We've run into some problems with NFSv3 on RH Linux systems, but
they seem to dissappear after forcing mount/automount to use tcp.
For us, the problems were most visible when the NFS traffic had to
traverse a Cisco router (which was not very loaded at the time).
Try "mount -o tcp" or set localoptions='tcp' in /etc/init.d/autofs
mike
>From: Gerry Creager N5JXS <gerry.creager@xxxxxxxx>
>Subject: Re: LDM/GEMPAK NFS Question
>
>Considering the origins of NFS, it probably makes sense... Still, I
>have yet to see a problem attributable to NFS mounts here.
>
>My LDM data does come in on one machine, and has to traverse a router or
>2 to get to my Gempak machine for processing. I'll be most interested
>in performance after I can phyically colocate the 2 machines on the same
>subnet.
>
>Robert Mullenax wrote:
>> I have used RH Linux boxes as well with out any
>> huge problems, but they are still noticeably
>> slower than Solaris x86 boxes doing the same
>> thing, especially as load and file size increases.
>>
>>
>>
>> -----Original Message-----
>> From: Gerry Creager N5JXS
>> To: Robert Mullenax
>> Cc: 'Patrick O'Reilly'; ldm-users@xxxxxxxxxxxxxxxx; gembud@xxxxxxxxxxxxxxxx
>> Sent: 11/14/2003 8:57 PM
>> Subject: Re: LDM/GEMPAK NFS Question
>>
>> Then again, the only problem I can attribute to using nfs on RH9 is some
>>
>> remaining virus traffic.
>>
>> I'm running v3 here on Mesonet. We do some "interesting" cross-mounts
>> without problems.
>>
>> gerry
>>
>> Robert Mullenax wrote:
>>
>>>Under RH 6.2, we would have to specify NFS version 2, but this
>>>version of RH supports version 3. I would maybe look at NFS
>>>versions to see which version each box is running.
>>>
>>>As an aside, based on a lot of NFS use here..
>>>I will say that you will be much happier in the long run
>>>if you just ran Solaris x86 instead of RH 9. NFS was
>>>and continues to be a problem on Linux. You will never get
>>>the same NFS performance with a Linux NFS server as you
>>>would with a Solaris box (SPARC or x86).
>>>
>>>Robert
>>>
>>>-----Original Message-----
>>>From: Patrick O'Reilly [mailto:patrick.oreilly@xxxxxxx]
>>>Sent: Friday, November 14, 2003 12:29 PM
>>>To: ldm-users@xxxxxxxxxxxxxxxx; gembud@xxxxxxxxxxxxxxxx
>>>Subject: LDM/GEMPAK NFS Question
>>>
>>>
>>>Hello,
>>>
>>>I am switching my LDM machine from a Sun Solaris 8 machine to a PC
>>
>> running
>>
>>>RedHat 9. This PC will feed the data via NFS to 4 Sun Solaris 8
>>
>> machines,
>>
>>>who in turn run GEMPAK scripts on the data and post imagery to the
>>
>> web. My
>>
>>>prior setup had the 4 Suns mounting the NFS share of another Sun and
>>
>> things
>>
>>>went well. I just made the switchover, i.e. made the PC/Redhat
>>
>> machine the
>>
>>>NFS server to the 4 Suns, and the mount went fine. The problem is,
>>
>> now the
>>
>>>GEMPAK scripts take a very long time to run on the Suns. A simple
>>
>> script
>>
>>>takes a couple minutes where it used to take 20 seconds. I am
>>
>> assuming that
>>
>>>this is because of the communication between the PC and the Suns via
>>
>> the NFS
>>
>>>mount. Are there any NFS settings I can use to make the data read
>>
>> faster?
>>
>>>Is this even the problem? Has anyone faced this problem? Any wisdom
>>
>> on the
>>
>>>situation would be helpful. Thanks!
>>>
>>>Patrick
>>>_______________________________________
>>>Patrick O'Reilly
>>>Meteorological Decision Support Scientist
>>>The STORM Project - University of Northern Iowa
>>>patrick.oreilly@xxxxxxx ~ ph: 319-273-3789
>>>http://www.uni.edu/storm
>>>
>>>"No trees were killed in the making of this e-mail...however,
>>>a large number of electrons were horribly inconvenienced."
>>
>>
>
>--
>Gerry Creager -- gerry.creager@xxxxxxxx
>Network Engineering -- AATLT, Texas A&M University
>Cell: 979.229.5301 Office: 979.458.4020 FAX: 979.847.8578
>Page: 979.228.0173
>Office: 903A Eller Bldg, TAMU, College Station, TX 77843
