Huge sidekik jobs pending for task: Jobs::UpdateGravatar

(Gautam Kachenga) #1

Our discourse set-up had been running fine for 2 months.
However since morning today we experienced that the forum has increased latency as well as gives timeout.

After checking the sidekik stats, we found that 0.5 million requests are queued in sidekik (dashboard gave a similar warning). Most requests are for Jobs::UpdateGravatar for every user.

Restarting the ec2 instance solves the problem for a while before the latency/time-out reappear.
Any idea how to mitigate this ?
Can we clear the sidekik queue - no problem if new post/topic data is lost because of this.