From owner-freebsd-current Mon Oct 12 21:05:23 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id VAA20347 for freebsd-current-outgoing; Mon, 12 Oct 1998 21:05:23 -0700 (PDT) (envelope-from owner-freebsd-current@FreeBSD.ORG) Received: from math.berkeley.edu (math.Berkeley.EDU [128.32.183.94]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id VAA20342 for ; Mon, 12 Oct 1998 21:05:22 -0700 (PDT) (envelope-from dan@math.berkeley.edu) Received: (from dan@localhost) by math.berkeley.edu (8.8.7/8.8.7) id VAA18347; Mon, 12 Oct 1998 21:05:08 -0700 (PDT) Date: Mon, 12 Oct 1998 21:05:08 -0700 (PDT) From: dan@math.berkeley.edu (Dan Strick) Message-Id: <199810130405.VAA18347@math.berkeley.edu> To: current@FreeBSD.ORG Subject: -current NFS problem Cc: dan@math.berkeley.edu Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG I haven't reported this before because it is kind of hard to be very specific about the cause and because I am under the impression that people are aware of NFS problems in a general way. I am reporting this now necause I believe this to be very important and I haven't heard any discussion of this specific problem. There is some sort of NFS problem that causes processes to wedge in a high priority (i.e. non-interruptable) wait state. I encounter this problem whenever I use the ncftp2 program on a 3.0-SNAP or BETA system to make copy of a FreeBSD release in a remotely mounted file system. It doesn't happen immediately, but it always happens before the copy is complete. The ncftp2 program just stops reporting any progress and a ps shows it to be in a "disk" wait state. The ncftp2 program is normally not killable and attempts (on the 3.0-SNAP or BETA system) to examine the directory it was copying into also wedge in a "disk" wait state. Eventually the entire 3.0-SNAP or BETA system will wedge. I suspect the NFS problem, which initially is localized to one file or directory, cascades through the automounter. The NFS server, a sun workstation running Solaris 2.5, does not appear to be affected. The directory that is wedged on the 3.0-SNAP or BETA FreeBSD system is visible and behaves normally on the server even after the FreeBSD NFS client system locks up. The only (temporary) fix is to reboot the FreeBSD 3.0-SNAP or BETA system and restart the ncftp2 program where it left off. Eventually it will finish. I normally avoid the problem entirely by doing my FreeBSD release copies on a 2.2.7-RELEASE system which never hangs. Note: after a NFS wedge, I never find a partially written file. The wedge always seems to happen "between" files. Dan Strick To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message