Re: kern/50707: rm: fts_read: No such file or directory

[BERTRAND Joël <joel.bertrand%systella.fr@localhost>]

The following reply was made to PR kern/50707; it has been noted by GNATS.

From: =?UTF-8?Q?BERTRAND_Jo=c3=abl?= <joel.bertrand%systella.fr@localhost>
To: gnats-bugs%NetBSD.org@localhost, kern-bug-people%netbsd.org@localhost, gnats-admin%netbsd.org@localhost,
        netbsd-bugs%netbsd.org@localhost
Cc: 
Subject: Re: kern/50707: rm: fts_read: No such file or directory
Date: Thu, 23 Jun 2016 20:33:21 +0200

 BERTRAND Joël a écrit :
 > The following reply was made to PR kern/50707; it has been noted by GNATS.
 >
 > From: =?UTF-8?Q?BERTRAND_Jo=c3=abl?= <joel.bertrand%systella.fr@localhost>
 > To: gnats-bugs%NetBSD.org@localhost, kern-bug-people%netbsd.org@localhost, gnats-admin%netbsd.org@localhost,
 >          netbsd-bugs%netbsd.org@localhost
 > Cc:
 > Subject: Re: kern/50707: rm: fts_read: No such file or directory
 > Date: Mon, 6 Jun 2016 11:31:48 +0200
 >
 >   David Holland a écrit :
 >   > The following reply was made to PR kern/50707; it has been noted by GNATS.
 >   >
 >   > From: David Holland <dholland-bugs%netbsd.org@localhost>
 >   > To: gnats-bugs%NetBSD.org@localhost
 >   > Cc:
 >   > Subject: Re: kern/50707: rm: fts_read: No such file or directory
 >   > Date: Mon, 30 May 2016 03:40:41 +0000
 >   >
 >   >   On Tue, Apr 26, 2016 at 02:40:01PM +0000, Robert Elz wrote:
 >   >    > I have a patch you can apply that will allow switching to
 >   >    > read-only mode, but you do need to be very careful - the ability to
 >   >    > do that was removed because it does not always work correctly, to
 >   >    > switch to read-only mode takes some care to make sure the
 >   >    > filesystem is in a suitable state first (the system should simply
 >   >    > make it work, but no-one ever found out why it didn't.)
 >   >
 >   >   Well, that's not quite true. We found one at least one set of
 >   >   problems, which is that the logic for delayed mtime/atime updates was
 >   >   not pushing pending mtime/atime updates out. That I believe has been
 >   >   fixed.
 >   >
 >   >   There was also another set of problems that came to light when someone
 >   >   discovered that -o async wasn't ever writing anything out at all
 >   >   unless forced to by buffer cache capacity, but I don't remember what
 >   >   the resolution of that was, if any.
 >   >
 >   >   --
 >   >   David A. Holland
 >   >   dholland%netbsd.org@localhost
 >   >
 >   >
 >
 >   	Hello,
 >
 >   	I have applied last patches. Now, my server runs 7.0.1. Last night,
 >   after upgrading pkgsrc from CVS tree, I have seen "to many open files"
 >   and fs was currupted again. Some directories do not contains '.' or '..'.
 >
 >   legendre$ uname -a
 >   NetBSD legendre.systella.fr 7.0.1 NetBSD 7.0.1 (GENERIC.201605221355Z) amd64
 >   legendre$ mount
 >   /dev/raid0a on / type ffs (asynchronous, log, NFS exported, local)
 >   /dev/raid0f on /var type ffs (asynchronous, log, local)
 >   /dev/raid0e on /usr type ffs (local)
 >   /dev/raid0g on /home type ffs (asynchronous, log, NFS exported, local)
 >   kernfs on /kern type kernfs (local)
 >   ptyfs on /dev/pts type ptyfs (local)
 >   procfs on /proc type procfs (local)
 >   tmpfs on /var/shm type tmpfs (local)
 >   legendre$
 >
 >   	If I remove async from /etc/fstab, I obtain a fts_read error. With
 >   aync, error differs but result is the same. I have to reboot with fs in
 >   ro mode to run fsck.
 >
 >   	Best regards,
 >
 >   	JKB
 >
 >
 
 	Hello,
 
 	Any news ? I have seen some other disk corruption since my last 
 message... I can do some tests but I don't know VFS subsystem enough to 
 debug.
 
 	Best regards,
 
 	JKB