bug-hurd
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: malloc() patches round 3


From: Chiaki Ishikawa
Subject: Re: malloc() patches round 3
Date: Fri, 24 Aug 2001 07:46:55 +0900 (JST)

X-PMC-CI-e-mail-id: 16396 

I am joining this discussion rather late, but

>I've had weird exim failures cause my incoming email to get hosed.  It
>was a serious problem, and it could have been avoided if the system
>had simply rebooted.  It doesn't matter to me whether exim is hosed
>because it hands out "memory exhausted" errors to incoming SMTP
>connections, or if it responds in some other incomprehensible way:
>*BOTH* are just as wrong.

At least, before the suggested reboot of OS, I would like the system to
keep record of the vmstat-like memory usage information, and
process table dump, etc. for the later investigation.
Whatever the method the OS uses is up to the implementor
barring the exhaustion of file system space.
(I am running ps [and other assortment of
system status dump programs] every couple of minutes and record it
on a remote NFS mounted server file so that
a production machine that crashes/hung mysteriously can be
analyzed for any tell-tale signs about what caused the problem.
This works rather well on Solaris/SunOS.)

BTW, the Apollo Domain OS, in which it was possible to
page out even part of the kernel (in the days of monolithic kernel,
that is), seemed to handle the virtual memory exhaustion rather
gracefully. (Second hand story heard from someone who built
large AI system using lisp and observed the Domain OS behavior.)
This was back in the 1980's.


Ishikawa






reply via email to

[Prev in Thread] Current Thread [Next in Thread]