Messages in this thread | | | Date | Tue, 7 Mar 2000 14:02:08 -0500 | From | Ian Peters <> | Subject | mystery process of death |
| |
Just earlier today, with 2.3.49 (along with tytso's one-line patch to softirq.c), processes like 'w' and 'ps' stopped returning. Instead they did the whole "claim to be chewing up cycles and killing the load average while really doing nothing because the system stays responsive" game.
I wasn't local to the box, so I just opened up a lot of ssh connections and tried to figure out what could be up before rebooting. I strace(1)'d both of them, and they both block with output like:
--- SNIP --- stat("/proc/217", {st_mode=S_IFDIR|0555, st_size=0, ...}) = 0 open("/proc/217/stat", O_RDONLY) = 7 read(7, "217 (apache) S 202 202 202 0 -1 "..., 511) = 179 close(7) = 0 open("/proc/217/statm", O_RDONLY) = 7 read(7, "90 10 7 5 0 5 3\n", 511) = 16 close(7) = 0 open("/proc/217/status", O_RDONLY) = 7 read(7, "Name:\tapache\nState:\tS (sleeping)"..., 511) = 419 close(7) = 0 open("/proc/217/cmdline", O_RDONLY) = 7 read(7, "/usr/sbin/apache\0", 2047) = 17 close(7) = 0 open("/proc/217/environ", O_RDONLY) = 7 read(7, "PWD=/\0BOOT_FILE=/boot/vmlinuz-2."..., 2047) = 370 brk(0x8192000) = 0x8192000 close(7) = 0 stat("/proc/3183", {st_mode=S_IFDIR|0555, st_size=0, ...}) = 0 open("/proc/3183/stat", O_RDONLY) = 7 read(7, "3183 (gpm) S 1 3183 3183 0 -1 32"..., 511) = 194 close(7) = 0 open("/proc/3183/statm", O_RDONLY) = 7 read(7, "24 13 8 8 0 5 5\n", 511) = 16 close(7) = 0 open("/proc/3183/status", O_RDONLY) = 7 read(7, "Name:\tgpm\nState:\tS (sleeping)\nPi"..., 511) = 406 close(7) = 0 open("/proc/3183/cmdline", O_RDONLY) = 7 read(7, "/usr/sbin/gpm\0-m\0/dev/psaux\0-t\0p"..., 2047) = 98 close(7) = 0 open("/proc/3183/environ", O_RDONLY) = 7 read(7, "PWD=/\0COLORFGBG=default;0\0XAUTHO"..., 2047) = 1183 close(7) = 0 stat("/proc/5700", {st_mode=S_IFDIR|0555, st_size=0, ...}) = 0 open("/proc/5700/stat", O_RDONLY) = 7 read(7,
Both 'ps' and 'w' blocked on the same process, 5700. What is process 5700, you might ask? I dunno, I couldn't figure it out either. If I tried to look in /proc myself, my shell died too, of course.
So that's my mysterious story. I had to reboot the box because, even though I'm running 2.3.x kernels, it is occasionally useful and having random processes getting stuck wasn't good. If anyone has ideas of why it happened, or how I can look harder the next time it does happen, etc., and/or just wants to fix whatever bug it could be, I'd love to hear it.
-- Ian Peters | GnuPG Key ID 5C23D20C | Those who know what's best itp@gnu.org | E584 2558 FAC3 BEAB EFAC | for us must rise and save itp@cmu.edu | FC74 CFED 7E24 5C23 D20C | us from ourselves...
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |