Oh, that's right! I do remember the discussion about changing to CNAMEs.
I wonder if Quarry also keeps audit logs of the queries it ran and how long they took. It would be useful to display them in form of a trend line, for instance.
On Sat, Dec 30, 2017 at 12:28 PM, Bryan Davis bd808@wikimedia.org wrote:
On Sat, Dec 30, 2017 at 9:51 AM, Huji Lee huji.huji@gmail.com wrote:
I feel like Quarry is slower than before (before being last week or last month). Queries almost always get queued, and once executed, simple
queries
take longer to result.
I have no idea how to investigate this though. Any thoughts?
Quarry uses the *.analytics.db.svc.eqiad.labsdb servers. Before 2017-12-13 these servers were only used by tools which had explicitly changed their configuration to point to the new servers. After that date the legacy *.labsdb service names became CNAMEs (DNS aliases) to the analytics.db service names. This greatly increased the load on the backing servers which would logically make them slower to return results.
Bryan
Bryan Davis Wikimedia Foundation bd808@wikimedia.org [[m:User:BDavis_(WMF)]] Manager, Cloud Services Boise, ID USA irc: bd808 v:415.839.6885 x6855
Wikimedia Cloud Services mailing list Cloud@lists.wikimedia.org (formerly labs-l@lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/cloud