topic Re: Performance in Operating System - HP-UX

Performance

KapilRaj — Thu, 28 Jul 2005 05:25:31 GMT

Sure my server is performing very very bad. sar shows alots of wio round abt 80-90 %. I know the problem is NFS. But is there a command which will point out which nfs filesystem or which process is stuck ?

Regds,

Kaps

Re: Performance

Pedro Cirne — Thu, 28 Jul 2005 05:33:01 GMT

Hi,

Check nfsstat. May it helps.

Enjoy :)

Re: Performance

Eknath — Thu, 28 Jul 2005 05:39:55 GMT

Hi Kapil,

Check which NFS is utilised more. The busiest NFS normally causes the performance issue. If its possible mount each NFS seperately and find out.

Cheers!!!
eknath

Re: Performance

Patrice Le Guyader — Tue, 02 Aug 2005 15:25:45 GMT

Hi Kaps,

Try nfstat -rc on the nfs clients, and look at the values of timeout and retrans.
If they are high, whereas the value of badxid is near 0 then the packets to the server are lost, try reducing the value of wsize and rsize to 4096 on the client's mount points parameters.
If timeout and badxids are at the same level then your server is probably to slow, try to increase the value of timeout on the client's mount point parameters.

An other possibility is that you don't have enough nfsd daemons on the server.
Try "nfsstat -s" anf if the value of "socket overflows" is high then probably your server hadn't enough nfsd processes. Try modify /etc/rc.config.d/nfsconf or with "/usr/sbin/nfsd number_processes". On the client side it's the biod processes.
I'm not an NFS expert, but I wonder if the default numbers of nfsd and biod are different between an 11 and 11i. To verify.

Regards,
Pat.

Re: Performance

melvyn burnard — Tue, 02 Aug 2005 15:57:04 GMT

Take a look at this document, it may help you:

http://docs.hp.com/en/1435/NFSPerformanceTuninginHP-UX11.0and11iSystems.pdf

Re: Performance

RAC_1 — Tue, 02 Aug 2005 15:58:54 GMT

As all other mentioned, you need to look at nfsstat. If you are using udp for nfs transport, you may want to look at increasing biod and nfsd. But chances are that you most probablly might be using tcp.

NFS patches is also very important thing to look at.

Also, a must read is as follows.

http://docs.hp.com/hpux/onlinedocs/netcom/NFS_perf_tuning_hpux110_11i.pdf

Re: Performance

generic_1 — Tue, 02 Aug 2005 17:35:28 GMT

What NFS options and tweaks are you using?
Tweaking some of the export parameters can help.

PS I think NFS stands for Nowork FileSystem :).

Re: Performance

generic_1 — Tue, 02 Aug 2005 17:37:07 GMT

Hey if anyone here works on the NFS distro for HPUX get them to add multithreading to this piece of junk. That would be very helpful.

Re: Performance

morganelan — Thu, 08 Sep 2005 21:46:32 GMT

Important early step in troubleshooting any Network performance issues:
#How many network hops (i.e. bridges, hubs, routers, switches, etc.) do network packets traverse between the client and the server systems?
#What is the speed of each link separating these systems?
#Does your network equipment use auto-negotiation to set speed and duplex settings?
#Are your network interfaces configured for half-duplex or full-duplex mode?
#Do your switch port settings match the speed and duplex settings of your host interfaces?
#What is the maximum transmission unit (MTU) size of the links between these systems?
#If the links are using different MTU sizes, how are the packets being translated? For example, if the NFS client resides in an FDDI ring and uses an MTU size of 4352 and the NFS server uses a 100BT interface with an MTU size of 1500, how are the 4352 byte packets from the client being fragmented into 1500 byte packets for the server?
#Do packets sent from the client to the server take the same route through the network as the packets sent from the server to the client?

You can man netstat to find what you want to know about your network statistics.The traceroute(1M) tool provides a simple means of determining the path through the network taken by packets sent from one system to another.Another tool shipping with HP-UX that can simplify the process of collecting network topology information is ping(1M).The lanadmin command allows a system administrator to display many useful statistics kept by the LAN driver subsystem, regardless of the interface type.

You can try using tools such as:
ttcp:http://ftp.arl.mil/ftp/pub/ttcp
netperf:http://netperf.org