Hello,

I use orangefs-2.8.5 and kernel 3.0.17 on my amd64 test host running
on Gentoo. When orangefs is configured as follows, I experience hang
on file creation on the mounted pvfs2 filesystem.

./configure \
        --disable-nptl-workaround \
        --disable-redhat24 \
        --enable-epoll \
        --enable-shared \
        --enable-thread-safety \
        --enable-verbose-build \
        --sysconfdir=/etc/pvfs2 \
        --enable-kernel-aio \
        --enable-perf-counters \
        --enable-segv-backtrace \
        --enable-fuse \
        --disable-karma \
        --enable-mmap-racache \
        --disable trusted-connections \
        --enable-kernel-sendfile \
        --enable-server \
        --disable-static \
        --disable-static-server \
        --without-openib \
        --without-mtrace \
        --without-mx \
        --with-ssl=/usr/ \
        --with-bmi-tcp \
        --without-valgrind \
        --enable-threaded-kmod-helper \
        --with-kernel=/usr/src/linux \
        --enable-fast \
        --without-berkdb-debug

Debug options does not matter: I enabled them only to provide
additional debug data in this report; with this options disabled
filesystem behaviour is the same.

Then I created config file (attached fs.conf) and generated
filesystem using --mkfs. Afterwards I started a server:
/usr/sbin/pvfs2-server -p /var/run/pvfs2.pid /etc/pvfs2/fs.conf
and created fstab entry:
tcp://subaru:3334/pvfs2-fs /mnt/cluster pvfs2
defaults,intr,noauto 0 0

pvfs2-check-config and pvfs2-ping confirm validity of my setup (see
attached check.log).

I load module and start the client:
$ lsmod
Module                  Size  Used by
pvfs2                 119742  2 

pvfs2-client -f -L /var/log/pvfs2/client.log --gossip-mask=0xFFFF

Mounting filesystem is ok:
# mount /mnt/cluster
# mount | grep cluster
tcp://subaru:3334/pvfs2-fs on /mnt/cluster type pvfs2 (rw)

I see an empty tree:
$ ls -1
lost+found

I can create and remove directories without a problem.
But when I try to create some file (via touch, cp, pvfs2-cp — this
doesn't matter at all) filesystem hangs, forever, until daemons are
killed with SIGKILL.

Server and client log files as well as system log messages relevant
to this problem are attached to this mail as well. Log are cut
between start and hang-forever points.

I'll try to disable some implicitly dangerous configure options in
order to get this stuff to work.

Best regards,
Andrew Savchenko
<Defaults>
        UnexpectedRequests 50
        EventLogging none
        EnableTracing no
        LogStamp datetime
        BMIModules bmi_tcp
        FlowModules flowproto_multiqueue
        PerfUpdateInterval 1000
        ServerJobBMITimeoutSecs 30
        ServerJobFlowTimeoutSecs 30
        ClientJobBMITimeoutSecs 300
        ClientJobFlowTimeoutSecs 300
        ClientRetryLimit 5
        ClientRetryDelayMilliSecs 2000
        PrecreateBatchSize 0,32,512,32,32,32,0
        PrecreateLowThreshold 0,16,256,16,16,16,0

        DataStorageSpace /pvfs2-storage-space
        MetadataStorageSpace /pvfs2-storage-space

        LogFile /var/log/pvfs2/server.log
</Defaults>

<Aliases>
        Alias subaru tcp://subaru:3334
</Aliases>

<Filesystem>
        Name pvfs2-fs
        ID 311495942
        RootHandle 1048576
        FileStuffing yes
        <MetaHandleRanges>
                Range subaru 3-4611686018427387904
        </MetaHandleRanges>
        <DataHandleRanges>
                Range subaru 4611686018427387905-9223372036854775806
        </DataHandleRanges>
        <StorageHints>
                TroveSyncMeta yes
                TroveSyncData no
                TroveMethod alt-aio
        </StorageHints>
</Filesystem>
[E 18:46:37.295230] PVFS Client Daemon Started.  Version 2.8.5-orangefs
[D 18:46:37.306830] [INFO]: Mapping pointer 0x7f447265a000 for I/O.
[D 18:46:37.307549] [INFO]: Mapping pointer 0x1755000 for I/O.
[D 03/02/2012 18:45:55] PVFS2 Server version 2.8.5-orangefs starting.
[E 03/02/2012 21:10:44] 
PVFS2 server got signal 15 (server_status_flag: 1032191)

Attachment: system.log.xz
Description: Binary data

$ pvfs2-check-config
Config: (null)
Check Complete.

$ pvfs2-ping -m /mnt/cluster

(1) Parsing tab file...

(2) Initializing system interface...

(3) Initializing each file system found in tab file: /etc/fstab...

   PVFS2 servers: tcp://subaru:3334
   Storage name: pvfs2-fs
   Local mount point: /mnt/cluster
   /mnt/cluster: Ok

(4) Searching for /mnt/cluster in pvfstab...

   PVFS2 servers: tcp://subaru:3334
   Storage name: pvfs2-fs
   Local mount point: /mnt/cluster

   meta servers:
   tcp://subaru:3334

   data servers:
   tcp://subaru:3334

(5) Verifying that all servers are responding...

   meta servers:
   tcp://subaru:3334 Ok

   data servers:
   tcp://subaru:3334 Ok

(6) Verifying that fsid 311495942 is acceptable to all servers...

   Ok; all servers understand fs_id 311495942

(7) Verifying that root handle is owned by one server...

   Root handle: 1048576
     Ok; root handle is owned by exactly one server.

=============================================================

The PVFS2 filesystem at /mnt/cluster appears to be correctly configured.

Attachment: pgpFRMvmnwZMS.pgp
Description: PGP signature

_______________________________________________
Pvfs2-users mailing list
[email protected]
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users

Reply via email to