably completely using up one core). Restarting the VM solved it.
What I want to say is, it happened so suddenly that it was almost like one minute it was business as usual, normal network traffic, then the next minute it fell off a cliff and dropped to zero. Not gradual at all.
Something is happening to trigger an infinite loop in Sidekiq or something…
Be sure to do a full rebuild we recently bumped ruby version in image
3 Likes
ok. I’ll rebuild and observe.
Just as a reminder to people: If you get a “Docker version too low” error during rebuild, read the following:
4 Likes
I’ve started seeing this after an update a few days ago @sam after the Amazon lifecycle patch I did a complete rebuild and now I’m seeing this everyday. Never happened before that.
I’m on v1.9.0.beta14 +200
When is the last time you did a
git pull
./launcher rebuild app
You could be on an older version of Ruby.
2 Likes
I did this for the build above beta14+200
I didn’t do a git pull, thought that rebuild does that.
EDIT: Okay I’m it again now, lets see how it goes.
That’s good to know. Also, I just noticed that vagrant is still on Rails 4.
No issues so far after a git pull + rebuild instead of just a rebuild. Somehow I thought a rebuild would also do a pull
this is the dashboard from what ? pls teach me .
Go to yourforumurl/sidekiq to see what sidekiq is doing
similarly
/logs to see the error log
/admin/upgrade#/processes to see processes
3 Likes
Just thought I’d share that updating to v2.0.0 beta1 a couple of weeks ago this problem started again.
Sidekiq is consuming too much memory (using: 513.13M)
I checked the dashboard and it’s currently using about 6M but somehow it ramps up to 500 about once a day I see this warning in the logs. Oddly it’s always around 4:20am to 4:50am in the mornings.
Also, if your server has enough total RAM, you can increase the memory allocation for Sidekiq.
Place this in the env: section of app.yml:
UNICORN_SIDEKIQ_MAX_RSS: 1000
The default maximum is, I believe, 500mb.
4 Likes
Closing the loop here, over the past 2 years we had one large change that reduced sidekiq memory. We made a change to the hosted v8 instance to ensure it cleans up used memory and runs a gc. This resulted in a 10-20% memory reduction in our hosting.
Except for that, yes, on mega busy multisites you need to adjust UNICORN_SIDEKIQ_MAX_RSS
2 Likes