HP-UX & Netapp connectivity

[Roy Francis]

I am trying to setup a HP_UX 11.11 boxes to mount NFS from a NetApp filer.

I can get the HP to mount from Solaris and Linux boxes with no problems. The HP will mount from the Filer but is extremely slow.

I have followed advice from NetApps who now say the problem is with HP-UX.

Any ideas on anything I can try?

[doug mielke]

I have 2 netapp FC's connected to HP/UX on N and K classes, and they're runningh great.
For speed look at a couple of things.

traceroute filername

to ensure you're not taking the long way to the filer.

of cource, ensure the ping times are adaquate.


and,
look for a dns error. Try nslookup filername for a delay.
to avoid DNS issues
put filer name in etc/hosts. then edit /etc/nsswitch.conf to look locally first.

[A. Clay Stephenson]

You don't give us much to go on but I have never had a problem doing NFS mounts from any source including PC-based NFS servers under HP-UX. My best guess is that you have a hostname resolution problem. For example, if /etc/nsswitch.conf is set to use DNS then NIS and finally files and your NetApp box is not defined in DNS or NIS then both of those must timeout before finally finding the entry in /etc/hosts.

Ping by the exact hostname you are using in both directions and see if the hostname is quickly resolved.

You might also play with the NFS mount options vers= and proto=. Man mount_nfs for details.


One other thought: Are these hard NFS mounts or are you using the automounter (perhaps in conjunction with NIS/NIS+ maps)? It always helps to knoiw the real problem we are fighting.

[Jayan_2]

dear ,

Did u check up the switch in between the Netapps server and HP server we had a similar issue in MCS?G environment and got resolved by replacing the backbone switch.

Apply the latest support plus patches.

With regards
Jayan

[Stefan Farrelly · Accepted solution]

Weve got our 11.11 boxes working fine over NFS to Netapps - but we had to follow all their recomendations here;

http://www.netapp.com/tech_library/3146.html

Follow it all - patches, kernel params, gigabit lan card settings.

[Roy Francis]

I have tried the suggestions.

The system is in DNS. nslookups, traceroute and pings are all very quick. nsswitch.conf states to use files then DNS.

The mounts are being hard mounted from the filer. Hard mounts due to the data being access all the time.

The switch is set to giggabit on the filer and 100/Full on the HP box.

I have tried the infomation at http://www.netapp.com/tech_library/3146.html but this makes no difference.

If I request a data copy from the filer to the local disk it goes in bursts. First 64K quickly then stops for a while before doing a bit more.

I have standard installs on all systems which is supposed to work but for my systems do not

[Elena Leontieva]

Roy,

Try to do cp and ftp of some relatively big file in many from/to combinations, like:

cp Netapp - HP
cp Netapp - Solaris
ftp HP - Solaris , etc.

It could help you identify where the mismatch is. Do you have just one NIC per box? Are the servers and Netapp on the same subnet?

I assume that your HP box is NFS-patched.

Elena.

[Niraj Kumar Verma]

Hi Roy,

I too have the similar problem !!

Can you please let me know what did u do to solve the problem ??

Thanks & Regards
-Niraj