Apache hangs

Message

jcampbell · Post by **jcampbell** » Sun Sep 25, 2005 4:13 am

This is strange... I start apache and it works. I restart it and get this:

tiger ~ # /etc/init.d/apache2 start
 * Starting apache2 ...                                                                         [ ok ]
tiger ~ # /etc/init.d/apache2 stop
 * Stopping apache2 ...
httpd (no pid file) not running                                                                 [ ok ]
tiger ~ # /etc/init.d/apache2 start
 * Starting apache2 ...
(98)Address already in use: make_sock: could not bind to address [::]:80
no listening sockets available, shutting down
Unable to open logs
tiger ~ # /etc/init.d/apache2 zap
 * Manually resetting apache2 to stopped state.
tiger ~ # killall -s SIGKILL apache2
tiger ~ #

The apache server stays running even after the init script dies. There is no apache.pid in /var/run but the process continues to run. Best part is that no matter how many times I start apache after the initial failure, the server hangs when I try to browse any sites that it hosts. The only way I've been able to fix it is to reboot.

I don't recall ever having this problem before...

jcampbell · Post by **jcampbell** » Sun Sep 25, 2005 4:50 am

okay... more info... viewing apache's error_log:

Code: Select all

[Sun Sep 25 01:16:39 2005] [notice] suEXEC mechanism enabled (wrapper: /usr/sbin/suexec2)
[Sun Sep 25 01:16:39 2005] [notice] Digest: generating secret for digest authentication ...

I commented out this line in httpd.conf:

Code: Select all

LoadModule auth_digest_module            modules/mod_auth_digest.so

and it cured the problem. It's okay for a work-around, but why is this causing apache to hang?

Apparently, this has been an issue before? It's a different apache version... I have 2.0.54-r31
http://forums.gentoo.org/viewtopic-t-15 ... odule.html

gbpack · Post by **gbpack** » Sun Sep 25, 2005 7:03 am

I'm still having this exact problem. I tried the workaround but that didn't work either. I posted this in IRC a little while ago but no one replied. If anyone has any ideas .... it would be greatly appreciated.

Janne Pikkarainen · Post by **Janne Pikkarainen** » Sun Sep 25, 2005 8:19 am

I had the same problem with Apache 1.3.x some years back, but I can't remember if I ever fixed it (in any other way than not using the mod_auth_digest).

Anyway, maybe this problem could be a classic "kernel runs out of entropy" problem? If your server runs mostly idle and mod_auth_digest gets its digest data from /dev/random, it could be that kernel is unable to generate random data fast enough. Check that with command cat /proc/sys/kernel/random/entropy_avail -- if it's 0, then your server has temporarily run out of entropy.

In that case you may emerge rng-tools and run rngd, which generates some less perfect entropy.

gbpack · Post by **gbpack** » Sun Sep 25, 2005 3:25 pm

cat /proc/sys/kernel/random/entropy_avail says 4096.

gbpack · Post by **gbpack** » Sun Sep 25, 2005 5:04 pm

installed rng-tools and started it up. Tried starting apache2 again and it still doesn't put a pid file in /var/run. Apache says it started ok, but my home page won't load. Error log says:

[Sun Sep 25 11:31:05 2005] [notice] suEXEC mechanism enabled (wrapper: /usr/sbin/suexec2)
[Sun Sep 25 11:31:07 2005] [notice] Digest: generating secret for digest authentication ...
[Sun Sep 25 11:31:07 2005] [notice] Digest: done

I have no idea what to do?

Troy · Post by **Troy** » Mon Sep 26, 2005 4:46 am

It's not just you. Since the update, I can't restart apache

I have no entropy and no physical access to the system to create entropy. I'd used MY_BUILTINS="--with-devrandom=/dev/urandom" emerge apache in the past to fix this. It's not fixing it this time. I added urandom to my USE flags as suggested by the apr-0.9.6 ebuild, rebuilt APR and apache2 with no success. Not good...

taiger · Post by **taiger** » Mon Sep 26, 2005 6:35 pm

I have the same problem

I have temporaly cammented:

modules/mod_auth_digest.so

chrismear · Post by **chrismear** » Sat Oct 01, 2005 9:17 am

This might be some help. On one of my servers, it generates the secret instantly, so I never noticed this problem. But on another server, it takes about two minutes to generate the secret. Once it's done that, though, apache runs perfectly:

[Sat Oct 01 10:00:36 2005] [notice] suEXEC mechanism enabled (wrapper: /usr/sbin/suexec2)
[Sat Oct 01 10:00:36 2005] [notice] Digest: generating secret for digest authentication ...
[Sat Oct 01 10:02:33 2005] [notice] Digest: done
[Sat Oct 01 10:02:34 2005] [notice] Apache configured -- resuming normal operations

I don't know how to make it work faster, but in the meantime I'm happy just to wait for a couple of minutes while it sorts itself out.

Taily · Post by **Taily** » Fri Oct 07, 2005 9:20 pm

Well for me the apache server would just "hang" at

Code: Select all

[Sat Oct 08 00:12:27 2005] [notice] suEXEC mechanism enabled (wrapper: /usr/sbin/suexec2)
[Sat Oct 08 00:12:28 2005] [notice] Digest: generating secret for digest authentication ...

However when you did /etc/init.d/apache2 stop.. the server would actually not die, it finnishes loading itself and is alive :/

This was all due to an error in apache configuration. I suggest running apache2 -k stop after you've started it and see if it reports some error.

j-m · Post by **j-m** » Fri Oct 07, 2005 10:10 pm

emerge sys-apps/clrngd

bytelab · Post by **bytelab** » Tue Nov 08, 2005 12:07 pm

I have the same problem but after a little testing I found it to be an entropy issue.
Thanks guys for a very informative thread

TJNII · Post by **TJNII** » Wed Nov 09, 2005 2:22 am

Wow. I never knew a computer could run out of entropy.

Code: Select all

PrISUm2 / # ps -ec | grep apache
25393 TS   14 ?        00:00:00 apache2
PrISUm2 / # emerge -s I_Like_Pudding
Searching...   
[ Results for search key : I_Like_Pudding ]
[ Applications found : 0 ]
 

PrISUm2 / # ps -ec | grep apache
25393 TS   23 ?        00:00:00 apache2
25482 TS   21 ?        00:00:00 apache2
25484 TS   20 ?        00:00:00 apache2
25485 TS   19 ?        00:00:00 apache2
25486 TS   18 ?        00:00:00 apache2
25487 TS   17 ?        00:00:00 apache2
25488 TS   15 ?        00:00:00 apache2
PrISUm2 / #

ohcnetwork · Post by **ohcnetwork** » Tue Dec 27, 2005 7:50 am

Hi, I know that this is not RHE or Apache forum, but after I spent many hours searching for a solution to my problem, I only found a few places that discuss the problem, and this seems to be one place that I can ask about the problem and its possible solution, so I'm posting here. Hope it's ok ...

I am having a similar problem, and Apache2 installed on my server hangs. Apache2 is installed on RHE and it happens periodically. I tried to find causes, but the reason why I'm posting here is that I see the following error message in my error log file (and this is mentioned in this thread):

Code: Select all

[Fri Dec 23 01:00:30 2005] [notice] Digest: generating secret for digest authentication ...
[Fri Dec 23 01:00:30 2005] [notice] Digest: done

I see the notice recorded number of times although this is not exactly when Apache2 hangs. My questions are as follows:

Q1. Does the notice indicate that Apache2 can hang periodically?

Q2. If so, does commenting out modules/mod_auth_digest.so resolve the problem (temporarily)?

Q2.2. What does modules/mod_auth_digest.so really do and do I need it?

Q3. A couple of people mention about entropy in this thread, but what is it (my apology if this is such a trivial question)?

Q4. If entropy is really the problem on my system, how do I know if it's really the case?

Q5. If it is the case, how can I fix it?

ocbMaurice · Post by **ocbMaurice** » Tue Dec 27, 2005 3:19 pm

Your Problem is not likely related to the entropy pool. The problem with the entropy pool is the automatic generation of the secret for mod_auth_digest. To create a new secret (cypher key) it needs some (real) random data which is taken from the entropy pool. If the entropy pool is empty, the generation will take a long time. But in your case the generation was done in a very short time.

ohcnetwork · Post by **ohcnetwork** » Tue Dec 27, 2005 6:46 pm

ocbMaurice wrote:Your Problem is not likely related to the entropy pool. The problem with the entropy pool is the automatic generation of the secret for mod_auth_digest. To create a new secret (cypher key) it needs some (real) random data which is taken from the entropy pool. If the entropy pool is empty, the generation will take a long time. But in your case the generation was done in a very short time.

... then what is causing Apache (on my server) to hang?

mizery de aria · Post by **mizery de aria** » Wed Feb 15, 2006 12:34 am

*bump*
having this problem too...

pirus3210 · Post by **pirus3210** » Fri Mar 03, 2006 1:43 pm

i had the same problem.. Apache2 wouldnt start.

Try
#ps aux

if you can see apache there do this
#pkill -KILL apache

then try restarting

LL0rd · Post by **LL0rd** » Sat Mar 04, 2006 11:38 am

Hi,

I have exactly the same problem.

Code: Select all

[Fri Dec 23 01:00:30 2005] [notice] Digest: generating secret for digest authentication ...
[Fri Dec 23 01:00:30 2005] [notice] Digest: done

That problem appears, when I try to restart apache with the restart command. You can do everything you want, you can't get apache up. But if I restart apache with stop and then start, apache runs without problems. To solve the problem with hanging apache, I restarted the whole OS, after that, I could start apache with the start command.

Thats a realy strange behavior, it's like...... windows

Bash[DevNull] · Post by **Bash[DevNull]** » Wed Mar 08, 2006 11:13 pm

I've removed "LoadModule auth_digest_module modules/mod_auth_digest.so" line from config,
BUT it's not HELP me because on restart/start apache2 freezes (50-200sec) on reading from /dev/random!!!
Why? I have no other digest modules....

LL0rd · Post by **LL0rd** » Wed Mar 08, 2006 11:21 pm

okay, I don't realy know, what the error realy is, but I had restarted my apache webserver. After the restart I had just one apache process. Then me and my girlfriend hat a little bit fun

After about half an hour I was back and apache was running

maverick6664 · Post by **maverick6664** » Fri Mar 24, 2006 1:57 pm

Thanks all. Having the same problem I found this thread today, and finally found tapping the keyboard (not on a remote console, but on the "real" console w/o X Window) seemed to help.

Maybe moving the mouse or other factors will also help.

The value of /proc/sys/kernel/random/entropy_avail on my http server is often low (around or below 100) and this value actually affects the freezing time period. My kernel is generic 2.6.16.

EDIT: Running X Server doesn't seem to matter at all. Keyboard and mouse always affect much. Moving mouse is the most efficient in my case.

eltech · Post by **eltech** » Thu Apr 13, 2006 7:24 pm

j-m wrote:
emerge sys-apps/clrngd

This did work for me today as i ran into this problem .. and i just ran clrngd and the problem has been fixed, but will this run again on its own if and when centropy memory runs out ?

how can i prevent such happening>?

okuhl · Post by **okuhl** » Mon Apr 24, 2006 2:11 pm

Hi,

this worked for me, too:

Code: Select all

# cat /proc/sys/kernel/random/entropy_avail
0
# emerge -a clrngd
# /etc/init.d/clrngd start
# cat /proc/sys/kernel/random/entropy_avail
256
# /etc/init.d/apache2 start

Regards,
Oliver.

soth · Post by **soth** » Wed Jul 26, 2006 7:51 am

yay! No more problems for me!

Apache hangs

Apache hangs

Solution for me