command working fine after machine reboot, but it (the command) is hang couple days later.

[Anh-Thu Tran]

I am able to use ps2pdf command after the machine (redhat ES3) is rebooted. After several days working fine, the ps2pdf command is hang and need the machine to reboot to use the command again. Does anybody have any ideas?

[A. Clay Stephenson]

Note that ps2pdf is simply a front-end script for ghostscript (gs). My best guess is that a somewhat corrupt input file can leave gs waiting for more input (on stdin) that it never sees. Rather than rebooting, I suspect you simply need to kill gs. It would probably also be wise to look for any gs patches.

[Anh-Thu Tran]

We have the second box (Older redhat version)with same gs version and it does not have any problem, so I am not sure if the patch can fix the problem.

[A. Clay Stephenson]

The patch could be to one of the many shared libraries that gs uses.

[melvyn burnard]

Posted in wrong forum, moved to correct forum

[Vitaly Karasik_1]

what about free disk space?

[dirk dierickx]

what about just killing the process, rebooting the whole machine is a bit overkill.

[Anh-Thu Tran]

Yes, I can kill the gs process to get out or control C. However, my question is why the ps2pdf command is working fine after the reboot and stop later (just hang). To use the command, we have to reboot the box, and of course it will hang sometime later. How can I fix it?

[dirk dierickx]

run strace when executing the command, you'll get a lot of stuff on your screen, and it should after a while show you what it is trying to do and is not working.

[Anh-Thu Tran]

The strace anh.ps anh.pdf stopped at:
open("anh.pdf", O_RDWR|O_CREAT|O_TRUNC, 0666) = 8
open("/dev/random", O_RDONLY) = 9
fstat64(9, {st_mode=S_IFCHR|0644, st_rdev=makedev(1, 8), ...}) = 0
ioctl(9, SNDCTL_TMR_TIMEBASE or TCGETS, 0xbfff8098) = -1 EINVAL (Invalid argument)
mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb75e5000
read(9,


Any ideas?

[Marek Podmaka]

Please do the strace again and when it stops at the read() syscall, from another console do:
ls -l /proc/PID/fd
This will find out which file/socket is open under this file descriptor. So if this time it hangs on "read(9,", you should find fd 9. If it is file, the symlink will point to the file's original location.

[Anh-Thu Tran]

The command hang again this moring, but I could not anything /proc/PID/fd. Actually, there is no PID under /proc directory. Our machine is Redhat ES3.

[Stuart Browne]

Err he meant the PID of the running process, i.e. /proc/3872/fd (but replace 3872 with the pid of the process you're stracing)