Back to the main page.

Bug 139 - smartcpu is triggered simultaneously on multiple slaves that share a linux node

Reported 2010-09-02 10:19:00 +0200
Modified 2011-01-05 12:01:00 +0100
Product: FieldTrip
Component: peer
Version: unspecified
Hardware: PC
Operating System: Mac OS
Importance: P1 minor
Assigned to: Robert Oostenveld
Depends on:
See also:

Robert Oostenveld - 2010-09-02 10:19:28 +0200

if I do n = 100; peercellfun(@pause, repmat({3}, 1, n) ) in one matlab, and while(1); peerlist; pause(1); end in another, I observe multiple peerslaves switching from idle to zombie. This suggests that smartcpu is triggered simultaneously. Previously I had also observed a race condition between slaves, which caused them to alternate between idle and zombie.

Robert Oostenveld - 2010-09-07 17:18:36 +0200

this has been resolved by accumulating evidence over multiple calls prior to switching mode. Furthermore, there is now a freeze flag in smartcpu and smartmem, which is used by tcpsocket to prevent the switching of mode and/or memory available while a new job is arriving.

Robert Oostenveld - 2011-01-05 11:56:55 +0100

selected a long list of resolved bugs from roboos and changed the status into "RESOLVED"

Robert Oostenveld - 2011-01-05 12:01:00 +0100

selected all old bugs from roboos with status RESOLVED and changed it into CLOSED