I have done quite a bit of research, and cant find any documentation on this. I may be using the wrong terminology with the term "thread dump", so I apologize for any lack of knowledge on this topic.
I am looking for a way to take a thread dump of an Apache webserver running on an RHEL AWS instance. The Apache webserver is running on the httpd process. From my understanding, tools like jstack and jmap are specific to taking thread dumps of JVM processes only, and httpd is not compatible.
When I use ps -aux | grep apache I receive the following output of processes.
ps -aux | grep apache
ssm-user 1539 0.0 0.0 112812 960 pts/1 S+ 17:26 0:00 grep apache
apache 1554 0.0 0.2 272752 15488 ? S 16:12 0:00 /usr/sbin/httpd -DFOREGROUND
apache 1713 0.0 0.1 270376 13452 ? S 16:13 0:00 /usr/sbin/httpd -DFOREGROUND
apache 1715 0.0 0.1 270376 13452 ? S 16:13 0:00 /usr/sbin/httpd -DFOREGROUND
apache 1956 0.0 0.1 270556 14080 ? S 13:16 0:00 /usr/sbin/httpd -DFOREGROUND
apache 5700 0.0 0.2 272752 15704 ? S May22 0:00 /usr/sbin/httpd -DFOREGROUND
apache 5701 0.0 0.2 272996 16064 ? S May22 0:00 /usr/sbin/httpd -DFOREGROUND
apache 5703 0.0 0.1 270676 14368 ? S May22 0:00 /usr/sbin/httpd -DFOREGROUND
apache 5704 0.0 0.2 272720 15636 ? S May22 0:00 /usr/sbin/httpd -DFOREGROUND
apache 11441 0.0 0.2 272776 15744 ? S 04:55 0:00 /usr/sbin/httpd -DFOREGROUND
apache 11442 0.0 0.2 272800 15832 ? S 04:55 0:00 /usr/sbin/httpd -DFOREGROUND
I am also unsure why there are so many of the same process, but I wanted to figure out if its possible to take a thread dump of one of the processes before I dive into solving the duplicate processes.
Are there any tools or commands I can use to accomplish a thread dump of an httpd process?
Related
I'm really struggling with how to find processes by name in linux. I'm sure it's probably something simple that I'm missing.
Are you looking for the command ps ?
Here an example
nabil#LAPTOP:~$ ps xua | grep python
rootwsl 327 0.0 0.1 29568 17880 ? Ss Jan30 0:02 /usr/bin/python3 /usr/bin/networkd-dispatcher --run-startup-triggers
rootwsl 411 0.0 0.1 108116 20740 ? Ssl Jan30 0:00 /usr/bin/python3 /usr/share/unattended-upgrades/unattended-upgrade-shutdown --wait-for-signal
nabil 106387 0.0 0.0 3444 736 pts/1 S+ 23:26 0:00 grep --color=auto python
This is the Dockerfile I created for installing httpd on centos:
#Installing HTTPD
FROM centos:latest
MAINTAINER xxx#gmail.com
RUN yum install -y httpd
EXPOSE 80
#ENTRYPOINT ["systemctl"]
ENTRYPOINT ["/usr/sbin/httpd"]
After building, when I run the container I can see too many httpd process running inside this container:
docker run -d -p 80:80 httpd:4.0 -DFOREGROUND
Output of Docker top command:
UID PID PPID C STIME TTY TIME CMD
root 2457 2443 0 04:26 ? 00:00:00 /usr/sbin/httpd -DFOREGROUND
apache 2474 2457 0 04:26 ? 00:00:00 /usr/sbin/httpd -DFOREGROUND
apache 2475 2457 0 04:26 ? 00:00:00 /usr/sbin/httpd -DFOREGROUND
apache 2476 2457 0 04:26 ? 00:00:00 /usr/sbin/httpd -DFOREGROUND
apache 2477 2457 0 04:26 ? 00:00:00 /usr/sbin/httpd -DFOREGROUND
apache 2478 2457 0 04:26 ? 00:00:00 /usr/sbin/httpd -DFOREGROUND
apache 2491 2457 0 04:26 ? 00:00:00 /usr/sbin/httpd -DFOREGROUND
apache 2492 2457 0 04:26 ? 00:00:00 /usr/sbin/httpd -DFOREGROUND
apache 2493 2457 0 04:26 ? 00:00:00 /usr/sbin/httpd -DFOREGROUND
root 2512 2500 0 04:27 pts/0 00:00:00 /bin/bash
apache 2532 2457 0 04:27 ? 00:00:00 /usr/sbin/httpd -DFOREGROUND
Please let me know why so many httpd processes running and how to have only one process with PID 1 ?
Apache runs multiple processes to be ready catch up a client request fast, because spawning a server process is slow, so it is better to have one ready when request comes in.
You can configure their number in httpd.conf through StartServers, MinSpareServers, MaxSpareServers and ServerLimit directives.
The following command
[cuser#vdev-b7-5461-4f3c-9d98-7fc3ec6 kafka]$ ps aux|grep apache
apache 3967 0.0 0.0 175412 3192 ? S 03:40 0:00 /usr/sbin/httpd -k start
apache 3968 0.0 0.0 175412 3192 ? S 03:40 0:00 /usr/sbin/httpd -k start
apache 3969 0.0 0.0 175412 3192 ? S 03:40 0:00 /usr/sbin/httpd -k start
apache 3970 0.0 0.0 175412 3192 ? S 03:40 0:00 /usr/sbin/httpd -k start
apache 3971 0.0 0.0 175276 2700 ? S 03:40 0:00 /usr/sbin/httpd -k start
apache 3972 0.0 0.0 175276 2700 ? S 03:40 0:00 /usr/sbin/httpd -k start
apache 3974 0.0 0.0 175412 3188 ? S 03:40 0:00 /usr/sbin/httpd -k start
apache 3975 0.0 0.0 175276 2696 ? S 03:40 0:00 /usr/sbin/httpd -k start
clduser 9566 0.0 0.0 103244 844 pts/1 S+ 16:41 0:00 grep apache
There are multiple apache process running in my machine. Is there a way for me to determine the files/directory each of these process uses to run it?
Also, would like to know the ports that they are using?
[user#vdev-37-5461-4f3c-9d98-7af28dfc3ec6 kafka]$ netstat -tulpn|grep 80
(No info could be read for "-p": geteuid()=500 but you should be root.)
tcp 0 0 :::80 :::* LISTEN -
This above command shows port 80 is being used but I want to know what process is using it?
[duser#vpsq-dev-4ac0b737-5461-4f3c-9d98-7af28dfc3ec6 kafka]$ find / -iname apache 2>/dev/null
/opt/logstash/vendor/bundle/jruby/1.9/gems/jruby-kafka-1.4.0-java/lib/org/apache
/opt/logstash/vendor/bundle/jruby/1.9/gems/logstash-output-elasticsearch-0.1.18-java/lib/org/apache
/opt/logstash/vendor/bundle/jruby/1.9/gems/jruby-kafka-1.1.2-java/lib/org/apache
/home/duser/logstash-1.5.0.rc2/vendor/bundle/jruby/1.9/gems/logstash-output-elasticsearch-0.1.18-java/lib/org/apache
/home/duser/logstash-1.5.0.rc2/vendor/bundle/jruby/1.9/gems/jruby-kafka-1.1.2-java/lib/org/apache
I'm trying to automatically start a process when I enter runlevel 1 by init 1. It's a watchdog which has to send a life signal all the time therefore in runlevel 1 too! But when runlevel 1 is entered each process is killed and the system switches to runlevel S. I tried to tell linux to start my process by update-rc.d -f watchdog 99 1 S .. The resulting entries in /etc/rc1.d/ and /etc/rcS.d/ are:
/etc/rc1.d/:
.
.
.
S30killprocs -> ../init.d/killprocs
S90single -> ../init.d/single
S99watchdog -> ../init.d/watchdog
.
.
.
/etc/rcS.d/:
.
.
.
S01glibc.sh -> ../init.d/glibc.sh
S02hostname.sh -> ../init.d/hostname.sh
S99watchdog -> ../init.d/watchdog
.
.
.
A ps ax after runlevel 1 was entered sadfully doesn't return my watchdog process. I have to start it manually.
PID TTY STAT TIME COMMAND
1 ? Ss 0:00 init [S]
2 ? S 0:00 [kthreadd]
3 ? S 0:00 [ksoftirqd/0]
4 ? S 0:00 [kworker/0:0]
5 ? S 0:00 [kworker/u:0]
6 ? S 0:00 [rcu_kthread]
7 ? S< 0:00 [khelper]
8 ? S 0:00 [kworker/u:1]
104 ? S 0:00 [sync_supers]
106 ? S 0:00 [bdi-default]
108 ? S< 0:00 [kblockd]
119 ? S 0:00 [khubd]
219 ? S 0:00 [kswapd0]
220 ? S 0:00 [fsnotify_mark]
221 ? S< 0:00 [aio]
314 ? S< 0:00 [scsi_tgtd]
347 ? S< 0:00 [kpsmoused]
349 ? S 0:00 [kworker/0:1]
366 ? S 0:00 [w1_bus_master1]
390 ? S 0:00 [mmcqd/0]
395 ? S 0:00 [jbd2/mmcblk0p2-]
396 ? S< 0:00 [ext4-dio-unwrit]
475 ? S 0:00 [flush-179:0]
4532 ttyS0 Ss 0:00 init [S]
4533 ttyS0 S 0:00 bash
4536 ttyS0 R+ 0:00 ps ax
The system I'm working on is Debian GNU/Linux 5.0 debarm (embedded). Any hints or solution? Thank you.
As said here:
Run Level 1 is known as 'single user' mode. A more apt description
would be 'rescue', or 'trouble-shooting' mode. In run level 1, no
daemons (services) are started.
You have several options to get around this limitation:
Don't use runlevel 1, this is not what it is meant for
Start the watchdog on login (if this is a valid option), e.g. .bashrc.
I solved it by adding following line into /etc/inittab:
~~:S:wait:/etc/init.d/watchdog start
~~:S:wait:/sbin/sulogin
Note: It must be before /sbin/sulogin.
I have a bunch of processes owned by apache that are running for days because they are stuck.
apache 11173 0.1 0.0 228248 27744 ? Ss Sep27 3:58 php /var/www/html/myproj/symfony cron:aggregation --env=prod
apache 12609 0.1 0.0 228244 27744 ? Ss Sep18 19:30 php /var/www/html/myproj/symfony cron:aggregation --env=prod
apache 14646 0.1 0.0 228244 27744 ? Ss Sep17 21:30 php /var/www/html/myproj/symfony cron:aggregation --env=prod
apache 15900 0.1 0.0 228244 27744 ? Ss Sep20 15:46 php /var/www/html/myproj/symfony cron:aggregation --env=prod
apache 16169 0.1 0.0 228248 27752 ? Ss Sep22 12:16 php /var/www/html/myproj/symfony cron:aggregation --env=prod
apache 16887 0.1 0.0 228244 27748 ? Ss Sep21 14:04 php /var/www/html/myproj/symfony cron:aggregation --env=prod
apache 16950 0.1 0.0 228244 27744 ? Ss Sep28 2:25 php /var/www/html/myproj/symfony cron:aggregation --env=prod
apache 19195 0.1 0.0 228244 27748 ? Ss Sep23 10:29 php /var/www/html/myproj/symfony cron:aggregation --env=prod
apache 24605 0.1 0.0 228248 27752 ? Ss Sep24 8:48 php /var/www/html/myproj/symfony cron:aggregation --env=prod
apache 26442 0.1 0.0 228244 27744 ? Ss 03:45 0:50 php /var/www/html/myproj/symfony cron:aggregation --env=prod
apache 29714 0.1 0.0 228248 27752 ? Ss Sep25 7:06 php /var/www/html/myproj/symfony cron:aggregation --env=prod
apache 31031 0.1 0.0 228248 27752 ? Ss Sep26 5:30 php /var/www/html/myproj/symfony cron:aggregation --env=prod
I need to kill them all. And obviously I want to do it safely.
Thus, ideally I should kill them as apache using something like this:
kill 11173
The problem is that the apache userdoesn't have a shell.
So it seems the only way is escalate to root and kill the process as root. But it is not safely at all (I may kill other processes by mistake).
Has anybody got a better solution?
Thanks,
Daniele
sudo -u apache kill 11173
This should belong on http://serverfault.com I guess... but if you want to kill all processes named apache, run killall apache as root. Alternatively, change identity to your apache user with su apache and kill your processes there using kill as you did.