Deleting or restoring a post in a large topic (the one I saw this happen with had about 56K posts) caused performance problems for nearly 10 minutes.
I was getting 504 timeouts during that period.
A discussion on the performance of deleting a post in a large topic revealed a performance issue. anon21958287 reported that deleting or restoring a post in a 56K post topic caused nearly 10 minutes of performance problems, resulting in 504 timeouts. accalia confirmed the issue, citing instance monitor data that showed response times exceeding 15 seconds during the outage. Two outages were reported, one at 0942(GMT-0400) and another at 1032(GMT-0400), both coinciding with the deletion activity. The issue has since been fixed with a commit as noted by sam.
Deleting or restoring a post in a large topic (the one I saw this happen with had about 56K posts) caused performance problems for nearly 10 minutes.
I was getting 504 timeouts during that period.
I’ll confirm that time frame.
our instance monitor captured an outage starting at 1032(GMT-0400) lasting until 1039(GMT-0400) today where response times for our discourse instance were > 15 seconds. I don’t have data on whether those were simply exceptionally long requests or whether they would generate a 504 error as the monitor times out at 15 seconds and records the site as offline.
An earlier outage occurred at roughly 0942(GMT-0400) and lasted until about 0954(GMT-0400) today according to our monitoring site.
Monitoring data is available for the affected day here (CSV formatted)
Yes, that period coincided with the same activity. The 1032EDT was a test to confirm that my action was the cause.
i always forget that. for maximum clarity edited with offsets to GMT