Horrible cpu usage on shard1



  • Can anyone shed some light on what's going on with the servers lately?

    This is my CPU usage graph for shard1 for the past 48 hours: shard1 cpu usage

    Notice how the graph drastically changes from being nice and smooth in the first part to being all over the damn place, super jittery, and averaging significantly more than before. I have not changed anything in my code or room layouts since. It all started after yesterday's server restarts when shard2 died (again). I am seeing similar problem on shard3, and many of my friends on shard1 and shard2 are reporting similar CPU troubles as well.

    This is much worse than we had before server migration...



  • I am seeing similar results as well. Shard 1-3 is displaying similar graphs despite drastically different CPU allocations. It looks like a 10-20% average increase in CPU usage with large uneven spikes in multiple places. Hopefully something can be found and resolved to fix this.



  • I've seen odd CPU usage on shard1 as well. Usage will sometimes spike for no reason I can work out, and when I look closely, methods that shouldn't take much time at all take an extremely long time. (My profiling once yesterday showed my rooms taking something like ~50 cpu, in one tick, to perform an operation that it typically does in under 1 CPU, and I wasn't able to work out any possible reason it could have taken so long.)

    I'm not certain the problem lies anywhere other than my code. But if we're taking inventory of people who are seeing recent flakiness in CPU usage that they haven't been able to account for, count me among them.



  • My CPU usage graph on shard3:

    0_1574286954181_a51440d6-7ba8-44f9-b7e5-92f62d020cb0-image.png



  • A couple of hours ago there was another global reset, and it would appear my CPU usage is more or less back to where it was two days ago. It is much smoother now, at the very least.



  • It looks like mine is doing much better as well. I am curious as to what caused these issues across all shards.



  • I don't have more cpu statistics than a floating average and the impact on my bucket (still onshard3 only). But I realised the cpu usage spike during the last days, to the point my bucket was drained and cpu saving kicked in. My bucket recovered today as well, but it broke down during the last hour(s) again. Does anyone of you see the impact on their graphs again?