[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: 980324: Solaris disks and netCDF

This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.

  • Subject: Re: 980324: Solaris disks and netCDF
  • Date: Wed, 25 Mar 1998 12:14:43 -0700
On Mar 25, 10:10am, John Keck wrote:
> Subject: Re: 980324: Solaris disks and netCDF
> Glenn,
>       Thanks for your reply.  The OS's we're using are
>
> Solaris: SunOS astro12 5.5 Generic sun4u sparc SUNW,Ultra-1
> SunOS: SunOS astro13 4.1.3 3 sun4c
>
> I've used "sum" on the data files on both SunOS and Solaris drives.
> They're definitely the same.
>
> > Are the tools built for the OS? (Or were they just copied,
> > without recompilation, from SunOS?)
> > Which version of the netcdf library is in use?
>
> The pipeline tools were compiled for the SunOS and reside on our network
> server.  Admittedly I do now know much about the workings of the network,
> I would think that it shouldn't make a difference what type of machine the
> data disk is mounted on if I'm running a pipeline tool compiled for SunOS
> on a SunOS machine.

Agreed. If you are running a SunOS binary on a SunOS machine,
it should work. Sounds to me like an NFS problem.

> Maybe that's just a naive presumption.  Nevertheless
> the type of machine the disk is mounted on seems to make all the
> difference.
>
> From netcdf.h it appears the version is
>
>                       v 1.3 1993/06/08 19:22:22
>
> If I need to patch the versions we have, how do I go about it?  Will the
> resulting executable then be usable on both SunOS and Solaris, or do I
> need to compile different versions for each?

This is a pretty old version but I don't think that is your problem.
If you do rebuild, you will want to build a version for each machine
you run on. It doesn't matter what architecture is providing the NFS
service.

We think your problem is at a lower level than netcdf.
Here is what my local expert has to say.


" Subject: SunOS 4.1.3 support question
" Glenn,
"
" Here are the two ideas I have regarding the SunOS 4.1.3 NFS question;
"
" 1) I'd apply the appropriate NFS jumbo patch from Sun;
"
"    100173-13    SunOS 4.1.3: NFS Jumbo Patch
"    102177-04    SunOS 4.1.3_U1: NFS Jumbo Patch
"
" 2) I'd make sure that UDP checksums are turned on in the kernel;
"
"    from Frequently Asked Questions for Comp.sys.sun.admin
"
"    ------------------------------------------------------------------
"    I keep getting "data corruption" when using NFS over a wan,
"    or slip/ppp link. What do I do?  or
"    Does anybody know how to enable UDP checksum on NFS?
"
"    The usaully cause for this(at least for SunOS4.x) is not
"    having udp checksumming turned on.
"
"    You can turn it at boot and while the machine is running
"    by using the following adb script:
"
"    #!/bin/sh
"    adb -w -k /vmunix /dev/mem << EOF
"    udp_cksum?W1
"    udp_cksum/W1
"    EOF
"
"    Install or call this script from rc.local.
"
"    Or edit /usr/kvm/sys/netinet/in_proto.c and change the
"    udp_cksum line (near the end from  udp_cksum = 0 to udp_cksum =1
"    and reconfigure your kernel and reboot)
"    ------------------------------------------------------------------