Comments on: How to Commit Genocide on Annoying Processes http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes pink is the new black Thu, 26 Jan 2012 17:17:54 +0000 hourly 1 http://wordpress.org/?v=3.2.1 By: My B Side http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/comment-page-1#comment-9546 My B Side Tue, 18 Nov 2008 17:43:33 +0000 http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/#comment-9546 ps .... | .... | ..... | xargs kill -9 ps …. | …. | ….. | xargs kill -9

]]> By: /dev/random http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/comment-page-1#comment-9394 /dev/random Sat, 28 Jun 2008 06:51:15 +0000 http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/#comment-9394 Haha, I actually encountered this just today. I came across this blog again looking for something that would be more final than a kill -9. Unfortunately there's nothing. The system's getting rebooted, and there's going to be alot of angry customers :) Haha, I actually encountered this just today. I came across this blog again looking for something that would be more final than a kill -9. Unfortunately there’s nothing. The system’s getting rebooted, and there’s going to be alot of angry customers :)

]]> By: Nessa http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/comment-page-1#comment-8797 Nessa Tue, 04 Dec 2007 14:49:19 +0000 http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/#comment-8797 Most of the hack processes I've come across were spawned by processes that no longer exist, but only a fraction of those are ones that won't respond to a sigterm. I forgot the actual representation, but I believe they usually have a 'T' next to them when you do a 'ps aux'. Your explanation makes sense though Most of the hack processes I’ve come across were spawned by processes that no longer exist, but only a fraction of those are ones that won’t respond to a sigterm. I forgot the actual representation, but I believe they usually have a ‘T’ next to them when you do a ‘ps aux’. Your explanation makes sense though

]]> By: /dev/random http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/comment-page-1#comment-8789 /dev/random Sun, 02 Dec 2007 01:55:54 +0000 http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/#comment-8789 Oh! I have seen zombie processes that refuse to be killed, but they aren't actually processes, just the proc tree from a child process has exited, but the return code of the process hasn't been read by the parent process. Maybe a process that's blocked waiting for IO wouldn't die either? Next time you find one of those, could you look at the status column of ps? I'm really curious now! Oh! I have seen zombie processes that refuse to be killed, but they aren’t actually processes, just the proc tree from a child process has exited, but the return code of the process hasn’t been read by the parent process. Maybe a process that’s blocked waiting for IO wouldn’t die either? Next time you find one of those, could you look at the status column of ps? I’m really curious now!

]]> By: Nessa http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/comment-page-1#comment-8761 Nessa Sat, 01 Dec 2007 06:11:01 +0000 http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/#comment-8761 Yea, it happens some times but usually if you let the process site for a few minutes it like 'unlocks' itself or something. Weird because that defeats the whole purpose of a sigterm. Yea, it happens some times but usually if you let the process site for a few minutes it like ‘unlocks’ itself or something. Weird because that defeats the whole purpose of a sigterm.

]]> By: BinaryPhalanx http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/comment-page-1#comment-8747 BinaryPhalanx Sat, 01 Dec 2007 02:49:49 +0000 http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/#comment-8747 I haven't seen anything stand up to a signal 9 since I started with Linux when I was 12, though I suppose that doesn't rule out the possibility. Now I feel old :( I haven’t seen anything stand up to a signal 9 since I started with Linux when I was 12, though I suppose that doesn’t rule out the possibility.

Now I feel old :(

]]> By: Nessa http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/comment-page-1#comment-8717 Nessa Fri, 30 Nov 2007 23:54:05 +0000 http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/#comment-8717 There are many times where a process will spawn and refuse to die via sigkill, there have been few cases where I've had to actually reboot a server because of stale processes that don't want to die. No insults intended. There are many times where a process will spawn and refuse to die via sigkill, there have been few cases where I’ve had to actually reboot a server because of stale processes that don’t want to die. No insults intended.

]]> By: /dev/random http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/comment-page-1#comment-8669 /dev/random Fri, 30 Nov 2007 09:46:06 +0000 http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/#comment-8669 I've never heard of a SIGKILL not killing a process. Also, if the process hides it's true name than doing a ps isn't going to show it's real name either, since at that point they've changed the contents of /proc/$PID/cmdline. While killall does read from /proc/$PID/stat which wouldn't necessarily be the same as output from 'ps aux', "kill -9 `pgrep name`" reads from both /proc/$PID/stat and /proc/$PID/cmdline. I also forgot about the pkill binary, which combines the powers of kill and pgrep into one sweet command. so it could be simplified even further into "pkill -9 $apparent_process_name". I always thought pgrep was a cool and underutilized tool, it supports regex, and various flags to limit the scope too ("pkill -u apache sendmail" is an example). I just wanted to be nice and let you know about it if you didn't already. No need to insult my admin skills, that's how flame wars get started. kthxbye I’ve never heard of a SIGKILL not killing a process. Also, if the process hides it’s true name than doing a ps isn’t going to show it’s real name either, since at that point they’ve changed the contents of /proc/$PID/cmdline. While killall does read from /proc/$PID/stat which wouldn’t necessarily be the same as output from ‘ps aux’, “kill -9 `pgrep name`” reads from both /proc/$PID/stat and /proc/$PID/cmdline.

I also forgot about the pkill binary, which combines the powers of kill and pgrep into one sweet command. so it could be simplified even further into “pkill -9 $apparent_process_name”.

I always thought pgrep was a cool and underutilized tool, it supports regex, and various flags to limit the scope too (“pkill -u apache sendmail” is an example). I just wanted to be nice and let you know about it if you didn’t already. No need to insult my admin skills, that’s how flame wars get started. kthxbye

]]> By: BinaryPhalanx http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/comment-page-1#comment-8650 BinaryPhalanx Thu, 29 Nov 2007 06:32:36 +0000 http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/#comment-8650 pwned pwned

]]> By: Nessa http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/comment-page-1#comment-8646 Nessa Thu, 29 Nov 2007 02:18:58 +0000 http://www.v-nessa.net/2007/11/21/how-to-commit-genocide-on-annoying-processes/#comment-8646 No, it doesn't do the same thing. We're talking about hack processes spawned outside of httpd that can't be killed with a simple killall command. I'm sure it sounds easy just to do a killall -9 <process> but any system admin would know that 1) some processes hide their true names and 2) some processes refuse to die that way. In other words, if a kill/killall would have fixed the problem, I wouldn't have even made this post. No, it doesn’t do the same thing. We’re talking about hack processes spawned outside of httpd that can’t be killed with a simple killall command. I’m sure it sounds easy just to do a killall -9 but any system admin would know that 1) some processes hide their true names and 2) some processes refuse to die that way. In other words, if a kill/killall would have fixed the problem, I wouldn’t have even made this post. ]]>