gzip sleeps until 'strace'ing the process

From: Henning Fehrmann
Date: Wed Sep 16 2009 - 10:52:16 EST


Hello,

If you have some hints, please put me in the CC.

We launched a job which forks a
gzip -cd
which is basically zcat.
This process is accessing a file within a NFS mounted directory:

# ps aux|grep gzip
user 32078 0.0 0.0 4140 552 pts/0 S+ 15:06 0:00 gzip -cd -f file_in_NFS_mounted_directory

The file is properly ziped.

The process is in the sleep state - also the file can be accessed.

If I 'strace' it the process finishes successfully.

# strace -p 32078
Process 32078 attached - interrupt to quit
write(1, "e-010 5.5565\n121.951418712 2.2249"..., 24578) = 24578
write(1, "3 -0.3036801 -3.1699e-011 5.30998"..., 32768) = 32768
write(1, "2015e-010 5.5565\n121.9529017461 2"..., 32768) = 32768
write(1, "\n121.9543512274 0.166872 -0.13840"..., 32768) = 32768
write(1, "-011 5.5565\n121.9558745249 5.2519"..., 32768) = 32768
write(1, "08331213 -4.2015e-010 5.80302\n121"..., 32768) = 32768
write(1, "1.9587130924 4.603975 -0.2485881 "..., 32768) = 32768
read(8, "\361\234a\325\305\3161\373x\307\324\312\240+\307\354\211Y\357cH\234\360\362\315\235NV\310O\\\212\236"..., 32768) = 14630
read(8, ""..., 18138) = 0
write(1, "600686259 5.408658 -0.5240481 -1."..., 32768) = 32768
write(1, "5.707316 -0.3587721 -4.2015e-010 "..., 32768) = 32768
write(1, "1.9627260083 5.766155 -0.3587721 "..., 30733) = 30733
close(8) = 0
exit_group(0) = ?
Process 32078 detached

I found it puzzling since strace should not have an influence on the state of a process.

In the /proc directory I found the follwing:

/proc/32078# cat status
Name: gzip
State: S (sleeping)
Tgid: 32078
Pid: 32078
PPid: 31592
TracerPid: 0
Uid: 5059 5059 5059 5059
Gid: 5000 5000 5000 5000
FDSize: 256
Groups: 5000 5001 5002
VmPeak: 4140 kB
VmSize: 4140 kB
VmLck: 0 kB
VmHWM: 552 kB
VmRSS: 552 kB
VmData: 492 kB
VmStk: 84 kB
VmExe: 52 kB
VmLib: 1432 kB
VmPTE: 32 kB
Threads: 1
SigQ: 0/68608
SigPnd: 0000000000000000
ShdPnd: 0000000000000000
SigBlk: 0000000000000000
SigIgn: 0000000001300001
SigCgt: 0000000000805002
CapInh: 0000000000000000
CapPrm: 0000000000000000
CapEff: 0000000000000000
CapBnd: fffffffffffffeff
voluntary_ctxt_switches: 55
nonvoluntary_ctxt_switches: 1

Do you have any idea?

Cheers,
Henning Fehrmann
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/