Wondering whether it is possible to easily and system-lite to add search system lag to the search results to a wiki/wikis.
I am trying to do a clean up of xwiki conflict of interest link spam, and a result is the same hours after a cleaning process across numbers of wikipedias (en, id, simple, sd, ur, id, pnb). Having a lag time would indicate to me when I should come back to check that I have done the clean up appropriately.
Thanks.
@EBernhardson would be the one to answer this. When we first built this, I was told there's around a 15-minute lag, but I think it might be shorter now.
Make sure you're clearing the cache, if necessary, which lasts for 10 minutes. If you get a cached result you'll see a "purge cache" link.
You can see the current lag here, it's currently sitting at about 6 hours: https://grafana.wikimedia.org/d/000000400/jobqueue-eventbus?orgId=1&from=1588244840010&to=1588266440010&var-site=eqiad&var-type=All
The problem right now is we are trying to reindex wikidata on a cluster that is orders of magnitude smaller than the production clusters. The machines are crying, some of them are crashing and we've had to restart a few times. It will clear up with time. The machine count in the cluster is being expanded by 50% to help with these problems, but unclear if that will be sufficient mitigation.