-
The startup scripts of the workflow system do not start services when there is a PID file. As in uas-core there should be an additional check whether there is a process with a pid from the file.
2009-11-02 09:56:17 UTC in UNICORE
-
golbi committed revision 5625 to the UNICORE SVN repository, changing 1 files.
2009-10-28 16:29:55 UTC in UNICORE
-
golbi committed revision 5620 to the UNICORE SVN repository, changing 4 files.
2009-10-28 13:43:46 UTC in UNICORE
-
golbi committed revision 5619 to the UNICORE SVN repository, changing 1 files.
2009-10-28 13:40:45 UTC in UNICORE
-
golbi committed revision 5618 to the UNICORE SVN repository, changing 1 files.
2009-10-28 13:37:07 UTC in UNICORE
-
golbi committed revision 5605 to the UNICORE SVN repository, changing 8 files.
2009-10-26 17:12:18 UTC in UNICORE
-
You are probably right in case of the brokering problems and the feature you described would be cool. However this is a completely another story. I'm pretty sure (after having few rounds of UNICORE hands on with the actual end users) that we badly need better *error reporting*. Please notice that currently when workflow fails user is presented with the red cross and nothing more. Yes - I can dig...
2009-10-26 10:45:58 UTC in UNICORE
-
After some more tests it seems that this one and 2885682 share the same reason.
2009-10-25 00:17:54 UTC in UNICORE
-
Whenever I try to submit a job which is killed by a BSS (Torque in my case, job is correctly killed as declared resource limits are exceeded) I got the following error (with a popup too) in the URC:
com.intel.gpe.clients.api.exceptions.GPEMiddlewareRemoteException: Unable to retrieve resource properties
at com.intel.gpe.clients.impl.WSRPClientImpl.getResourceProperty(WSRPClientImpl.java:95)
2009-10-25 00:13:54 UTC in UNICORE
-
Sometimes I get an error in the URC when it tries to update job status (the reason of this error is subject of an another bug report). After such an error the client correctly updates the job's icon in the grid browser. But the read-only snapshot of the job in the editor still has 'running' state in the tab name.
Refreshing the job also doesn't help.
2009-10-25 00:08:09 UTC in UNICORE