From owner-freebsd-questions Thu Oct 19 15:14:32 1995 Return-Path: owner-questions Received: (from root@localhost) by freefall.freebsd.org (8.6.12/8.6.6) id PAA04097 for questions-outgoing; Thu, 19 Oct 1995 15:14:32 -0700 Received: from ponder (ponder.csci.unt.edu [129.120.3.16]) by freefall.freebsd.org (8.6.12/8.6.6) with SMTP id PAA04092 for ; Thu, 19 Oct 1995 15:14:12 -0700 Received: from sunra.csci.unt.edu by ponder (5.61/1.36) id AA08601; Thu, 19 Oct 95 17:14:37 -0500 Received: (jason@localhost) by sunra.csci.unt.edu (8.6.11/8.6.4) id RAA08802 for questions@freebsd.org; Thu, 19 Oct 1995 17:18:21 -0500 Date: Thu, 19 Oct 1995 17:18:21 -0500 From: Jason Brazile Message-Id: <199510192218.RAA08802@sunra.csci.unt.edu> To: questions@freebsd.org Subject: Question: Sharing mail spool Sender: owner-questions@freebsd.org Precedence: bulk We have a "cluster" of FreeBSD machines and we want them to share the same mail spool. I understand that FreeBSD doesn't have NFS locking yet so I was trying to think of ways to get around the obvious problem. Currently, we have set up a mail hub and all "client" machines just forward incoming mail to it - hence there is only one machine's sendmail writing to the spool which is a disk local to that machine (i.e. that it can lock). So as far as delivery is concerned, I think we are OK. But I see a remaining problem where clients nfs mount the mail spool directory and mail user agents not running on the mail hub could have problems with the mail hub's sendmail process. Any suggestions? The only thing I could think of was to see if sendmail could support lock files (instead of flock/lockf) and only allow mail user agents that adhere to the same lock file protocol. Of course, it seems like there still could be problems because of nfs client side caching. How do you guys get around this problem? Thanks in advance. --- Jason Brazile jason@sunra.csci.unt.edu "People say I'm apathetic but I don't care" brazile@math.utexas.edu