Part of operations Q3 2017-2018 goal is to serve 50% of Mathoid service via kubernetes. This is the very first service we are going to be serving via kubernetes so it will serve as a template for all other templates. This task is about tracking that work.
All Mathoid requests are essentially "internal", as in created by our infrastructure which does remove some variability from the work making it easier.
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | akosiaris | T184462 Serve one production service via Kubernetes | |||
| Resolved | akosiaris | T184919 Serve at least 50% of Mathoid via kubernetes |
Change 410489 had a related patch set uploaded (by Alexandros Kosiaris; owner: Alexandros Kosiaris):
[operations/puppet@production] Prepare kubernetes nodes for serving mathoid traffic
Change 410964 had a related patch set uploaded (by Alexandros Kosiaris; owner: Alexandros Kosiaris):
[operations/deployment-charts@master] Add the helm chart for mathoid
Change 410964 merged by Alexandros Kosiaris:
[operations/deployment-charts@master] Add the helm chart for mathoid
Yes. We don't have a proper helm repo yet (TBD) but the chart itself is here https://gerrit.wikimedia.org/g/operations/deployment-charts/+/31562f04cb8836693afc25453ee33ebc06c775cf
Ok. I was just asking because I will be giving a talk on Math rendering in Wikipedia on Mar. 8th and it would be nice to have some usage statistics;-)
Some years ago there was https://grafana.wikimedia.org/dashboard/db/service-mathoid?orgId=1 but it looks quite empty at this moment.
Ah I think we had a small misunderstanding. By chart I meant the overloaded/abused by helm[1] term of https://en.wikipedia.org/wiki/Nautical_chart, whereas I guess you meant https://en.wikipedia.org/wiki/Chart.
You do raise a valid point. I 've updated that graph a bit to show the request rates. I am not sure how to get the latencies though. @mobrovac, any ideas ?
[1] https://helm.sh/
The latencies as witnessed by RESTBase clients are available on this panel, which include the Math end points as well. We actually do collect all the data for Mathoid itself and some of it is available on its dashboard, but it needs some love.
Change 415605 had a related patch set uploaded (by Mobrovac; owner: Mobrovac):
[operations/deployment-charts@master] Mathoid chart: Use port 10042
Change 416950 had a related patch set uploaded (by Alexandros Kosiaris; owner: Alexandros Kosiaris):
[operations/puppet@production] Populate kubeconfigs on deployment server
Change 416950 merged by Alexandros Kosiaris:
[operations/puppet@production] Populate kubeconfigs on deployment server
Change 410489 merged by Alexandros Kosiaris:
[operations/puppet@production] Prepare kubernetes nodes for serving mathoid traffic
Mentioned in SAL (#wikimedia-operations) [2018-03-19T14:42:55Z] <akosiaris> T184919 pool all kubernetes for service mathoid.
Mentioned in SAL (#wikimedia-operations) [2018-03-26T11:16:57Z] <akosiaris> depool scb hosts for mathoid service. T184919
This has been achieved successfully and even surpassed the goal by achieving 100%. I 'll happily resolve this