- Mentioned In
- T188368: Show Event page in loading state if a job is running or is queued
T192954: Add cronjob to regularly spawn any unstarted jobs that have been submitted to the queue
T192743: Display of event stats is in the wrong order
- Mentioned Here
- T192954: Add cronjob to regularly spawn any unstarted jobs that have been submitted to the queue
The only event I observed this issue was for https://tools.wmflabs.org/grantmetrics/programs/Women_in_Red/Women_in_Red_-_February_hackathon
I noticed when I hit "update stats" it reloaded instantly. For that many participants/wikis, it shouldn't go that fast. Sure enough, it seems the job wasn't actually firing (or something). The "Last updated" timestamp even said February 2018. So, I went into the console and manually updated the stats, and now it's fixed...
My guess is the job got stuck somehow. I think this can be solved but using a cronjob to run the app:spawn-jobs console command. We had talked about doing this when we first set up the job queue system, but I guess we never created the cronjob. It looks like Kubernetes cronjobs are a thing, but not officially supported on Toolforge and they are a bit complicated to set up: https://wikitech.wikimedia.org/wiki/Help:Toolforge/Kubernetes#Kubernetes_cronjobs. The normal jsub grid engine job I don't think will work because it will use an older version of PHP. I have created T192954
@Shouston_WMF I guess I should ask... are you aware of any other events where the retention stats looks off (e.g. is more than the number of "new editors")? The one we looked at WMCON I believe was https://tools.wmflabs.org/grantmetrics/programs/Women_in_Red/Women_in_Red_-_February_hackathon which has now been fixed.