1 Nov
2019
1 Nov
'19
2:32 p.m.
On 11/1/2019 8:30 AM, Matt Riedemann wrote:
Looks like around the time of slowness mysqld is the top consuming process which is probably not surprising.
Oh I also see this in the mysql error log [1] around the time of the slow period, right after things pick back up: 2019-10-31T16:55:08.603773Z 0 [Note] InnoDB: page_cleaner: 1000ms intended loop took 161281ms. The settings might not be optimal. (flushed=201 and evicted=0, during the time.) So obviously something is going very wrong with mysql (or the node in general) during that time. [1] https://13cf3dd11b8f009809dc-97cb3b32849366f5bed744685e46b266.ssl.cf5.rackcd... -- Thanks, Matt