Lucas_Werkmeister_WMDE added a comment.

  > Can I take one step back and ask what are you using this max lag measure 
for, and how?
  
  At a high level, the purpose of checking the query service update lag is the 
same as for SQL lag: to make bots slow down if the database (broadly speaking) 
can’t keep up with the current edit rate. The query service update lag is 
integrated into the standard API `maxlag` mechanism (with a conversion factor – 
the query service is allowed to lag 60× as much as MySQL replication) so that 
bots don’t need any custom code for this – following the usual API etiquette is 
enough. (Unlike with SQL replication lag, though, we don’t forcibly make the 
wiki read-only if the lag grows too high.)
  
  One reason this is important is that bot edits are often based on results 
from the query service, so if the query service is too outdated, the bot edits 
can start being redundant or even wrong. For that reason, I think your 
suggestion of directly tying the metric to the number of queries the server has 
received makes a lot of sense. If a server isn’t handling user queries at the 
moment, we don’t care how outdated it is, as long as we think it will be able 
to catch up eventually – and as far as I’m aware, depooled servers that are 
processing updates normally don’t usually have issues keeping up with edits 
(but @dcausse would know more about that).

TASK DETAIL
  https://phabricator.wikimedia.org/T331405

EMAIL PREFERENCES
  https://phabricator.wikimedia.org/settings/panel/emailpreferences/

To: Lucas_Werkmeister_WMDE
Cc: Joe, Aklapper, dcausse, Lucas_Werkmeister_WMDE, Astuthiodit_1, AWesterinen, 
Arnoldokoth, karapayneWMDE, Invadibot, MPhamWMF, maantietaja, wkandek, 
JMeybohm, CBogen, ItamarWMDE, Akuckartz, Nandana, Namenlos314, jijiki, Lahi, 
Gq86, GoranSMilovanovic, QZanden, EBjune, merbst, LawExplorer, _jensen, 
rosalieper, Scott_WUaS, Jonas, Xmlizer, jkroll, Wikidata-bugs, Jdouglas, aude, 
Tobias1984, Manybubbles, Mbch331
_______________________________________________
Wikidata-bugs mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to