MySQL and ElasticSearch performance

[ghukill]

On an instance of Combine with 6 million + records - an 8gb VM - beginning to see some strain on mostly vanilla configurations for ElasticSearch and MySQL.

ElasticSearch

• heap size bumped to 1gb
• periodacally crashing
• at one point, lost a bunch of indices

MySQL

• some tweaks to innodb settings, but not having much effect
• job deletions are very, very slow, and slow down other operations

Initiating this issue to note progress on this front of hardening these configurations.

[ghukill]

Made some cursory headway on ES going down:

• reduced all indices to 1 shard, 0 replicas
• dropped heap size to 512m
  • this seems to be the primary cause of going down, running out space on VM

As for MySQL, scraping the bottom of the memory bucket on an 8gb VM, saw that a Job delete of a sizable job gobbles up about 300mb of memory. With only 600mb or so to spare, this is not insignificant.

More efficient / speedier MySQL deletions would still be ideal.

[ghukill]

Closing - switch to MongoDB for high IO tables has helped dramatically.