You are browsing a read-only backup copy of Wikitech. The primary site can be found at wikitech.wikimedia.org

Incident documentation/2021-11-25 eventgate-main outage: Difference between revisions

From Wikitech-static
Jump to navigation Jump to search
imported>Jelto
No edit summary
imported>Krinkle
mNo edit summary
Line 6: Line 6:
-->
-->
==Summary==
==Summary==
During the helm3 migration of eqiad Kubernetes cluster the service eventgate-main had reduced availability. The service was not available between 7:32 and 7:35 UTC.  
During the [[Helm|helm3]] migration of eqiad Kubernetes cluster the service eventgate-main experience an outage. The service was not available between 7:32 and 7:35 UTC.  


For the helm3 migration the service had to be removed and re-deployed to the cluster. Most Kubernetes services were explicitly pooled in codfw-only during the re-deployments. eventgate-main was also falsely assumed to be served by codfw but was still pooled in eqiad. So during the time of removing and re-creating the pods, no traffic could be served for this service.
For the helm3 migration the service had to be removed and re-deployed to the cluster. Most Kubernetes services were explicitly pooled in codfw-only during the re-deployments. eventgate-main was also falsely assumed to be served by [[Codfw cluster|Codfw]] but was still pooled in [[Eqiad cluster|Eqiad]]. So during the time of removing and re-creating the pods, no traffic could be served for this service.


The commands used to migrate and re-deploy codfw (see [[phab:T251305#7492328|T251305#7492328]]) were adapted and re-used for eqiad (see [[phab:T251305#7526591|T251305#7526591]]). Due to a small difference in what Kubernetes services are pooled as active-active and what are active-passive, eventgate-main was missing in the depooling command (as is it not pooled in codfw currently).
The commands used to migrate and re-deploy codfw (see [[phab:T251305#7492328|T251305#7492328]]) were adapted and re-used for eqiad (see [[phab:T251305#7526591|T251305#7526591]]). Due to a small difference in what Kubernetes services are pooled as active-active and what are active-passive, eventgate-main was missing in the depooling command (as is it not pooled in codfw currently).


 
'''Impact''': For about 3 minutes (from 7:32 to 7:35 UTC), eventgate-main was unavailable. This resulted in 25,000 unrecoverable MediaWiki backend errors due to inability to queue new jobs. About 1,000 user-facing web requests and API requests failed with an HTTP 500 error. Event intake processing rate measured by eventgate briefly dropped from ~3000/second to 0/second during the outage.
'''Impact''': Between 7:32 and 7:35 UTC approximately 25.000 MediaWiki exceptions were created (see MediaWiki dashboard below). For around 5 minutes increased user-facing 5XX rates could be seen (see Varnish dashboard below), with a maximum of around ~230 5XX errors per minute. The produced events in eventgate were drastically reduced during the outage (see eventgate dashboard below).


'''Documentation''':
'''Documentation''':
*[https://grafana.wikimedia.org/d/000000438/mediawiki-exceptions-alerts?orgId=1&var-datasource=eqiad%20prometheus%2Fops&viewPanel=18&from=1637825199597&to=1637826150301 Grafana Dashboard MediaWiki Exceptions]
*[https://grafana.wikimedia.org/d/000000438/mediawiki-exceptions-alerts?orgId=1&var-datasource=eqiad%20prometheus%2Fops&viewPanel=18&from=1637825199597&to=1637826150301 Grafana Dashboard MediaWiki Exceptions]
*[https://grafana.wikimedia.org/d/ZB39Izmnz/eventgate?orgId=1&from=1637821829000&to=1637829749000 Grafana eventgate statistics]
*[https://grafana.wikimedia.org/d/ZB39Izmnz/eventgate?orgId=1&from=1637824200000&to=1637827200000 Grafana eventgate statistics]
*[https://grafana.wikimedia.org/d/000000503/varnish-http-errors?viewPanel=7&orgId=1&from=1637825257550&to=1637826384984 Grafana Varnish http errors]
*[https://grafana.wikimedia.org/d/000000503/varnish-http-errors?viewPanel=7&orgId=1&from=1637825257550&to=1637826384984 Grafana Varnish http errors]
<gallery mode="packed">
File:2021-11-25-mediawiki-exceptions.png
File:2021-11-25-eventgate-statistics.png
File:2021-11-25-varnish-http500.png
</gallery>


==Actionables==
==Actionables==


* automate maintenance and proper de-depooling of Kubernetes services using a cookbook [[phab:T277677|T277677]] and [[phab:T260663|T260663]]
*automate maintenance and proper de-depooling of Kubernetes services using a cookbook [[phab:T277677|T277677]] and [[phab:T260663|T260663]]
* reduce snowflake services which need special treatment and make most/all of them active-active (for example [[phab:T288685|T288685]])
*reduce snowflake services which need special treatment and make most/all of them active-active (for example [[phab:T288685|T288685]])
* optional: create a lvs/pybal/k8s service dashboard to see which service is pooled in which DC (will create a task)
*optional: create a lvs/pybal/k8s service dashboard to see which service is pooled in which DC (will create a task)
*[[phab:T296699|T296699: Pool eventgate-main in both datacenters (active/active)]]
*[[phab:T296699|T296699: Pool eventgate-main in both datacenters (active/active)]]

Revision as of 23:43, 1 December 2021

document status: in-review

Summary

During the helm3 migration of eqiad Kubernetes cluster the service eventgate-main experience an outage. The service was not available between 7:32 and 7:35 UTC.

For the helm3 migration the service had to be removed and re-deployed to the cluster. Most Kubernetes services were explicitly pooled in codfw-only during the re-deployments. eventgate-main was also falsely assumed to be served by Codfw but was still pooled in Eqiad. So during the time of removing and re-creating the pods, no traffic could be served for this service.

The commands used to migrate and re-deploy codfw (see T251305#7492328) were adapted and re-used for eqiad (see T251305#7526591). Due to a small difference in what Kubernetes services are pooled as active-active and what are active-passive, eventgate-main was missing in the depooling command (as is it not pooled in codfw currently).

Impact: For about 3 minutes (from 7:32 to 7:35 UTC), eventgate-main was unavailable. This resulted in 25,000 unrecoverable MediaWiki backend errors due to inability to queue new jobs. About 1,000 user-facing web requests and API requests failed with an HTTP 500 error. Event intake processing rate measured by eventgate briefly dropped from ~3000/second to 0/second during the outage.

Documentation:

Actionables