Date: Fri, 10 Feb 1995 21:06:08 -0600 (CST) From: Peter da Silva <peter@bonkers.taronga.com> To: terry@cs.weber.edu (Terry Lambert) Cc: davidg@Root.COM, jmb@kryten.atinc.com, hackers@FreeBSD.org Subject: Re: MIT SHM X11 extensions? (fwd) Message-ID: <199502110306.VAA26420@bonkers.taronga.com> In-Reply-To: <9502110149.AA14264@cs.weber.edu> from "Terry Lambert" at Feb 10, 95 06:49:45 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> Mount /usr over NFS and run apps from it from the an NFS client. Paging from an executable over NFS is a problem in and of itself anyway, since NFS doesn't handle locking the executable in any way, and it doesn't even have hooks. Try opening a tmp file on a nfs mounted /tmp and use the normal "open and unlink using the open file to keep the link around" trick. I don't think the fact that NFS is fundamentally broken is a good design criterion. There's so much else that's broken on NFS anyway. Executables opened over NFS should be copied into swap. > I consider the sudden explosion of core files on the client to be an > even more suprising consequence, especially sonsidering that you can > largely ignore swap unless you actually run out. And if you suddenly copy all your running text segments to swap you may. > Realistically, you could watermark it and start randomly crashing things > at 90% swap utilization on the client instead of putting up "out of > swap messages. And you were worried about all those core files?
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199502110306.VAA26420>