ksh: hist_flush error when logging into HP-UX

[John Moorhead_2]

We support a NIS-based mixed environment of HP-UX servers and Linux RHEL4WS desktops. The user's NIS home directories reside on an 11.23 HPUX server. The users often login from one Linux box to another, or from a Linux box to an HPUX box or from HPUX to Linux, so the users $HOME is mounted sometimes on Linux, sometimes on HP-UX, sometimes on both at the same time.

Several users are reporting that they are getting the following error message popping up at login or after issuing commands:
"ksh: hist_flush: EOF seek failed errno=116".

There is very little I can find on the Web about this error (there was one case involving Apache where a user got this same error, but nothing was mentioned about how to resolve this issue).

It appears that users encounter this error more when they hop from a Linux system to an HPUX system rather than from HPUX to HPUX, HPUX to Linux or from Linux to Linux.

Is there any known fix or recommendation? Would increasing HISTSIZE help?

[D Block 2]

John- check out linux errno.h for 116. The error points to a stale file handle from NFS. NOthing to do w/ ksh history file Size.

Sounds like issues related to Stale file handles in NFS - see

http://linux.derkeiler.com/Mailing-Lists/Kernel/2003-11/2122.html