home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!olivea!sgigate!odin!fido!zola!zuni!anchor!olson
- From: olson@anchor.esd.sgi.com (Dave Olson)
- Newsgroups: comp.sys.sgi
- Subject: Re: algorithm for kill due to memory
- Keywords: vhand killed SYSLOG
- Message-ID: <njg95dc@zuni.esd.sgi.com>
- Date: 22 Jul 92 01:37:14 GMT
- References: <2810@contex.contex.com>
- Sender: news@zuni.esd.sgi.com (Net News)
- Organization: Silicon Graphics, Inc. Mountain View, CA
- Lines: 29
-
- In <2810@contex.contex.com> frank@marvin.contex.com (Frank Perdicaro) writes:
-
- | Could anybody out there point me to information on the alogrithm used
- | to erase a process that is killed due to lack of memory/swap. I am
- | talking about the function associated with the kernel message
- | "Process [%s] pid %d killed due to insufficient memory/swap."
- | ( this is from strings /unix | fgrep -i kill ), and not the one
- | associated with "Process [%s] pid %d killed due to %s (errno:%d)".
- | These messages are send to SYSLOG.
-
- It varies from release to release. It is biased towards large processes.
-
- | I would like to react to this action; the process returns with status
- | indicating a signal 9 was sent to it. Is the process really given a
- | signal? As there appears to be no way to prevent any particular
-
- Yes, and SIGKILL is not catchable.
-
- | process from being expunged, is there any way predict which process
- | gets hit?
-
- Not in the general case.
-
- You could configure the system so this doesn't happen, see
- availsmem_accounting, in master.d/kernel.
- --
- Let no one tell me that silence gives consent, | Dave Olson
- because whoever is silent dissents. | Silicon Graphics, Inc.
- Maria Isabel Barreno | olson@sgi.com
-