LDAP sync sends cache clear messages in a tight loop

Description

The errors in the logs correlate to the CPU spikes, which happens during the LDAP syncs

Issue created from a message in Mattermost.

QA Test Steps

  • Create an HA setup with 5000 teams, 50000 channels. (use the mattermost sampledata command)

  • Run LDAP sync without this patch and notice that during the sync there is a CPU spike and an insane amount of cluster traffic

  • Apply this patch and verify that there is no such spike and the cluster traffic is flattened.

Activity

Show:
Rohitesh Gupta
August 6, 2020, 4:50 PM

Tested on the latest 5.26 test instance and there were no issues found related to LDAP sync. Closing the ticket.

Katie Kusjanovic
July 23, 2020, 7:14 PM
Amy Blais
July 22, 2020, 5:17 PM

Latest update: after testing and investigating further, dev has a better idea what the issue is; aiming to have a fix available for QA testing by tomorrow latest.

Katie Kusjanovic
July 22, 2020, 2:25 PM

looks like we had a brief moment of sustained CPU load last night. is there anything i need/can to do aside from providing this very important diagram?

Stu Doherty
July 20, 2020, 1:37 PM

Great, thanks and ! Sven let the customer know as well.

Done

Mana

None

Assignee

Agniva De Sarker

QA Assignee

Rohitesh Gupta

Reporter

Agniva De Sarker

Epic Link

None

Fix versions

Mattermost Team

Sustained Engineering

Sprint

None

Labels

QA Testing Areas

Other (write in QA test steps)

GitHub Issue

None

Components

None

Severity

S1_data_loss_or_crash