[Swift-user] coasters on eureka - block task ends prematurely

[Michael Wilde]

Im getting the following on eureka for a 1-job cat sanity test of coasters:

eur$ swift -tc.file tc -sites.file sites.xml cats.swift
Swift svn swift-r3186 cog-r2577

RunID: 20091117-1031-71txxj43
Progress:
Worker task failed: 1117-311023-000000Block task ended prematurely



Progress:  Active:1
Failed to transfer wrapper log from cats-20091117-1031-71txxj43/info/q 
on coast
Progress:  Submitted:1
Worker task failed: 1117-311023-000001Block task ended prematurely



Progress:  Active:1
Failed to transfer wrapper log from cats-20091117-1031-71txxj43/info/s 
on coast
Progress:  Submitted:1
Worker task failed: 1117-311023-000002Block task ended prematurely



Progress:  Active:1
Failed to transfer wrapper log from cats-20091117-1031-71txxj43/info/u 
on coast
Execution failed:
	Exception in cat:
Arguments: []
Host: coast
Directory: cats-20091117-1031-71txxj43/jobs/u/cat-u6s7zkjj
stderr.txt:

stdout.txt:

----

Caused by:
	Task failed: 1117-311023-000002Block task ended prematurely



Cleaning up...
Shutting down service at https://10.40.9.151:58810
Got channel MetaChannel: 845296226 -> null
+ Done
eur$

--

tc is:

coast	cat	/bin/cat	null	null	null

sites.xml is:

<config>
   <pool handle="coast">
     <filesystem provider="local" />
     <execution provider="coaster" jobmanager="local:cobalt"/>
     <profile namespace="globus" key="slots">1</profile>
     <profile namespace="globus" key="nodeGranularity">1</profile>
     <profile namespace="globus" key="workersPerNode">8</profile>
     <profile namespace="globus" key="maxNodes">1</profile>
     <profile namespace="globus" key="project">JGI-Pilot</profile>
     <profile namespace="globus" key="kernelprofile">zeptoos</profile>
     <profile namespace="globus" key="maxtime">1800</profile>
     <profile namespace="globus" key="alcfbgpnat">true</profile>
     <profile namespace="karajan" key="jobThrottle">0.63</profile>
     <profile namespace="karajan" key="initialScore">100000</profile>
     <workdirectory >/home/wilde/swiftwork</workdirectory>
     <scratch>/scratch</scratch>
   </pool>
</config>

--

Ive also tested with maxtime 3000 as in prior examples from Mihael.

Latest logs are on Eureka in:

eur$ pwd
/home/wilde/swift/lab
eur$ ls *log
23683.cobaltlog  23684.cobaltlog  cats-20091117-1101-oapf33ye.0.rlog 
cats-20091117-1101-oapf33ye.log  swift.log
eur$


Moving logs to logs/ as I test further.

First sign of trouble (that I can see) in the log above (*ye.log) is:

2009-11-17 11:01:48,582-0600 INFO  BlockQueueProcessor Plan time: 1
2009-11-17 11:01:50,785-0600 INFO  BlockQueueProcessor Updated 
allocsize: 8.66447649575794
2009-11-17 11:01:50,786-0600 INFO  BlockQueueProcessor allocsize = 
8.66447649575794, queuedsize = 1.0660596665516473, qsz = 1
2009-11-17 11:01:50,786-0600 INFO  BlockQueueProcessor Plan time: 1
2009-11-17 11:01:51,940-0600 DEBUG TaskImpl Task(type=JOB_SUBMISSION, 
identity=urn:cog-1258477276784) setting status to Completed
2009-11-17 11:01:51,941-0600 INFO  Block Block task status changed: 
Completed
2009-11-17 11:01:51,941-0600 WARN  Block Worker task failed: 
1117-011117-000000Block task ended prematurely


--

- Mike