1833599 Members
3536 Online
110061 Solutions
New Discussion

NFS Slow

 
Fernando Jose P de Souz
Regular Advisor

NFS Slow

Hi,

Servers: rx2620
HP-UX 11.23
MC Service Guard 11.17

I would like to just mount together the file systems a NFS mount point in both servers.

but my problem is when i try to change the package to another host in the cluter is VERY SLOW to mount the NFS.

My package.sh in atach.


pec1-/# cmviewcl

CLUSTER STATUS
cluster1 up

NODE STATUS STATE
pec1 up running

PACKAGE STATUS STATE AUTO_RUN NODE
sappec up running enabled pec1

NODE STATUS STATE
pec2 up running
pec1-/#


pec1:
pec1-/# bdf
Filesystem kbytes used avail %used Mounted on
/dev/vg00/lvol3 573440 420536 151720 73% /
/dev/vg00/lvol1 311296 140280 169776 45% /stand
/dev/vg00/lvol8 4718592 2350448 2350528 50% /var
/dev/vg00/lvol7 6356992 2547792 3779488 40% /usr
/dev/vg00/lvol9 10240000 5778675 4182526 58% /usr/sap
/dev/vg00/lvol4 2048000 1425328 618520 70% /tmp
/dev/vg00/lvol6 6275072 3960128 2296872 63% /opt
/dev/vg00/lvol5 32768 9976 22640 31% /home
/dev/sapmnt/lvol1 6275072 4805718 1377549 78% /export/sapmnt/PEC
/dev/vg01/lvol2 507904 3813 472645 1% /oracle
/dev/vg01/lvol3 4096000 133055 3715266 3% /oracle/client
/dev/vg01/lvol4 10240000 4858260 5045388 49% /oracle/stage/102_64
/dev/vg01/lvol5 6291456 585851 5354574 10% /oracle/PEC
/dev/vg01/lvol6 6144000 4234421 1790791 70% /oracle/PEC/102_64
/dev/vg01/lvol7 81920000 28684812 49907990 36% /oracle/PEC/sapdata1
/dev/vg01/lvol8 81920000 27471524 51045449 35% /oracle/PEC/sapdata2
/dev/vg01/lvol9 81920000 35630948 43395989 45% /oracle/PEC/sapdata3
/dev/vg01/lvol10 81920000 32251780 46563958 41% /oracle/PEC/sapdata4
/dev/vg02/lvol1 2048000 119376 1808092 6% /oracle/PEC/mirrlogA
/dev/vg02/lvol2 2048000 119376 1808092 6% /oracle/PEC/mirrlogB
/dev/vg02/lvol3 40960000 7550620 31321513 19% /oracle/PEC/oraarch
/dev/vg02/lvol4 2048000 136417 1792116 7% /oracle/PEC/origlogA
/dev/vg02/lvol5 2048000 136417 1792116 7% /oracle/PEC/origlogB
/dev/vgsappec/lvol1
10469376 3215482 6800718 32% /usr/sap/PEC
/dev/vg02/lvol6 5210112 1308198 3658062 26% /oracle/PEC/saptemp1
/dev/vg02/lvol7 15351808 20236 14373356 0% /oracle/PEC/sapreorg
pec0:/export/sapmnt/PEC
6275072 4805720 1377544 78% /sapmnt/PEC
pec1-/#


pec2:
pec2-/# bdf
Filesystem kbytes used avail %used Mounted on
/dev/vg00/lvol3 573440 283136 288048 50% /
/dev/vg00/lvol1 1835008 122760 1698976 7% /stand
/dev/vg00/lvol8 5120000 2877160 2225384 56% /var
/dev/vg00/lvol7 7274496 2554280 4683448 35% /usr
/dev/vg00/lvol10 12288000 21448 11499896 0% /usr/sap
/dev/vg00/lvol4 524288 19792 501544 4% /tmp
/dev/vg00/lvol6 7274496 4066480 3182968 56% /opt
/dev/vg00/lvol5 114688 10024 103872 9% /home
pec0:/export/sapmnt/PEC
6275072 4805720 1377544 78% /sapmnt/PEC
pec2-/#


4 REPLIES 4
Fernando Jose P de Souz
Regular Advisor

Re: NFS Slow

My script in attach.
Bill Hassell
Honored Contributor

Re: NFS Slow

If NFS is normal speed after the mount is complete, I would suspect DNS problems. To eliminate DNS problems, change your nsswitch.conf file for hosts: to use files then dns, like this:

hosts: files [NOTFOUND=continue UNAVAIL=continue] dns

then put all of your production machine addresses into /etc/hosts. Production systems won't change very much and you avoid delays and problems with another group maintaining DNS.


Bill Hassell, sysadmin
Fernando Jose P de Souz
Regular Advisor

Re: NFS Slow

The problem isn't in the DNS.

The problem is:

The NFS(/export/sapmnt/PEC) is in the package .

The IP Package:
192.210.1.216 pec0 # package Hostname

Message in the log:
/sapmnt/PEC: stat: Stale NFS file handle
Stephen Doud
Honored Contributor

Re: NFS Slow

The script you provided shows:
#Revision: 1.10
however no HP copyright is present.

HP's scripts contain a copyright - so your script is home-grown.

What version of HPUX is running?

Make certain the Volume Group minor number matches between servers.

A stale handle indicates the NFS clients' linkage to the real file system located on the NFS server has changed or otherwise become invalid. This may happen as a result of moving an NFS package from one server to another, and the LVM volume group minor number differing between servers (as an example).

As a diagnostic in the applic_halt_cmds() in your script, use 'nfsstat -m' to identify any NFS file systems that may have stale handles.