Date: Sun, 22 Feb 2009 13:00:52 +0200 From: Kostik Belousov <kostikbel@gmail.com> To: Carl <k0802647@telus.net> Cc: freebsd-fs@freebsd.org Subject: Re: UFS2 and/or sparse file bug causing copy process to land in 'D'' state? Message-ID: <20090222110052.GH41617@deviant.kiev.zoral.com.ua> In-Reply-To: <49A10626.8060705@telus.net> References: <49A10626.8060705@telus.net>
next in thread | previous in thread | raw e-mail | index | archive | help
--sjel/IY1pyoUgMMX Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Sun, Feb 22, 2009 at 12:00:38AM -0800, Carl wrote: > I've come across what I'm thinking may be a bug in the context of=20 > FreeBSD 7.0 with a pair of gmirrored drives and gjournaled partitions=20 > when copying a large number of files into a file-backed memory device. >=20 > The consequence of this problem is that a process enters the 'D' state=20 > (process in disk) indefinitely, cannot be killed, and the system cannot= =20 > be shutdown. The only solution is to cold reboot the system, which is a= =20 > really big problem for remote systems. This is happening to me=20 > intermittently with the standard tar-tar pipeline form of copying, but=20 > has happened with the rsync 3.0.4 port as well. >=20 > I would appreciate it if some of you would see if you can repeat this=20 > problem. Here is a sequence of tcsh shell commands which manifest the=20 > problem (on occasion but not every time), which I will refer to as the=20 > "truncate sequence" (depends on fully populated /usr/src tree as data set= ): >=20 > # truncate -s 671088640 target > # mdconfig -f target -S 512 -y 255 -x 63 -u 7 > # bsdlabel -w /dev/md7 auto > # newfs -O2 -m 0 -o space /dev/md7a > # mount /dev/md7a /media > # tar -cvf - -C /usr/src . | tar -xvpof - -C /media > # umount /media ; mdconfig -d -u 7 ; rm target >=20 > An alternate version has yet to fail for me and involves replacing the=20 > first line with this one: >=20 > # dd if=3D/dev/zero of=3Dtarget bs=3D1M count=3D640 >=20 > I'll call that the "dd sequence". Here is an ordered series of tests I=20 > just completed: >=20 > a) Repeated truncate sequence 7 times - 1st, 5th, and 7th failed. > b) Repeated dd sequence 7 times - no failures. > c) Repeated truncate sequence 6 time - no failures. > d) Used following sequence to ensure all disk caches flushed: >=20 > # dd if=3D/dev/random of=3Dtarget bs=3D1M count=3D4096 > # dd if=3Dtarget of=3D/dev/null bs=3D1M > # rm target >=20 > e) Repeated truncate sequence 4 times - no failures. > f) Performed orderly reboot. > g) Repeated truncate sequence 2 times - 2nd failed. > h) Performed orderly reboot. > i) Repeated dd sequence 7 times - no failures. >=20 > All failures involve the second tar in the pipeline hanging in the 'D'=20 > state. In each case I do a cold reboot before proceeding with the next te= st. >=20 > It's tempting to speculate that a bug exists in code related to handling= =20 > sparse files specifically, but perhaps it just raises the probability of= =20 > tripping a bug that would eventually manifest in the dd sequence as=20 > well. OTOH, I don't know how to rule out a physical disk or disk=20 > firmware problem. >=20 > This problem has occurred with different data sets and different sized=20 > memory disks, but only with the source and destination filesystems being= =20 > UFS2. I have done similar sequences with EXT2 and FAT16 destinations=20 > with no failures thus far, but the memory disks and data sets were=20 > smaller so it's conceivable that probability worked against me. >=20 > I should note that the drives are Seagate ST31000340AS Barracudas, but=20 > both drives have been upgraded to firmware version SD1A and are=20 > therefore supposedly free of the infamous little horror Seagate=20 > inflicted on so many of us. smartctl tells me that both disks still have= =20 > a raw value of 0 for Reallocated_Sector_Ct and both pass the "short"=20 > self test. Please, see http://www.freebsd.org/doc/en_US.ISO8859-1/books/developers-handbook/kernel= debug-deadlocks.html for instructions on how to gather the required information to diagnose the issue. --sjel/IY1pyoUgMMX Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (FreeBSD) iEYEARECAAYFAkmhMGMACgkQC3+MBN1Mb4iqVACePL6IH3cjmuxS/fBbA662oa6o 1oMAnjiIFXx8lUDtxWyr9TdEWDfnF5xf =7grU -----END PGP SIGNATURE----- --sjel/IY1pyoUgMMX--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20090222110052.GH41617>