Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!sdd.hp.com!hplabs!hpl-opus!hpcc05!hpgva1!hpuamsa!franks
From: franks@hpuamsa.neth.hp.com (Frank Slootweg CRC)
Newsgroups: comp.sys.hp
Subject: Re: sar reporting incorrectly on 9000/822
Message-ID: <28510047@hpuamsa.neth.hp.com>
Date: 16 May 91 14:49:16 GMT
References: <1991May13.110328.28522@minyos.xx.rmit.oz.au>
Organization: HP-Sales Office-The Netherlands
Lines: 18

Re: This "bug" is actually a feature.

  The inode table is used for both open inodes and for used_to_be_open_
but_now_closed inodes. For open inodes it is a real table: If it is
full and a new inode must be opened it overflows. For ex-open inodes it
is a cache: I.e. old entries are kept untill a new inode must be opened
and there is no free entry.

  So like any good cache it is (nearly) full (nearly) all of the time
and that is what "sar" reports. I understand that HP proprietary (sp?)
programs like GlancePlus/UX will report both open and ex-open cached
inodes in future. I do not know if the UNIX standard bodies allow us to
add such features to the "sar" output.

  B.T.W. The open-file table is a table, not a cache, so if *it* is full
then no more files can be opened and the kernel will say so.

Frank Slootweg, HP, Dutch Customer Response Center.