| Gehel added a comment. |
A few notes:
The limitations we hit on WMCS are:
- disk space: we have a 2 special images with 300GB disk, but we are already hitting that limit. It is probably possible to get larger disks on labs, but we will at some point hit the limit.
- performance: the test instance we have running on WMCS is significantly slower than our production servers, to an extend that it makes testing painful. I'm not entirely clear on what measure of "slower" we have. The most probable cause is lower IO throughput. (@Smalyshev could you expand on that?) This is going to be hard to address while staying on WMCS.
Deploying both elasticsearch and wdqs on the relforge servers is ugly and will make maintenance more complex. It might be a short term solution, but not one I'm comfortable pursuing in the longer term. If we go that way, we should get dedicated servers for a wdqs testing cluster (probably 2 servers, with similar specs as the current wdqs production servers).
Relforge servers (as any other physical server we could get) are in the production network, and as such constrained by the same security limitations as other production hosts (security upgrade, more strict access control, ...). Having test servers on which we experiment in that production network is not necessarily the best idea. Yes, we have an exception for relforge, and this is a very similar use case. This still need to be discussed / approved by our core SRE team.
Cc: Aklapper, dcausse, EBernhardson, Smalyshev, Gehel, Lahi, Gq86, Darkminds3113, Lucas_Werkmeister_WMDE, GoranSMilovanovic, QZanden, EBjune, merbst, LawExplorer, Avner, Jonas, FloNight, Xmlizer, jkroll, Wikidata-bugs, Jdouglas, aude, Tobias1984, Manybubbles, Mbch331
_______________________________________________ Wikidata-bugs mailing list [email protected] https://lists.wikimedia.org/mailman/listinfo/wikidata-bugs
