Date: Wed, 19 Feb 2014 13:19:48 +0200 From: Andriy Gapon <avg@FreeBSD.org> To: Kai Kockro <k.kockro@myphotobook.de>, freebsd-fs@FreeBSD.org Subject: Re: ZFS directory not accessable Message-ID: <53049354.3040700@FreeBSD.org> In-Reply-To: <530490A8.70209@myphotobook.de> References: <530487FD.10209@myphotobook.de> <53048D43.1010201@FreeBSD.org> <530490A8.70209@myphotobook.de>
next in thread | previous in thread | raw e-mail | index | archive | help
on 19/02/2014 13:08 Kai Kockro said the following: > Hello Andriy, > > > ps -p 4883 > PID TT STAT TIME COMMAND > 4883 1- D+ 0:00,00 ls -G -f > > procstat -kk 4883 > PID TID COMM TDNAME KSTACK > 4883 100558 ls - mi_switch+0x194 > sleepq_wait+0x42 _sx_slock_hard+0x3b0 _sx_slock+0x3d > zap_get_leaf_byblk+0xbd zap_deref_leaf+0x68 fzap_cursor_retrieve+0xe7 > zap_cursor_retrieve+0x155 zfs_freebsd_readdir+0x2d8 VOP_READDIR_APV+0x78 > kern_getdirentries+0x212 sys_getdirentries+0x23 amd64_syscall+0x5ea > Xfast_syscall+0xf7 Can you please also obtain the following info? 1. Find out an inode number of the affected directory and its underlying ZFS filesystem, e.g. by using stat(1). Then execute `zdb -ddddddd <dataset> <inum>`. 2. Run 'kgdb', issue 'tid 100558' command and 'bt' command. Examine the stack trace and try to get as much information as possible from frames with _sx_slock and zap_get_leaf_byblk calls. For pointers please print objects that they point to (e.g. 'print *ptr'). In particular I am interested in *sx, *zap, *l. > ps -p 5236 > PID TT STAT TIME COMMAND > 5236 3- D+ 0:00,00 ls -G -la > > procstat -kk 5236 > PID TID COMM TDNAME KSTACK > 5236 101104 ls - mi_switch+0x194 > sleepq_wait+0x42 __lockmgr_args+0x9a0 vop_stdlock+0x39 > VOP_LOCK1_APV+0x84 _vn_lock+0x47 vacl_get_acl+0x55 > sys___acl_get_link+0x10d amd64_syscall+0x5ea Xfast_syscall+0xf7 > > > > Best, > Kai > > Am 19.02.2014 11:53, schrieb Andriy Gapon: >> on 19/02/2014 12:31 Kai Kockro said the following: >>> Hello, >>> >>> i have an 9.2-STABLE r262153 installation on a T420s laptop. Its a ZROOT >>> on a INTEL SSDSA2BW160G3L drive. >>> >>> I have a folder ( skype profil ), which is not accessable anymore. I can >>> do ls, du and so on, but the states are in D+ after this. >>> >>> STRG + T shows: >>> >>> [user@host 11:01am] ~/.SkypeOLD/user_profil/>ls -f >>> load: 0.13 cmd: ls 4883 [] 2.76r 0.00u 0.00s 0% 2124k >>> load: 0.13 cmd: ls 4883 [] 3.50r 0.00u 0.00s 0% 2124k >>> load: 0.13 cmd: ls 4883 [] 3.92r 0.00u 0.00s 0% 2124k >>> load: 0.20 cmd: ls 4883 [] 4.13r 0.00u 0.00s 0% 2124k >>> load: 0.20 cmd: ls 4883 [] 4.31r 0.00u 0.00s 0% 2124k >>> >>> What can i do? Any debug infos needed? >> You can start with getting procstat -kk output for a stuck process. >> > -- Andriy Gapon
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?53049354.3040700>