You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
#' to finish sending large data (in case an exit signal is received).
Suggest: (when a worker exits after reaching a timeout or having completed a certain number of tasks)
As (i) there is no exit signal and (ii) if for example daemons(0) closes the socket on the client side, the exit is immediate with no linger.
The current default for seconds_exit of 0.1 is quite low and could be problematic for sending/receiving large objects. The default in mirai has been 1000 ms = 1L for the past couple of builds. I'd assume this might only be an issue re. autoscaling, but I'd take the performance hit for the extra safety.
The text was updated successfully, but these errors were encountered:
Thanks for explaining, and especially for catching the low default for seconds_exit. Should all be fixed now. With the new seconds_exit, I am going to retest for hanging just in case it helped that particular issue.
Prework
Description
On the docs, a couple of things I came across re.
mirai
:crew/R/crew_launcher.R
Line 43 in cb976e0
The argument name is 'timerstart'.
crew/R/crew_launcher.R
Line 33 in cb976e0
Suggest: (when a worker exits after reaching a timeout or having completed a certain number of tasks)
As (i) there is no exit signal and (ii) if for example
daemons(0)
closes the socket on the client side, the exit is immediate with no linger.seconds_exit
of 0.1 is quite low and could be problematic for sending/receiving large objects. The default inmirai
has been 1000 ms = 1L for the past couple of builds. I'd assume this might only be an issue re. autoscaling, but I'd take the performance hit for the extra safety.The text was updated successfully, but these errors were encountered: