merge 6.8: Fix pool preloading [CUSTOM] #43
Conversation
Fix PCLAS-300
|
@mpaulmier Le préloading c'est un truc qui n'est que chez nous il me semble |
| def pre_load_database(name): | ||
| def load(): | ||
| with Transaction().start(name, 0): | ||
| Pool(name).init() | ||
| return load |
There was a problem hiding this comment.
Je ne comprends pas le lien entre ce changement et le fait que la pool ne soit pas chargée deux fois.
Car a priori ce qui fait le travail c'est le test sur WERKZEUG_RUN_MAIN juste un poil au dessus.
There was a problem hiding this comment.
Le fait de ne pas créer de transaction fait que le thread de cache n'est pas spawn au bon moment. Du coup, comme pour l'autre problème, il commence par reset la pool au démarrage
| # The problem is that the cache listener will be started as soon as the | ||
| # transaction is created, and its first action will be to stop the pool | ||
| # / clear the caches. |
There was a problem hiding this comment.
I thought that this concurrency had been fixed by f258162 could it be the Transaction.start from pre_load_database from this patch that added back another way to have the issue?
…TOM] PR: coopengo#43 Fix #7859
nicoe
left a comment
nicoe
left a comment
There was a problem hiding this comment.
Au final la solution sera probablement plutôt https://foss.heptapod.net/tryton/tryton/-/issues/11401#note_331051
Mais go go
Real fix will probably be https://foss.heptapod.net/tryton/tryton/-/issues/11401#note_331051 Fix PCLAS-300
…TOM] PR: coopengo#43 Fix #7859
Real fix will probably be https://foss.heptapod.net/tryton/tryton/-/issues/11401#note_331051 Fix PCLAS-300
…TOM] PR: coopengo#43 Fix #7859
…TOM] PR: coopengo#43 Fix #7859
…TOM] PR: coopengo#43 Fix #7859
4 participants
Fix PCLAS-300