From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ryusuke Konishi Subject: Re: nilfs2 weird issue - snapshots are gone, cleanerd not running Date: Tue, 10 Jul 2012 17:51:31 +0900 (JST) Message-ID: <20120710.175131.21311203.konishi.ryusuke@lab.ntt.co.jp> References: <51D5FCEA-7103-4D4A-BADA-99A9780D9B68@dubeyko.com> <20120710.105315.33988123.konishi.ryusuke@lab.ntt.co.jp> <1341904734.1980.17.camel@slavad-ubuntu-11> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <1341904734.1980.17.camel@slavad-ubuntu-11> Sender: linux-nilfs-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-ID: Content-Type: Text/Plain; charset="iso-2022-jp" To: Vyacheslav Dubeyko Cc: szarpaj-TbOm9Ca2r9GrDJvtcaxF/A@public.gmane.org, linux-nilfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org Hi, On Tue, 10 Jul 2012 11:18:54 +0400, Vyacheslav Dubeyko wrote: > Hi Ryusuke, > > Unfortunately, my kernel is compiled with enabled CONFIG_SHMEM and > CONFIG_TMPFS options. Moreover, I can see that /dev/shm mounted with > tmpfs type: > > linux-2.6.34-gentoo-r6 # mount > /dev/sda3 on / type xfs (rw,noatime) > proc on /proc type proc (rw,nosuid,nodev,noexec,relatime) > sysfs on /sys type sysfs (rw,nosuid,nodev,noexec,relatime) > udev on /dev type tmpfs (rw,nosuid,relatime,size=10240k,mode=755) > devpts on /dev/pts type devpts > (rw,nosuid,noexec,relatime,gid=5,mode=620) > shm on /dev/shm type tmpfs (rw,noexec,nosuid,nodev) > usbfs on /proc/bus/usb type usbfs > (rw,noexec,nosuid,devmode=0664,devgid=85) > /dev/sdb2 on /mnt/nilfs2 type nilfs2 (rw,gcpid=6331) > > And it is possible to see it from strace output also (tmpfs magic - > 0x1021994): > > statfs("/dev/shm", {f_type=0x1021994, f_bsize=4096, f_blocks=64251, > f_bfree=64250, f_bavail=64250, f_files=64251, f_ffree=64249, f_fsid={0, > 0}, f_namelen=255, f_frsize=4096}) = 0 > > However, I have not last stable kernel on its machine > (2.6.34-gentoo-r6). Maybe, it can make some side effects. > > I am going to continue investigation of the issue. I attach to the > e-mail full strace output (10-07-2012-nilfs-cleanerd-issue.txt). Ok, this looks a different problem. How is CONFIG_POSIX_MQEUEU ? Is it enabled in your kernel ? Regards, Ryusuke Konishi > With the best regards, > Vyacheslav Dubeyko. > > On Tue, 2012-07-10 at 10:53 +0900, Ryusuke Konishi wrote: > > Hi Vyacheslav, > > On Mon, 9 Jul 2012 22:55:40 +0400, Vyacheslav Dubeyko wrote: > > > Hi Piotr, > > > > > > You are right. I can reproduce this issue very simply. The nilfs_cleanerd doesn't started during mount really. > > > > > > I can detect some suspicious output of strace during mount and next trying to start of nilfs_cleanerd: > > > > > > .... > > > set_tid_address(0xb76a0768) = 21036 > > > set_robust_list(0xb76a0770, 0xc) = 0 > > > futex(0xbfdd4f90, FUTEX_WAKE_PRIVATE, 1) = 0 > > > futex(0xbfdd4f90, FUTEX_WAIT_BITSET_PRIVATE|FUTEX_CLOCK_REALTIME, 1, NULL, bfdd4fa0) = -1 EAGAIN (Resource temporarily unavailable) > > > > > > .... > > > mq_open("nilfs-cleanerq-2066", O_RDONLY|O_CREAT, 0600, {mq_maxmsg=6, mq_msgsize=4096}) = -1 ENOSYS (Function not implemented) > > > > > > But maybe it is not reason of the problem. It needs to investigate the issue more deeply. > > > > Your problem looks that of FAQ #8 on http://www.nilfs.org/en/faq.html > > > > > 8. cleanerd (or chcp/mkcp command) fails with an error: ``cannot open > > > nilfs on /dev/xxx: Function not implemented''. > > > > > > Confirm whether tmpfs (former shm fs) is mounted on /dev/shm. POSIX > > > semaphores do not work if the filesystem on /dev/shm is wrong, > > > which causes the above failure. > > > > > > Some systems are using ramfs instead of tmpfs. You may need to > > > change kernel configuration and rebuild kernel to enable tmpfs. > > > > Please confirm if tmpfs is mounted on /dev/shm. > > > > The same issue is reported on the following thread: > > > > http://marc.info/?t=133190016900003&r=1&w=2 > > > > > > Regards, > > Ryusuke Konishi > > > > > Thanks, > > > Vyacheslav Dubeyko. > > > > > > On Jul 9, 2012, at 8:56 PM, Piotr Szymaniak wrote: > > > > > > > On Mon, Jul 09, 2012 at 01:28:32PM +0400, Vyacheslav Dubeyko wrote: > > > >> Hi Piotr, > > > >> > > > >> Does system journals on your machines contain any interested details > > > >> about reported issue? Could you try to extract some error or warning > > > >> messages from system journal? > > > > > > > > (resend as I replied only to Vyacheslav) > > > > > > > > If by journals you mean logs then no. I'm only able to find some like > > > > this: > > > > Jul 3 10:32:45 wloczykij nilfs_cleanerd[1434]: resume (clean check) > > > > Jul 3 10:41:37 wloczykij nilfs_cleanerd[1434]: pause (clean check) > > > > > > > > That's all about nilfs in the last week and current log has only manual > > > > runs related to those operation described before. > > > > > > > > Piotr Szymaniak. > > > > > > > > > > > >> On Mon, 2012-07-09 at 09:33 +0200, Piotr Szymaniak wrote: > > > >>> Hi. > > > >>> > > > >>> I've upgraded nilfs-utils (running Gentoo) on 29 july. Today I ran out > > > >>> of space on my / and found that nilfs_cleanerd isn't working. When I > > > >>> start it from the command line it exits instantly. Also, all previous > > > >>> checkpoints on / (also on two other mountpoints on different machine) > > > >>> are gone. > > > >>> > > > >>> What I did? Downgraded nilfs-utils to 2.1.1, remounted mountpoints. On > > > >>> the second machine it's runnig fine (cleaned _all_ checkpoints), on the > > > >>> first one with disk space issue it exits just like 2.1.3. > > > >>> > > > >>> Here are some fs details. Machine with disk space issues, rootfs: > > > >>> CNO DATE TIME MODE FLG NBLKINC ICNT > > > >>> 147688 2012-07-09 08:38:14 cp - 11075 242915 > > > >>> 147689 2012-07-09 08:38:14 cp - 60 242895 > > > >>> (…) > > > >>> 148999 2012-07-09 09:13:46 cp - 60 242888 > > > >>> 149000 2012-07-09 09:19:45 cp - 44 242888 > > > >>> > > > >>> Filesystem Size Used Avail Use% Mounted on > > > >>> rootfs 24G 13G 11G 56% / > > > >>> > > > >>> mount shows: > > > >>> /dev/sda2 on / type nilfs2 (rw,noatime,nodiratime,gcpid=15356) > > > >>> > > > >>> There's no nilfs_cleanerd with pid 15356. > > > >>> > > > >>> > > > >>> Second machine rootfs: > > > >>> CNO DATE TIME MODE FLG NBLKINC ICNT > > > >>> 92246 2012-07-09 08:16:58 cp - 118 44669 > > > >>> (…) > > > >>> 92439 2012-07-09 09:19:14 cp - 29 44668 > > > >>> 92440 2012-07-09 09:19:46 cp - 33 44668 > > > >>> > > > >>> Filesystem Size Used Avail Use% Mounted on > > > >>> rootfs 3.7G 888M 2.6G 26% / > > > >>> > > > >>> (it should be around 3G used) > > > >>> > > > >>> Second machine second mountpoint: > > > >>> CNO DATE TIME MODE FLG NBLKINC ICNT > > > >>> 1496 2012-07-09 03:31:23 cp - 8837 132766 > > > >>> 1497 2012-07-09 03:31:26 cp - 468 132766 > > > >>> 1498 2012-07-09 03:41:27 cp - 1474 132765 > > > >>> > > > >>> (this fs should containt *all* 1498 checkpoints) > > > >>> > > > >>> Filesystem Size Used Avail Use% Mounted on > > > >>> /dev/dm-2 117G 58G 54G 76% /mnt/home_backup > > > >>> > > > >>> (in this one it should be around 100G of used space) > > > >>> > > > >>> mount: > > > >>> /dev/dm-2 on /mnt/home_backup type nilfs2 (rw,gcpid=13135) > > > >>> /dev/sda3 on / type nilfs2 (rw,noatime,nodiratime,gcpid=1363) > > > >>> > > > >>> Both cleaners running (the second mountpoint - /mnt/home_backup - is under > > > >>> heavy load and I suppose it will end with around 20G used space). > > > >>> > > > >>> Where to go from this point? How to debug nilfs_cleanerd issue? > > > >>> > > > >>> > > > >>> Piotr Szymaniak. > > > >> > > > >> > > > >> -- > > > >> To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in > > > >> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org > > > >> More majordomo info at http://vger.kernel.org/majordomo-info.html > > > > > > > > -- > > > > Marriage is like a coffin and each kid is like another nail. > > > > -- Homer Simpson > > > > > > -- > > > To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in > > > the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org > > > More majordomo info at http://vger.kernel.org/majordomo-info.html > -- To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html