Page MenuHomePhabricator

Deploy Chart service in production
Open, HighPublic

Description

Now that T371278 is done, and seems to be working well, deploy the chart-renderer service in production.

Steps below follow https://wikitech.wikimedia.org/wiki/Kubernetes/Add_a_new_service

Steps are described in more detail either in a linked task, or, at the above link.

  • Begin New Service Request process T376939
  • Create k8s namespace and deployment users (@CDanis to do)
  • Write Helm chart for chart-renderer (Charts to do) T376948
  • Deploy the service to staging and production k8s clusters (Charts to do w/ help)
  • Deploy extension (T378127) (Charts to do w/ help)

Post test.wikipedia.org deploy:

  • make sure we have appropriate logging in place (Chris task)
  • Set up Grafana dashboard for monitoring, starting with making a copy of the template (Charts to do)

Doing it after testwiki is totally fine IMO.

We can get latency and also request success rates from different queries based on the one existing Prom metric.

  • Add alerts around latency ("under 5 seconds" perhaps), and one around 5xx result ratio
  • Configure Ingress (@CDanis to do no taskforce assistance needed)
  • Add chart-renderer to service mesh (@CDanis to do)

Event Timeline

Change #1079350 had a related patch set uploaded (by CDanis; author: CDanis):

[operations/deployment-charts@master] Add chart-renderer namespace

https://gerrit.wikimedia.org/r/1079350

Change #1079345 had a related patch set uploaded (by CDanis; author: CDanis):

[operations/puppet@production] Add chart-renderer deployment server profile

https://gerrit.wikimedia.org/r/1079345

CDanis renamed this task from Deploy Chart service in production (placeholder, not actionable yet) to Deploy Chart service in production.Thu, Oct 10, 7:53 PM
CDanis raised the priority of this task from Low to Medium.
CDanis updated the task description. (Show Details)

Change #1079345 merged by CDanis:

[operations/puppet@production] Add chart-renderer deployment server profile

https://gerrit.wikimedia.org/r/1079345

Change #1079350 merged by jenkins-bot:

[operations/deployment-charts@master] Add chart-renderer namespace

https://gerrit.wikimedia.org/r/1079350

  • Configure LVS service (@CDanis to do)

What are the expectations regarding traffic here? Could this use Ingress instead of LVS?

What are the expectations regarding traffic here? Could this use Ingress instead of LVS?

Discussed with Janis on serviceops IRC and our best guess is that Ingress is totally viable for this service. This removes/simplifies a couple steps which is nice.

CDanis updated the task description. (Show Details)
CCiufo-WMF raised the priority of this task from Medium to High.Mon, Oct 21, 5:43 PM
CCiufo-WMF moved this task from Needs Triage to Sprint 9 on the Charts board.
CCiufo-WMF edited projects, added Charts (Sprint 9); removed Charts.

Change #1085607 had a related patch set uploaded (by CDanis; author: CDanis):

[operations/puppet@production] deployment group: add aude

https://gerrit.wikimedia.org/r/1085607

Change #1085607 merged by CDanis:

[operations/puppet@production] deployment group: add aude

https://gerrit.wikimedia.org/r/1085607