Problems when upgrade version 1.2.0 to 1.2.6

[arturoth]

Dear guys,

I had a big trouble with mod_cluster because when upgrade my version from 1.2.0-Final to 1.2.6-Final "dmesg" throw this error:

httpd[14408]: segfault at 18 ip 00007f0f6e60c92e sp 00007f0f6d8bec88 error 4 in mod_manager.so[7f0f6e603000+e000]

When i try to find more specifications for this error, i found this URL http://www.cyberciti.biz/tips/segmentation-fault-on-linux-unix.html but when i read this i think the problem perhaps must be a bug of mod_manager.

I share for you my specs of my server (i had 2 VM and work with active-passive load balancing cluster behind heartbeat) :

OS:      CentOS release 6.3 (Final)

Architecture:          x86_64

CPU op-mode(s):        32-bit, 64-bit

Byte Order:            Little Endian

CPU(s):                4

CPU MHz:               2000.000

Mem:         6 GB

Swap:      2 GB

And my httpd and mpm_worker_module important settings are :

<IfModule mpm_worker_module>

    ThreadLimit         120

    ServerLimit 16

    StartServers        8

    MinSpareThreads     5

    MaxSpareThreads     20

    MaxClients          1920

    ThreadsPerChild     120

    MaxRequestsPerChild 16384

</IfModule>

  <VirtualHost *:36666>

    KeepAliveTimeout 300

    MaxKeepAliveRequests 0

   AdvertiseFrequency 5

    EnableMCPMReceive

Please any help will may work for me.

Best regards

[jfclere]

Where did you get mod_cluster libraries?

[arturoth]

Man,

I get libraries from http://downloads.jboss.org/mod_cluster/1.2.6.Final/linux-x86_64/mod_cluster-1.2.6.Final-linux2-x64-ssl.tar.gz and make clean install and replace conf files with the other ones.

New scenery:

In version 1.2.0-Final i get the same error:

httpd[7474]: segfault at 18 ip 00007fb65bc4592e sp 00007fb65aef7cc8 error 4 in mod_manager.so[7fb65bc3c000+e000]

:/ Please help us.

[mbabacek]

Hi Arturo, the most crucial thing here is the version of your Apache HTTP Server. The libraries you downloaded work well only with httpd 2.2.2x.

If you tell us the version of the httpd you need to run, we will be able to help you i.e. by guiding you through compiling your own binaries, preferably of mod_cluster 1.2.9.Final or 1.3.x., because mod_cluster 1.2.6.Final contains an unpleasant performance bug.

K.

[arturoth]

Michael.

We had Apache/2.2.21 and jboss-eap-6.0 for apps. If you could help me to solve this fix, would be very greatful.

Thanks for all and I wait 'ur answer.

Best regards,

[jfclere]

So you use the httpd from /opt/jboss/httpd/sbin/httpd of the mod_cluster-1.2.6.Final-linux2-x64-ssl.tar.gz and you have a core, correct?

What I did  download and install (opt/jboss/httpd/sbin/installhome.sh)  in my fedora20:

+++

[jfclere@jfcpc TMP]$ /home/jfclere/TMP/TMP/opt/jboss/httpd/sbin/apachectl -V
Server version: Apache/2.2.21 (Unix)
Server built:   Oct  1 2013 10:20:37
Server's Module Magic Number: 20051115:30
Server loaded:  APR 1.4.5, APR-Util 1.3.12
Compiled using: APR 1.4.5, APR-Util 1.3.12
Architecture:   64-bit
Server MPM:     Worker
  threaded:     yes (fixed thread count)
    forked:     yes (variable process count)

+++

Basic tests are OK.  You are probably mixing httpd from centos and modules from the bundle or new httpd and old httpd. you should clean your box and make a fresh installation of the bundle.

If that still doesn't help try to debug httpd (LogLevel debug) in httpd.conf

[rhusar]

In case you want to build your own modules as suggested by Michal, see instructions here: mod_cluster Documentation. Don't worry, it's easy.

[arturoth]

Man,

There is the specifications of MC :

/opt/jboss/httpd/sbin/apachectl -V

Server version: Apache/2.2.21 (Unix)

Server built:   Oct  1 2013 10:20:37

Server's Module Magic Number: 20051115:30

Server loaded:  APR 1.4.5, APR-Util 1.3.12

Compiled using: APR 1.4.5, APR-Util 1.3.12

Architecture:   64-bit

Server MPM:     Worker

  threaded:     yes (fixed thread count)

    forked:     yes (variable process count)

Server compiled with....

-D APACHE_MPM_DIR="server/mpm/worker"

-D APR_HAS_SENDFILE

-D APR_HAS_MMAP

-D APR_HAVE_IPV6 (IPv4-mapped addresses enabled)

-D APR_USE_SYSVSEM_SERIALIZE

-D APR_USE_PTHREAD_SERIALIZE

-D SINGLE_LISTEN_UNSERIALIZED_ACCEPT

-D APR_HAS_OTHER_CHILD

-D AP_HAVE_RELIABLE_PIPED_LOGS

-D DYNAMIC_MODULE_LIMIT=128

-D HTTPD_ROOT="/opt/jboss/httpd/httpd"

-D SUEXEC_BIN="/opt/jboss/httpd/sbin/suexec"

-D DEFAULT_SCOREBOARD="logs/apache_runtime_status"

-D DEFAULT_ERRORLOG="logs/error_log"

-D AP_TYPES_CONFIG_FILE="conf/mime.types"

-D SERVER_CONFIG_FILE="conf/httpd.conf"

Another thing, the install was clean and i had new error:

httpd[24802]: segfault at 64696e6f91 ip 00007f7cfe40429e sp 00007f7cfd7c0c88 error 4 in mod_slotmem.so[7f7cfe402000+3000]

In this moment, i will try with new release (1.3.0) and advise you if all is OK.

Best regards,

[arturoth]

Guys.

When i tried to compile the last version (modcluster/mod_cluster · GitHub) with mod_cluster Documentation I get this error :

[Mon Jun 16 11:32:17 2014] [emerg] ap_lookup_provider mod_cluster_slotmem failed Configuration Failed

And my settings in "httpd.conf" are.

LoadModule manager_module /opt/jboss/httpd/lib/httpd/modules/mod_manager.so

LoadModule slotmem_module /opt/jboss/httpd/lib/httpd/modules/mod_slotmem.so

LoadModule proxy_cluster_module /opt/jboss/httpd/lib/httpd/modules/mod_proxy_cluster.so

LoadModule advertise_module /opt/jboss/httpd/lib/httpd/modules/mod_advertise.so

Help please!!! Or somebody have the kindness to send me a compile version.

[arturoth]

At the end, finally i repair the problem with this step .

:::::::::::::::::::::::::::::::::::

dw repo

:::::::::::::::::::::::::::::::::::

/tmp/mod_cluster

git clone https://github.com/modcluster/mod_cluster.git

git checkout -b 1.3.0.Final

git checkout 1.3.0.Final

:::::::::::::::::::::::::::::::::::

Repeat this steps

:::::::::::::::::::::::::::::::::::

sh buildconf

./configure --with-apxs=/opt/jboss/httpd/sbin/apxs

make

cp *.so /opt/jboss/httpd/lib/httpd/modules

::::::::::::::::::::::::::::::::::

in 4 paths

::::::::::::::::::::::::::::::::::

/tmp/mod_cluster/native/advertise/

/tmp/mod_cluster/native/mod_cluster_slotmem/

/tmp/mod_cluster/native/mod_manager/

/tmp/mod_cluster/native/mod_proxy_cluster/

::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

Finally in /opt/jboss/httpd/httpd/conf/httpd.conf

::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

#LoadModule slotmem_module /opt/jboss/httpd/lib/httpd/modules/mod_slotmem.so

LoadModule cluster_slotmem_module /opt/jboss/httpd/lib/httpd/modules/mod_cluster_slotmem.so

:::::::::::::::::::::::::::::::::::

Restart apachectl service

:::::::::::::::::::::::::::::::::::

When i pass this steps to productive servers, i comment this.

[jfclere]

#LoadModule slotmem_module /opt/jboss/httpd/lib/httpd/modules/mod_slotmem.so

LoadModule cluster_slotmem_module /opt/jboss/httpd/lib/httpd/modules/mod_cluster_slotmem.so

That change is need for httpd-2.4.x which brings a different mod_slotmem.so than the one mod_cluster is using.

[rhusar]

There is also Jira to use slotmem from httpd:

[MODCLUSTER-54] Use slotmems from httpd-trunk - JBoss Issue Tracker

(if anyone want to pick that work up from the community).

For the 2.4.x i have created Jira to update the documentation:

[MODCLUSTER-421] Document LoadModule mod_cluster_slotmem.so change - JBoss Issue Tracker

so that we dont forget.