This part here looks like your problem. lustre2 should have a different host 
id. When it starts it's using the same handle range as lustre1.

  There was also a fix in 2.7.x or 2.8.x for the BDB issue below. I think the 
fix requires adding DBT_USERMEM to all the flags.

kevin

> 
> *pvfs2-server.conf-lustre1*
> 
> StorageSpace /pvfs2-storage-space
> HostID "tcp://lustre1:3334"
> LogFile /tmp/pvfs2-server.log
> 
> *pvfs2-server.conf-lustre2*
> 
> StorageSpace /pvfs2-storage-space
> HostID "tcp://lustre1:3334"
> LogFile /tmp/pvfs2-server.log
> 
> MDS + I/O server log:
> [D 05/11 23:13] PVFS2 Server version 1.5.1 starting.
> 
> 2nd I/O server log
> [D 05/11 23:14] PVFS2 Server version 1.5.1 starting.
> [E 05/11 23:15] TROVE:DBPF:Berkeley DB: DB_THREAD mandates memory allocation 
> flag on key DBT
> [E 05/11 23:15] TROVE:DBPF:Berkeley DB: DB_THREAD mandates memory allocation 
> flag on key DBT
> 
> Berkeley version for lustre1
> Name       : db4-devel
> Arch       : x86_64
> Version    : 4.3.29
> Release    : 10.el5
> 
> for lustre2
> i  libdb-dev                         4.7.25.3                          
> Berkeley Database Libraries [development]
> ii  libdb4.7                          4.7.25-7ubuntu2                   
> Berkeley v4.7 Database Libraries [runtime]
> ii  libdb4.7-dev                      4.7.25-7ubuntu2                   
> Berkeley v4.7 
> Thanks a lot,
> Andriy
> 
> On Tue, May 11, 2010 at 8:47 PM, Kevin Harms <[email protected]> wrote:
> 
>  Can you supply your fs.conf file? And perhaps the pvfs2-server.log from both 
> nodes?
> 
> kevin
> 
> On May 11, 2010, at 10:45 AM, Andriy Chut wrote:
> 
> > Hello everyone,
> >
> > Is there any way to deal with the problem of root handle ownership?
> > My installation:
> > MDS+ i/o server
> > 1) SL 5.0 with 2.6.17-CITI_NFS4_ALL-pnfs-1 from
> > http://www.citi.umich.edu/projects/asci/pnfs/linux/ and PVFS2 1.5.1
> > 2) i/o server: Ubuntu Server 9.04 with 2.6.31-14-server kernel.
> > PVFS2 1.5.1 (the same version)
> >
> > 1 installation with 1 MDS + 1 i/o configuration works just fine, but as 
> > soon as I add a second node (re-generating configuration, deleting and even 
> > re-formating the storage), I have a small issue while executing pvfs2-ping:
> >
> > (7) Verifying that root handle is owned by one server...
> >
> >    Root handle: 1048576
> > Failure: check root handle failed
> > PVFS_mgmt_setparam_all: Detailed per-server errors are available
> >
> > Per-server errors:
> >     Server: tcp://lustre1:3334: Reports ownership of root handle
> >
> >
> > # 2 handles for server 0
> > # 0 handles for server 1
> > File: <Root>
> >   handle = 1048576, type = Directory, server = 0
> > remaining handles:
> >
> > Is there (possibly) any way to get a clue on where the problem might be?
> > (I've been checking whether there's just one server instance running on 
> > each machine).
> >
> > I know that is irrelevant for PVFS2 2.8.1 at least, but, unfortunately, 
> > 1.5.1 seems to be the last (for now) version supporting PVFS-pNFS, so I 
> > kind of have no option, but to stick to it.
> > Any suggestions?
> > Thanks,
> > Andriy
> > _______________________________________________
> > Pvfs2-users mailing list
> > [email protected]
> > http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users
> 
> 

Attachment: smime.p7s
Description: S/MIME cryptographic signature

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

Reply via email to