Re: NFS file truncating

From: Alan Vidmar (Alan.Vidmar@COLORADO.EDU)
Date: Wed Feb 11 2004 - 18:48:20 EST


As a side note to this discussion: how do people feel about defragging RAID 5
arrays?

Not needed? Unnecessary? What do you think?
Alan

On 11 Feb 2004 at 18:29, cbaker@GOODYEAR.COM wrote:

Date sent: Wed, 11 Feb 2004 18:29:18 -0500
Send reply to: IBM AIX Discussion List <aix-l@Princeton.EDU>
From: cbaker@GOODYEAR.COM
Subject: Re: NFS file truncating
Originally to: IBM AIX Discussion List <aix-l@Princeton.EDU>
To: aix-l@Princeton.EDU

> /dev/nfs5lv00 1700134912 73600168 96% 337928 3% /hpc/light-5
>
> rdsnfs5:root: /
> > defragfs -q /hpc/light-5
> statistics before running defragfs:
> number of free fragments 9199989
> number of allocated fragments 203316875
> number of free spaces shorter than a block 88510
> number of free fragments in short free spaces 460609
>
> rdsnfs5:root: /
> > defragfs -r /hpc/light-5
> statistics before running defragfs:
> number of free fragments 9200753
> number of allocated fragments 203316111
> number of free spaces shorter than a block 88506
> number of free fragments in short free spaces 460607
>
> { Hung up on this report..........}
>
> Went ahead and typed in the "defragfs /hpc/light-5" after the report is
> done. Had to leave. Will send the results of the report and the defrag
> tomorrow.
>
> Thanks,
>
> Christopher M. Baker
> Senior Technical Support Analyst
> DSE/TCO
> Goodyear Tire and Rubber Company
>
> =================================================
> Contains Confidential and/or Proprietary Information.
> May not be copied or disseminated without the expressed
> written consent of The Goodyear Tire & Rubber Company.
> =================================================
>
>
>
>
> pSeries AIX Geek
> <aixgeek@YAHOO.C To: aix-l@Princeton.EDU OM>
> cc: (bcc: Chris Baker/NA/GDYR)
> Sent by: IBM AIX Subject: Re: NFS file truncating
> Discussion List <aix-l@Princeton .EDU>
>
>
> 02/11/2004 05:01
> PM
> Please respond
> to IBM AIX
> Discussion List
>
>
>
>
>
>
> Interesting. Still, I'd suspect fragmentation. If
> you run defragfs with the "what-if" option (sorry,
> don't have man pages in front of me), what does it
> report?
>
> [There's a -q and a -r flag on defragfs. I can't
> remember what each does, but the man page for defragfs
> will show you what options you have to run to show
> what it WOULD do, if you let it.]
>
> Another good test would be to use dd if see if you can
> replicate the same error by creating a 4MB+ file on
> the server itself (i.e., on the JFS file system, not
> on the NFS mount). If you get the same error, I'll
> still opine that it's a fragmentation issue.
>
> - pAG
>
>
> --- cbaker@GOODYEAR.COM wrote:
> > No JFS errors in the errpt on the P655. (Actually,
> > at the time of the last
> > few failed file copies, there have been no errpt
> > entries at all.)
> >
> > Christopher M. Baker
> > Senior Technical Support Analyst
> > DSE/TCO
> > Goodyear Tire and Rubber Company
> >
> > =================================================
> > Contains Confidential and/or Proprietary
> > Information.
> > May not be copied or disseminated without the
> > expressed
> > written consent of The Goodyear Tire & Rubber
> > Company.
> > =================================================
> >
> >
> >
> >
> > pSeries AIX Geek
> > <aixgeek@YAHOO.C To:
> > aix-l@Princeton.EDU
> > OM> cc:
> > (bcc: Chris Baker/NA/GDYR)
> > Sent by: IBM AIX
> > Subject: Re: NFS file truncating
> > Discussion List
> > <aix-l@Princeton
> > .EDU>
> >
> >
> > 02/11/2004 12:21
> > PM
> > Please respond
> > to IBM AIX
> > Discussion List
> >
> >
> >
> >
> >
> >
> > Are you getting JFS fragmentation errors in the
> > error
> > report? On the 4MB boundary, you go from using
> > single
> > indirection to double indirection and if you don't
> > have enough CONTIGUOUS free space, you'll get this
> > issue.
> >
> > Look in your error report for fragmentation errors,
> > and look into running defragfs.
> >
> > - pAG
> >
> >
> > --- cbaker@GOODYEAR.COM wrote:
> > > Folks,
> > >
> > > We have a P655 running as a NFS server. Has been
> > > running for months
> > > without any problems. But, about a week ago, one
> > of
> > > our applications
> > > started failing when copying files into one of the
> > > NFS mounted filesystems.
> > >
> > > Sometime it works and some times it only partially
> > > copies.
> > >
> > > Strange thing is, many times when it fails, the
> > > files that do not
> > > completely copy are all the exact same size when
> > > they fail. That size is
> > > 4,194,304 bytes.
> > > In other words, we copy multiple files, the first
> > > one is less than 4 MBytes
> > > and it copies fine. The next three files are
> > bigger
> > > than 4 MB, some as big
> > > as 600 MB, but each copy gives an I/O error then
> > > moves on to the next file.
> > > When it is all done, the three larger files are
> > each
> > > only 4,194,304 bytes
> > > big.
> > >
> > >
> > > DETAILS:
> > > ========
> > >
> > > 1) The copy is really a "rcp" running on an HP
> > J6000
> > > (HP-UX). It is remote
> > > copying a file from a third box ( a Tru64 system)
> > > back to itself. But it
> > > is copying to a filesystem that is the NFS mounted
> > > filesystem off of the
> > > P655.
> > >
> > > 2) We see no errors on the Tru64 box, the HP-J6000
> > > or the P655.
> > >
> > > 3) The files being copied are mainly binary files.
> > >
> > > 4) The copy is done using a rcp that is part of
> > LSF
> > > (Load Share Facility).
> > > It seems to fail with this error:
> > >
> > > "copyFile: write(802816) failed, No such
> > > file or directory."
> > >
> > > This is followed by: "lsrcp: main()
> > > failed, try rcp...."
> > >
> > > Then it appears to try to use "rcp": and
> > that
> > > fails also with an
> > > error of: "I/O error"
> > >
> > > 5) Strange thing is, the copy (lsrcp) worked just
> > > seconds before on a small
> > > ASCII log file (the first file) from and to the
> > same
> > > directories??
> > >
> > > 6) This problem started about the same time we
> > > increased the load on the
> > > NFS server. But, system does not seem to be
> > > anywhere near loaded.
> > >
> > > 7) The filesystem in question resides on EMC
> > > Clariion disks that are
> > > attached via fiber channel to the P655.
> > >
> > > 8) AIX 5.1 ML04 on NFS server.
> > >
> > >
> > > Any help, suggestion or sly remarks would be
> > greatly
> > > appreciated.
> > >
> > > Thanks,
> > >
> > > Christopher M. Baker
> > > Senior Technical Support Analyst
> > > DSE/TCO
> > > Goodyear Tire and Rubber Company
> > >
> > > =================================================
> > > Contains Confidential and/or Proprietary
> > > Information.
> > > May not be copied or disseminated without the
> > > expressed
> > > written consent of The Goodyear Tire & Rubber
> > > Company.
> > > =================================================
> >
> >
> > __________________________________
> > Do you Yahoo!?
> > Yahoo! Finance: Get your refund fast by filing
> > online.
> > http://taxes.yahoo.com/filing.html
>
>
> __________________________________
> Do you Yahoo!?
> Yahoo! Finance: Get your refund fast by filing online.
> http://taxes.yahoo.com/filing.html

"I don't have time to be impatient."
Alan R. Vidmar Assistant Director of IT
Office of Financial Aid University of Colorado
Alan.Vidmar@Colorado.EDU (303)492-3598
*** This message printed with 100% recycled electrons ***



This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 22:17:35 EDT