Michael Kublin has posted comments on this change. Change subject: engine: Fixing RecoveryStoragePool and EventQueue ......................................................................
Patch Set 1: "I guess that we can find cases, of course each one of them has 1% probability, but this is a rare flow, error flow - so, do we really mind having here canDoAction() running again and DB load to prevent errors? In my opinion - not. yes, it's running twice (basically we also run it twice when running multiple actions), but in my opinion we have much more safety this way and we don't have to think on each change how it will affect it or the flow - we have a queue, an event and when running we check if we can run. " totally disagree, or prevent all races or not try to prevent some rare race,especially even if it will happen, it will not be a cause to any problem in the system "example - just random case , one recovery fails because of network error, other recovery in the queue tries to reconstruct using the same version and failes as well because the master in vdsm is already in the same version or even can cause to worse behaviour." - You are wrong, reconstruct usually loads all data. And recovery which we are talking about can be triggered only by user. -- To view, visit http://gerrit.ovirt.org/11201 To unsubscribe, visit http://gerrit.ovirt.org/settings Gerrit-MessageType: comment Gerrit-Change-Id: I3b58f095c69c76116f0af44b08ccaf41d1ff1fb8 Gerrit-PatchSet: 1 Gerrit-Project: ovirt-engine Gerrit-Branch: master Gerrit-Owner: Michael Kublin <[email protected]> Gerrit-Reviewer: Allon Mureinik <[email protected]> Gerrit-Reviewer: Ayal Baron <[email protected]> Gerrit-Reviewer: Barak Azulay <[email protected]> Gerrit-Reviewer: Liron Aravot <[email protected]> Gerrit-Reviewer: Michael Kublin <[email protected]> Gerrit-Reviewer: Yair Zaslavsky <[email protected]> _______________________________________________ Engine-patches mailing list [email protected] http://lists.ovirt.org/mailman/listinfo/engine-patches
