Bug 4331 - Jobs disappear from state machine under heavy loads.
: Jobs disappear from state machine under heavy loads.
Status: RESOLVED FIXED
: GRAM
wsrf managed execution job service
: 4.0.1
: PC Linux
: P3 normal
: 4.0.2
Assigned To:
:
:
:
: 4197
  Show dependency treegraph
 
Reported: 2006-04-06 09:50 by
Modified: 2006-04-06 09:53 (History)


Attachments


Note

You need to log in before you can comment on or make changes to this bug.


Description From 2006-04-06 09:50:29
Jobs seem to be disappearing from the state machine before they can reach Done
or Failed when a large number of staging jobs are requested.
------- Comment #1 From 2006-04-06 09:53:10 -------
The remove() Destroy opertaion callback wasn't making sure that the external
state of the job was Done or Failed (just the internal state) before removing
the resource. Since the state machine sets the internal state before it
actually processes that state, the job could disappear before the terminal
states could be processed or a terminal state notification could be sent. A fix
for the race condition is in the globus_4_0_branch.