On Jun 16, 2006, at 6:44 PM, Pete Wyckoff wrote:
[EMAIL PROTECTED] wrote on Fri, 16 Jun 2006 17:05 -0500:
I haven't been able to reproduce this specific error or get dbench to
fail for me, but Murali just committed a bunch of fixes to bugs in
the keyval changes I made recently that I could imagine causing a lot
of weird behaviors. If you find that these commits fix the dbench
errors you were seeing, let me know. I'd prefer to wait to cleanup
the error handling on the server (db->err and the weird EIO case) as
well as the msgpairarray problems until after the release.
That does change things, but it's still a bit funky.
My configure line looks like (straight tcp, no ib):
CFLAGS=-g ../pvfs2/configure \
--prefix=/usr/local/pvfs2-test \
--enable-shared \
--enable-segv-backtrace \
--enable-epoll \
--with-kernel=/usr/src/kernel/linux-2.6.16
Then I start 1 server with this config:
<Defaults>
UnexpectedRequests 50
LogFile /tmp/pbstmp.1225/pvfs2.log
EventLogging none
LogStamp usec
BMIModules bmi_tcp
FlowModules flowproto_multiqueue
PerfUpdateInterval 1000
ServerJobBMITimeoutSecs 30
ServerJobFlowTimeoutSecs 30
ClientJobBMITimeoutSecs 300
ClientJobFlowTimeoutSecs 300
ClientRetryLimit 5
ClientRetryDelayMilliSecs 2000
</Defaults>
<Aliases>
Alias ib14 tcp://ib14:3334
</Aliases>
<Filesystem>
Name pvfs2-fs
ID 1990198648
RootHandle 1048576
<MetaHandleRanges>
Range ib14 4-4294967297
</MetaHandleRanges>
<DataHandleRanges>
Range ib14 4294967298-8589934591
</DataHandleRanges>
<StorageHints>
TroveSyncMeta no
TroveSyncData no
CoalescingHighWatermark infinity
CoalescingLowWatermark 1
</StorageHints>
</Filesystem>
I'm using dbench-3.03 from the robl io tests distribution, with
command line, from a different machine, single client:
./dbench -D /pvfs -t 20 -c ./client.txt 10
Everything finishes now, but all the threads complain. They had
not previously done this in the past:
/bin/rm: cannot remove directory `/pvfs/clients/client0': No such
file or directory
/bin/rm: cannot remove directory `/pvfs/clients/client5': No such
file or directory
/bin/rm: cannot remove directory `/pvfs/clients/client8': No such
file or directory
/bin/rm: cannot remove directory `/pvfs/clients/client1': No such
file or directory
/bin/rm: cannot remove directory `/pvfs/clients/client3': No such
file or directory
/bin/rm: cannot remove directory `/pvfs/clients/client6': No such
file or directory
/bin/rm: cannot remove directory `/pvfs/clients/client7': No such
file or directory
/bin/rm: cannot remove directory `/pvfs/clients/client9': No such
file or directory
Hope this helps; at least it does not lock up my machines anymore.
I'm only picking on you since I saw some dbpf checkins that seem to
have been related (the -EIO issue).
I'm able to get something similar with dbench now after that commit,
so I'll look into that. I'm not sure about the EIO thing, it doesn't
seem like we should return EIO if the get of a dspace object fails,
but I think that may be code left over from Neill's first
implementation, so he may have done it for some reason that I've missed.
-sam
-- Pete
_______________________________________________
Pvfs2-developers mailing list
[email protected]
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-developers
