On Mon, 31 Mar 2014, sf...@users.sourceforge.net wrote: > According to this netconsole.log, > - ln:1567 enters aufs, and aufs tries copying-up a file by calling > kworker/5:1:73. > - kworker/5:1:73 tries opening a file on afs and gets stuck. > - rs:main Q:Reg:1571 is also running and issues access(2), enters aufs > and gets stuck. > > What is the process "rs:main Q:Reg" whose pid is 1571? > And which file is it going to access(2)?
In this case, Debian's 'savelog' command tried to rename a logfile, which existed only in the ro-Branch (in AFS). I tried to create a minimalistic example: http://users.minet.uni-jena.de/~erik/aufs/mv.log In the last line, the 'mv' command stalls. The corresponding kernel messages are in http://users.minet.uni-jena.de/~erik/aufs/netconsole.log > How do you mount aufs and openafs? Similar to the old nfsroot-client this is an OpenAFS-rooted client, the directories are bind mounts, p. e.: mount -v -o bind -r afs/mirz/linuxclients/linuxpool-jessie64/var /var … The superjacent AUFS-mount for this directory is (/run/shm/aufs/var ist just an example for an empty directory, usually it is the local disk): mount -t aufs -o br:/run/shm/aufs/var=rw:/afs/mirz/linuxclients/linuxpool-jessie64/var=ro none /var > If you make something like recursive mount, ie. mount aufs under > openafs, then your system may not work since it looks like that openafs > has a global mutex (based upon just my current guess). These are bad news, because it worked for long years. > ---------------------------------------- > - /proc/mounts (instead of the output of mount(8)) > - /sys/module/aufs/* > - /sys/fs/aufs/* (if you have them) > - /debug/aufs/* (if you have them) > - linux kernel version > if your kernel is not plain, for example modified by distributor, > the url where i can download its source is necessary too. > - aufs version which was printed at loading the module or booting the > system, instead of the date you downloaded. > - configuration (define/undefine CONFIG_AUFS_xxx) > - kernel configuration or /proc/config.gz (if you have it) > ---------------------------------------- You find all these files (including 600 MB unpacked kernel source code) and outputs in http://users.minet.uni-jena.de/~erik/aufs/ The output of the branches in /sys/fs/aufs/ stalls, when the above-mentioned 'mv' commando is entered. >In order to make sure this, if you enable CONFIG_MAGIC_SYSRQ, try >SysRq+d and SysRq+w after the deadlock. >If the global mutex in openafs is related to this problem, then kernel >prints "afs_global_lock" somewhere in the message. Unfortunately there seems to be a bug in Debians actual OpenAFS release, which is why OpenAFS did not compile with CONFIG_LOCKDEP (bug report is issued). In netconsole.log ist the output of SysRq+w. The afs_global_lock is not in there. Thank you, Erik ------------------------------------------------------------------------------
