fix: previous delay 0 could still cause us to recruit a tlog before processing disk errors
This commit is contained in:
parent
64e0260085
commit
588eaf4b36
|
@ -2020,7 +2020,7 @@ ACTOR Future<Void> tLog( IKeyValueStore* persistentData, IDiskQueue* persistentQ
|
|||
}
|
||||
|
||||
//Disk errors need a chance to kill this actor.
|
||||
Void _ = wait(delay(0.0));
|
||||
Void _ = wait(delay(0.000001));
|
||||
|
||||
if(recovered.canBeSet()) recovered.send(Void());
|
||||
|
||||
|
|
Loading…
Reference in New Issue