Hello all,

In stress testing of JFS V1.0.1, I'm still experiencing the lockup/sleeping problem originally discussed on June 30th with V1.0.0. The discussion was 'Weird "lockup" problem with JFS v1.0.0 (kernel2.2.19)'

The scenario is: RH7.1, V2.4.6 SMP kernel, A PERL script to create ~5.1 million files, each of 8 bytes, in a ~25.2G volume, with 255 files in each directory, and each directory having 255 directories.

After an hour of processing, and ~4.05M files, the JFS/PERL processes went into 'uninterruptible sleep'. This has happened twice now (2 for 2), the first time it processed ~3 million files.
 

SER       PID %CPU %MEM   VSZ  RSS TTY      STAT START   TIME COMMAND
root         1  0.0  0.0  1368   52 ?        S    Jul22   0:05 init [3]
root         2  0.0  0.0     0    0 ?        SW   Jul22   0:00 [keventd]
root         3  0.0  0.0     0    0 ?        SW   Jul22   0:33 [kswapd]
root         4  0.0  0.0     0    0 ?        SW   Jul22   0:00 [kreclaimd]
root         5  0.0  0.0     0    0 ?        SW   Jul22   0:08 [bdflush]
root         6  0.0  0.0     0    0 ?        DW   Jul22   0:12 [kupdated]
root         7  0.0  0.0     0    0 ?        SW   Jul22   3:04 [jfsIO]
root         8  0.4  0.0     0    0 ?        DW   Jul22  18:10 [jfsCommit]
root         9  0.0  0.0     0    0 ?        DW   Jul22   0:44 [jfsSync]
root       382  0.0  0.0  1428  176 ?        S    Jul22   0:00 syslogd -m 0
root       387  0.0  0.0  1916    4 ?        S    Jul22   0:00 klogd -2
daemon     443  0.0  0.0  1400    4 ?        S    Jul22   0:00 /usr/sbin/atd
root       462  0.0  0.0  2240    4 ?        S    Jul22   0:00 xinetd -stayalive -reuse -pidfile /var/run/xinetd.pid
root       474  0.0  0.0  1552  112 ?        S    Jul22   0:00 crond
root       486  0.0  0.0  2968   68 ?        S    Jul22   0:00 smbd -D
root       491  0.0  0.0  2308  420 ?        S    Jul22   0:01 nmbd -D
root       514  0.0  0.0  2296    4 tty1     S    Jul22   0:00 login -- root
root       515  0.0  0.0  2296    4 tty2     S    Jul22   0:00 login -- root
root       516  0.0  0.0  1340    4 tty3     S    Jul22   0:00 /sbin/mingetty tty3
root       517  0.0  0.0  1340    4 tty4     S    Jul22   0:00 /sbin/mingetty tty4
root       518  0.0  0.0  1340    4 tty5     S    Jul22   0:00 /sbin/mingetty tty5
root       519  0.0  0.0  1340    4 tty6     S    Jul22   0:00 /sbin/mingetty tty6
root       522  0.0  0.0  2360    4 tty1     S    Jul22   0:00 -bash
root       561  0.0  0.1  2360  560 tty2     S    Jul22   0:00 -bash
root      2006 14.9  0.0  2708  296 tty1     D    19:23  10:10 /usr/bin/perl ./test2.pl
root      2028  0.0  0.1  2736  832 tty2     R    20:31   0:00 ps -auxw

Extras:

1) Stress testing of SGIs XFS had no problems. (Same machine, new JFS kernel, just reformatted volume with JFS)
2) The same problem of JFS and uninterruptible sleeping processes has also been found on a RH6.1 box, V2.2.19 kernel NO SMP, BUT with an Adaptec 2100 raid card, the problem is actually much more acute here as it will only create a few thousand files.

Hope someone can help.