Hi Sage and ceph-developers,
I reproduced the error again and captured file descriptors and log information.
I've quoted them below. Here is a brief summary:
I have ceph 0.48 running on a single mulitprocessor node. I have two clients
that have mounted the cephfs over infiniband (same error happens under GB
though). I start an parallel program with mpi on the two clients. It starts 7
processes on each client. One process (id 1728 below) opens 43 files at start
up and the remaining 13 processes open up two files each. I've quotes the
output of the system file descriptors below. The processes run fine with I/O
going to files for about 25s when the process with 43 files locks up. At 33s
the remaining processes are locked and accesses to the mounted ceph fs also
lock.
I hope these details along with the mds log quotes in my last message help
isolate this problem.
Is it recommended to upgrade from 0.48, or are there backports to 0.48 I should
incorporate?
thanks,
-john
Here's the debug feedback you asked for:
# cat /sys/kernel/debug/ceph/*/mdsc
1028 mds0 getattr #100000003ee
The main process with 43 files:
#+begin_quote stat /proc/1728/fd/*
[root@eaps-80-35e ~]# stat /proc/1728/fd/*
File: `/proc/1728/fd/0' -> `pipe:[29056]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25862 Links: 1
Access: (0500/lr-x------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/1' -> `pipe:[29057]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25863 Links: 1
Access: (0300/l-wx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/10' -> `anon_inode:[infinibandevent]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25872 Links: 1
Access: (0500/lr-x------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/11' -> `pipe:[29058]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25873 Links: 1
Access: (0300/l-wx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/12' -> `/dev/infiniband/uverbs0'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25874 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/13' -> `anon_inode:[infinibandevent]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25875 Links: 1
Access: (0500/lr-x------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/14' -> `/'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25876 Links: 1
Access: (0500/lr-x------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/15' ->
`/dev/shm/ib_shmem-kvs_2025_0-eaps-80-35e.acesgrid.org-50329.tmp (deleted)'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25877 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/16' ->
`/dev/shm/ib_pool-kvs_2025_0-eaps-80-35e.acesgrid.org-50329.tmp (deleted)'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25878 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/17' ->
`/dev/shm/ib_shmem_coll-kvs_2025_0-eaps-80-35e.acesgrid.org-50329.tmp (deleted)'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25879 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/18' -> `/mnt/ceph/runtemplate/STDERR.0000'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25880 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/19' -> `/mnt/ceph/runtemplate/STDOUT.0000'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25881 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/2' -> `pipe:[29058]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25864 Links: 1
Access: (0300/l-wx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/20' -> `/mnt/ceph/runtemplate/stats/PO4.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25882 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/21' -> `/mnt/ceph/runtemplate/stats/NO3.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25883 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/22' -> `/mnt/ceph/runtemplate/stats/FeT.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25884 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/23' -> `/mnt/ceph/runtemplate/stats/Si.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25885 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/24' -> `/mnt/ceph/runtemplate/stats/DOP.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25886 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/25' -> `/mnt/ceph/runtemplate/stats/DON.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25887 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/26' -> `/mnt/ceph/runtemplate/stats/DOFe.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25888 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/27' -> `/mnt/ceph/runtemplate/stats/POP.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25889 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/28' -> `/mnt/ceph/runtemplate/stats/PON.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25890 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/29' -> `/mnt/ceph/runtemplate/stats/POFe.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25891 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/3' -> `socket:[22813]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25865 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/30' -> `/mnt/ceph/runtemplate/stats/POSi.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25892 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/31' -> `/mnt/ceph/runtemplate/stats/NH4.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25893 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/32' -> `/mnt/ceph/runtemplate/stats/NO2.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25894 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/33' -> `/mnt/ceph/runtemplate/stats/DIC.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25895 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/34' -> `/mnt/ceph/runtemplate/stats/DOC.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25896 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/35' -> `/mnt/ceph/runtemplate/stats/POC.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25897 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/36' -> `/mnt/ceph/runtemplate/stats/PIC.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25898 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/37' -> `/mnt/ceph/runtemplate/stats/ALK.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25899 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/38' -> `/mnt/ceph/runtemplate/stats/O2.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25900 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/39' -> `/mnt/ceph/runtemplate/stats/ZooP.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25901 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/4' -> `/dev/infiniband/uverbs0'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25866 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/40' -> `/mnt/ceph/runtemplate/stats/ZooN.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25902 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/41' -> `/mnt/ceph/runtemplate/stats/ZooFe.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25903 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/42' -> `/mnt/ceph/runtemplate/stats/ZooSi.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25904 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/43' -> `/mnt/ceph/runtemplate/stats/Phy.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25905 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/44' -> `/mnt/ceph/runtemplate/stats/ZooC.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25906 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/45' -> `/mnt/ceph/runtemplate/stats/PP.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25907 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/46' -> `/mnt/ceph/runtemplate/stats/PAR.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25908 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/47' ->
`/mnt/ceph/runtemplate/stats/Diver1.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25909 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/48' ->
`/mnt/ceph/runtemplate/stats/Diver2.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25910 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/49' ->
`/mnt/ceph/runtemplate/stats/Diver3.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25911 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/5' -> `socket:[29055]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25867 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/50' ->
`/mnt/ceph/runtemplate/stats/Diver4.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25912 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/51' ->
`/mnt/ceph/runtemplate/stats/DICTFLX.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25913 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/52' ->
`/mnt/ceph/runtemplate/stats/DICCFLX.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25914 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/53' ->
`/mnt/ceph/runtemplate/stats/DICOFLX.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25915 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/54' ->
`/mnt/ceph/runtemplate/stats/DICPCO2.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25916 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/55' ->
`/mnt/ceph/runtemplate/stats/DICPHAV.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25917 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/56' ->
`/mnt/ceph/runtemplate/stats/ChlCloer.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25918 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/57' ->
`/mnt/ceph/runtemplate/stats/ChlGeide.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25919 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/58' ->
`/mnt/ceph/runtemplate/stats/ChlDoney.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25920 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/59' ->
`/mnt/ceph/runtemplate/stats/Shannon.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25921 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/6' -> `pipe:[29056]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25868 Links: 1
Access: (0500/lr-x------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/60' ->
`/mnt/ceph/runtemplate/stats/Simpson.0000000000.txt'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25922 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/7' -> `anon_inode:[infinibandevent]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25869 Links: 1
Access: (0500/lr-x------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/8' -> `/dev/infiniband/uverbs0'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25870 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
File: `/proc/1728/fd/9' -> `pipe:[29057]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25871 Links: 1
Access: (0300/l-wx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:32:40.468849356 -0400
Modify: 2012-09-04 12:32:40.468849356 -0400
Change: 2012-09-04 12:32:40.468849356 -0400
Birth: -
#+end_quote
An example of the other 13 processes:
#+begin_quote stat /proc/1729/fd/*
[root@eaps-80-35e ~]# stat /proc/1729/fd/*
File: `/proc/1729/fd/0' -> `/dev/infiniband/uverbs0'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25840 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/1' -> `pipe:[29061]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25841 Links: 1
Access: (0300/l-wx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/10' -> `pipe:[29058]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25850 Links: 1
Access: (0500/lr-x------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/11' -> `pipe:[29061]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25851 Links: 1
Access: (0300/l-wx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/12' -> `anon_inode:[infinibandevent]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25852 Links: 1
Access: (0500/lr-x------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/13' -> `pipe:[29062]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25853 Links: 1
Access: (0300/l-wx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/14' -> `/'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25854 Links: 1
Access: (0500/lr-x------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/15' ->
`/dev/shm/ib_shmem-kvs_2025_0-eaps-80-35e.acesgrid.org-50329.tmp (deleted)'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25855 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/16' ->
`/dev/shm/ib_pool-kvs_2025_0-eaps-80-35e.acesgrid.org-50329.tmp (deleted)'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25856 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/17' ->
`/dev/shm/ib_shmem_coll-kvs_2025_0-eaps-80-35e.acesgrid.org-50329.tmp (deleted)'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25857 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/18' -> `/mnt/ceph/runtemplate/STDERR.0002'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25858 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/19' -> `/mnt/ceph/runtemplate/STDOUT.0002'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25859 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/2' -> `pipe:[29062]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25842 Links: 1
Access: (0300/l-wx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/3' -> `socket:[22813]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25843 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/4' -> `anon_inode:[infinibandevent]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25844 Links: 1
Access: (0500/lr-x------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/5' -> `/dev/infiniband/uverbs0'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25845 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/6' -> `socket:[29060]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25846 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/7' -> `anon_inode:[infinibandevent]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25847 Links: 1
Access: (0500/lr-x------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/8' -> `pipe:[29057]'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25848 Links: 1
Access: (0500/lr-x------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
File: `/proc/1729/fd/9' -> `/dev/infiniband/uverbs0'
Size: 64 Blocks: 0 IO Block: 1024 symbolic link
Device: 3h/3d Inode: 25849 Links: 1
Access: (0700/lrwx------) Uid: (50329/jcwright) Gid: ( 101/ mit)
Access: 2012-09-04 12:31:32.957857330 -0400
Modify: 2012-09-04 12:31:32.957857330 -0400
Change: 2012-09-04 12:31:32.957857330 -0400
Birth: -
#+end_quote
#
On Aug 31, 2012, at 4:03 PM, Sage Weil wrote:
> Hi John,
>
> On Fri, 31 Aug 2012, John C. Wright wrote:
>> An update,
>> While looking into how to switch over to a different network on my ceph
>> cluster - another question altogether, I discovered during my upgrade from
>> 0.47.2 to 0.48 on the three nodes somehow my 'git checkout 0.48' wasn't done
>> correctly on two nodes and wound up reinstalling 0.47.2 on those so somehow
>> had a heterogenous cluster running 0.47.2 on the osds and 0.48 running on
>> the mds (all three running monitors).
>>
>> So wiped out and started fresh with 0.48 and still got the error, but with
>> more info this time.
>> This MPI code is running on 14 processes on two client nodes. Each process
>> writes to its own 'stdout' file and some other data files are created by
>> process '0'. The program starts up and creates its initial files and begins
>> to write to its stdout files. Normally this proceeds with writes to the
>> stdout at periodic intervals during the run, but on the ceph volume, this
>> freezes up after about one minute.
>>
>> Symptoms: initially can still access ceph volume from other clients. Listing
>> the working directory of the mpi code is very slow and soon is unresponsive
>> but can still list other directories on the ceph volume. After another
>> minute or so, ceph clients can no longer access the volume at all without
>> locking up in a trace that ends in a 'fastpath' kernel call. If I CTRL-C out
>> of the mpirun call within the first minute, everything recovers, but waiting
>> longer than that requires a reboot of mounted nodes and a restart of ceph to
>> clear things up.
>>
>> Below are relevant (I hope) process traces from dmesg and ceph logs. Any
>> help on diagnosing this would be greatly appreciated. We're hoping to use
>> ceph as a parallel file system on a scientific workload beowulf cluster,
>> initially with a buyer-beware policy an for only transient reproducible data
>> and more general usage as ceph gets stable and reaches the 1.0 milestone.
>
> It looks a bit like this is a problem on the MDS side of things, since you
> have both a hung request and a writer waiting for caps. Can you generate
> an MDS log that goes with this workload? With
>
> debug ms = 1
> debug mds = 20
>
> in the [mds] section of your config? Also, once it hangs, it would be
> helpful to see what the hung request is (cat
> /sys/kernel/debug/ceph/*/mdsc) and the inode number for the hung writer
> (stat /proc/$pid/fd/NNN). Hopefully the stat won't hang.. but if it does,
> hopefully you can identify which file ino or filename it is some other
> way.
>
> Thanks!
> sage
>
>
>>
>> Thanks very much.
>>
>> -john wright
*Removed old quoted part of thread*
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html