Krinkle added a comment.

  In T223410#6233012 <https://phabricator.wikimedia.org/T223410#6233012>, 
@Catrope wrote:
  
  > What's the rationale behind doing the branch cut on Friday, then waiting 
over 72 hours to deploy it to even test wikis on Monday?
  
  From my perspectice, the same rationale for why +2 doesn't go to production 
automatically within the hour, either. We should imho stop cutting and 
deploying on the same day regardless of any other changes.
  
  Note that I'm not against more rapid deployments in general. I just think 
that right now we are far too disfynctional and undisciplined to do that 
responsibly.
  
  For changes merged Tue-Fri there are currently two working days for 
maintainers, product owners and QA to find out about what changes are made, 
whether they approve, and if so what amount of QA is appropiate to do locally 
or on Beta prior to deployment, and then to have performed that QA - all before 
the production deploy on Tuesday.  The same applies to cross-cutting risks as 
well, e.g. TechCom, Security, Performance finding logical mistakes or 
misunderstands that are incompatible with current database, CDN, or parser 
cache semantics. Areas that are frequently misunderstood and rarely found even 
on group0.
  
  A similarly magical and undocumented/informal QA cycle also repeats itself 
for the Wednesday group0-1 and Thursday group1-2. None of this is formalised or 
explicitly expected or waited for, and thus assumes nobody is on vacation, or 
pulled away to other work on any given day.
  
  I elaborated on this at T118212: Take heat off day before the weekly 
branch-cut? <https://phabricator.wikimedia.org/T118212>.
  
  In short, I think we need to "logically slow down" (better understand and 
consistently expect/ensure), after which we can then try speeding it up, or 
simply removing time-bound bottlenecks alltogether and see how long it takes in 
general for people to do their thing.
  
  A long-term idea I have is that perhaps Beta shouldn't run master but rather 
the next branch. E.g. we'd cut it on Thursday for Beta, and promote onwards 
from there until the following Thursday, and repeat.
  
  There are a lot of moving parts though, so I'm definitely in favour of 
incremental changes and regular re-evaluation.

TASK DETAIL
  https://phabricator.wikimedia.org/T223410

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

To: Krinkle
Cc: Catrope, Majavah, ArielGlenn, Krinkle, DannyS712, BEANS-X2, Amorymeltzer, 
RhinosF1, Ladsgroup, Michael, Jakob_WMDE, Pablo-WMDE, Addshore, 
Lydia_Pintscher, darthmon_wmde, WMDE-leszek, Lucas_Werkmeister_WMDE, brennen, 
mmodell, greg, hashar, Aklapper, Jdforrester-WMF, Blissjay007, Oblanco79, 
Alter-paule, Beast1978, CBogen, Un1tY, Hook696, Daryl-TTMG, RomaAmorRoma, 
E.S.A-Sheild, Kent7301, Meekrab2012, joker88john, CucyNoiD, Nandana, 
NebulousIris, Gaboe420, Versusxo, Majesticalreaper22, Giuliamocci, Adrian1985, 
Cpaulf30, Lahi, Gq86, Af420, Darkminds3113, Bsandipan, Lordiis, 
GoranSMilovanovic, Adik2382, Th3d3v1ls, Ramalepe, Liugev6, QZanden, 
LawExplorer, WSH1906, Lewizho99, Maathavan, Poyekhali, _jensen, rosalieper, 
Taiwania_Justo, Liudvikas, Scott_WUaS, Ixocactus, Wong128hk, Wikidata-bugs, 
aude, El_Grafo, Dinoguy1000, Steinsplitter, Mbch331, Keegan
_______________________________________________
Wikidata-bugs mailing list
Wikidata-bugs@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikidata-bugs

Reply via email to