2 tests from Apertium MT were failing in Production/master CI (See: https://integration.wikimedia.org/ci/job/npm-node-6-docker/3282/console for full log) due to dependencies on apertium-apy service running on Labs. This is not OK and it is hard to debug and can block merge changes of cxserver in Production and updating it.
Description
Details
Subject | Repo | Branch | Lines +/- | |
---|---|---|---|---|
Use TestClient for tests, instead of a real Apertium instance | mediawiki/services/cxserver | master | +9 -4 |
Related Objects
- Mentioned In
- T206777: Create Wikipedia Shan
T203160: CX2: Highlight (and skip) references with a template that could not be adapted
T203077: Performance analysis for translate API - Mentioned Here
- rGCXSb7cdb2646168: Add shn
T203077: Performance analysis for translate API
T203160: CX2: Highlight (and skip) references with a template that could not be adapted
T206777: Create Wikipedia Shan
Event Timeline
ssh language-apertium2.eqiad.wmflabs service apertium-apy status ● apertium-apy.service - Translation server and API for Apertium Loaded: loaded (/lib/systemd/system/apertium-apy.service; disabled; vendor preset: enabled) Active: inactive (dead) Docs: http://wiki.apertium.org/wiki/Apertium-apy
@KartikMistry Are there docs for this service? It took a while for me to figure out the service name etc.
I restarted it, but why wasn't it running by default?
Also, it's not nice that our tests depend on a labs service. They should be network-less by default.
https://www.mediawiki.org/wiki/Content_translation/Machine_Translation/Apertium/Service
Also, it's not nice that our tests depend on a labs service. They should be network-less by default.
Agree. I'll rename this task and reuse it.
Change 472103 had a related patch set uploaded (by Santhosh; owner: Santhosh):
[mediawiki/services/cxserver@master] Use TestClient for tests, instead of a real Apertium instanace
Change 472103 merged by jenkins-bot:
[mediawiki/services/cxserver@master] Use TestClient for tests, instead of a real Apertium instance
Once this is deployed to production, the (Apertium) requests in https://grafana.wikimedia.org/dashboard/db/service-cxserver?orgId=1&from=now%2Fw&to=now should go down dramatically.
Mentioned in SAL (#wikimedia-operations) [2018-11-09T12:16:23Z] <kartik@deploy1001> Started deploy [cxserver/deploy@fc21164]: Update cxserver to 01686f6 (T208831)
Mentioned in SAL (#wikimedia-operations) [2018-11-09T12:17:32Z] <kartik@deploy1001> Finished deploy [cxserver/deploy@fc21164]: Update cxserver to 01686f6 (T208831) (duration: 01m 09s)