Cachefs messages

[Gadi]

Hi,

We have a lot off B1000 and B2000 running 10.20. We use NIS and NFS and the cachfs is configured too.

While normal working the output of dmesg is something like this:
cachefs_getattr: vap->va_nodeid: 61177
cachefs_getattr: backvp: 37e6004
cachefs_getattr: vap->va_nodeid: 61263
cachefs_getattr: backvp: 37e6404
cachefs_getattr: vap->va_nodeid: 61177
cachefs_getattr: backvp: 37e6004
cachefs_getattr: vap->va_nodeid: 61263
cachefs_getattr: backvp: 3a1b004

I checked in /var/adm/syslog/syslog.log and I see it's happends every 2 Min.

I'm not sure this is a real problem or not.

So how can I get out this messages?

[Alex Glennie]

[PHNE_21108/PACHRDME/English] and [PHNE_20313/PACHRDME/English] introduced behavior with CacheFS that can cause
excessive logging to /var/adm/syslog/syslog.log.This
annoyance can become a major problem if it consumes all the
available space on the file system.
- For every file created in a CacheFS directory, two messages
will be logged to syslog.log. The messages are similar to:
vmunix: cachefs_getattr: vap->va_nodeid:
vmunix: cachefs_getattr: backvp: 191c804
- This behavior also exists in superseding patches [PHNE_20957/PACHRDME/English]
and PHNE_21108. This behavior does not exist in the patch
that precedes [PHNE_20313/PACHRDME/English] , which is PHNE_20021.
- The only workaround to avoid the logging with one of these
patches is to disable CacheFS. If disabling CacheFS or
removing the patches is not an option, it is recommended to
monitor the size of the /var/adm/syslog/syslog.log file to
avoid filling up the file system.

This will be fixed in the next NFS Kernel patch release (this was written 14
Jun 00) .... this may go some way to explain your problem ?

[Cheryl Griffin]

Gadi,

This is a known issue and will be fixed in the next NFS cumulative patch.
If you would like I can register your email address in the SR notify database so that you will automatically be notified when the fix is released for this issue.

Please let me know if you want me to do this.

Best Wishes,
Cheryl

[Anthony deRito]

Cahcfs is a Solaris filesystem type. Are you running this on HP-UX 10.20? If you somehow are, it just looks like they are some sort of debug statements regarding the cache state. Please explain what cachfs you are using if I am on the wrong track. Sorry couldn't help more.

Tony

[Gadi]

Thank you Alex and Chryl for your help.
Cheryl please register my e-mail address in the SR notify database. My e-mail address is:
gadikatz@hotmail.com.

Best wishes,
Gadi

[Rita C Workman]

Just a thought...I get similar output into syslog because I have inetd logging enabled.
It's a pain to read the syslog with all the registrar entries. I found by just running a baby script every morning, it made reading the logs easier and kept /var down.
Just do a cat syslog.log | grep -v -e 'stmt you want to remove' -e 'next stmt you want removed'.
Then just copy the results to a tmp file. Then just copy the tmp file back to syslog.log.
Not fancy coding I know...but it's quick and it cleans out the messages you don't need to see anyway.
Regards,

[Cheryl Griffin]

Gadi,

I have you registered.
Take Care,
Cheryl

[Gadi]

For a few days I'm out of my office. I want to thank everyone for the help I get from you.

Regards,

Gadi Katz.

[Cheryl Griffin]

This is just an update to this thread.

The patch for this issue has been released:

Patch Name: PHNE_22117

Patch Description: s700_800 10.20 NFS Kernel General Rel & Perf Patch

Best Wishes!
Cheryl