Timeout set by client is only respected by top-level grain
#4.328 geöffnet am 27. März 2018
Repository-Metriken
- Stars
- (10.777 Stars)
- PR-Merge-Metriken
- (Durchschn. Merge 2T 2h) (64 gemergte PRs in 30 T)
Beschreibung
According to https://github.com/dotnet/orleans/issues/1693#issuecomment-212984821 the timeout set by the client should be respected and propagated down to grains invoked by the grain running on the silo. I have a (fairly) minimal project here that demonstrates it is not: https://github.com/pmsanford/orleans-timeout-propagation-demo
It appears that the request to the first grain completes even if it takes longer than the SiloMessagingOptions.ResponseTimeout. However, any grain calls beyond the initial call only respect the SiloMessagingOptions.ResponseTimeout. Have I misinterpreted the comment on the linked issue or is this a bug?
Should a timeout set on MessagingOptions (or ClientMessagingOptions (??)) on the client side propagate through all grain activations on the silo?
Edited to add: This is on OSX, with dotnet --version 2.1.4 and Orleans version 2.0.0-rc2