You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@tomcat.apache.org by jkv <j....@gmail.com> on 2010/03/04 06:34:11 UTC
Re: Apache 2.2.3(mod_proxy_ajp) - Tomcat 6.0.13 Loadbalancing -
error logs in apache
Thanks for the reply Ster,
But we don't have the privilege to upgrade Apache, because we are using Red
Had Enterprise Linux and we have to go with the default httpd installation
in it, i.e., 2.2.3, but is there a possibility for us to use mod_jk instead
of mod_proxy for load balancing? I read mod_proxy was quite new when 2.2.3
was released and is this issue related to mod_proxy rather than Apache
Server as a whole? I definitely can think of upgrading Tomcat.
We have a eye popping requirement to handle 15000 concurrent https users
simultaneously, an I am not sure a single Apache Server and five Tomcat
instances (what we now have) can take this?
Regards
Pid Ster wrote:
>
> On 26/02/2010 06:36, jkv wrote:
>>
>> We are using the above setup to load balance http and https request, for
>> https request
>
> Apache HTTPD 2.2.3 was released on 28 Jul 2006, you should definitely
> upgrade to the latest version, there have been *many* important updates
> since then.
>
> Tomcat 6.0.13 was released on 14 May 2007, you should definitely upgrade
> to the latest version.
>
> Please let us know if the upgraded applications still display the same
> problem.
>
>
> p
>
>> Apache is configured to serve the certificates and the request is
>> actually
>> being processed
>> by 3 tomcat instances (TomcatA, TomcatB, TomcatB)running behind. We are
>> getting a strange log in apache
>>
>> [error] ajp_read_header: ajp_ilink_receive failed
>> [error] (120006)APR does not understand this error code: proxy: read
>> response failed from (null) (localhost)
>>
>> I have another question like, when Apache forwards a http/https request
>> to
>> tomcat and suppose that tomcat takes too long to respond! will the same
>> request be routed to a different tomcat??
>> As we have not configured sticky sessions in Apache, We are having many
>> instances where in the java application in TomcatA takes too long (throws
>> exception because it waits to connect to another host which takes too
>> long)
>> in responding back to Apache and I can see logs in other tomcats, say
>> TomcatB and TomcatC with a session Id xxxxxxxxxxxxxxxxxxxxx.NodeA, with
>> empty body which throws exceptions in my application level, why does this
>> occur and can this be eliminated?? I am not sure that this occurs only
>> for
>> https request where we get empty body.
>>
>> Apache configurations are as follows
>>
>> <Proxy balancer://tomcatcluster>
>> Order deny,allow
>> Allow from all
>> BalancerMember ajp://localhost:8109/test route=NodeA
>> BalancerMember ajp://localhost:8209/test route=NodeB
>> BalancerMember ajp://localhost:8309/test route=NodeC
>> </Proxy>
>>
>> ProxyPass /test balancer://tomcatcluster lbmethod=byrequests
>> nofailover=On
>> ProxyPassReverse /test ajp://localhost:8109/test
>> ProxyPassReverse /test ajp://localhost:8209/test
>> ProxyPassReverse /test ajp://localhost:8309/test
>>
>> Thanks in advance.
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
> For additional commands, e-mail: users-help@tomcat.apache.org
>
>
>
--
View this message in context: http://old.nabble.com/Apache-2.2.3%28mod_proxy_ajp%29---Tomcat-6.0.13-Loadbalancing---error-logs-in-apache-tp27714229p27777097.html
Sent from the Tomcat - User mailing list archive at Nabble.com.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Apache 2.2.3(mod_proxy_ajp) - Tomcat 6.0.13 Loadbalancing - error
logs in apache
Posted by Christopher Schultz <ch...@christopherschultz.net>.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
To whom it may concern,
On 3/4/2010 12:34 AM, jkv wrote:
> We have a eye popping requirement to handle 15000 concurrent https users
> simultaneously, an I am not sure a single Apache Server and five Tomcat
> instances (what we now have) can take this?
15000 concurrent users or concurrent requests? The former is manageable
but the latter will probably require some load balancing, as you are
trying to do.
- -chris
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iEYEARECAAYFAkuVFYoACgkQ9CaO5/Lv0PApKgCeMJEoNmIhUKcb9gvl2R5niCsG
iIUAn0kw/KlF0hnh9APwK+YlAdLXUWdR
=fWLT
-----END PGP SIGNATURE-----
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Apache 2.2.3(mod_proxy_ajp) - Tomcat 6.0.13 Loadbalancing - error
logs in apache
Posted by Pid <pi...@pidster.com>.
On 05/03/2010 08:01, jkv wrote:
>
> Thanks for the reply Ster,
>
> "Find someone who does.", sorry when I said privilege I actually meant
> option, that is I cannot install a new Apache over the older 2.2.3 and we
> have to go with the default configuration. Because they have chosen RHEL 5.4
> as the platform and it comes with 2.2.3 version, I will see if I can
> convince them!
You don't have to install it over the top, you can install a different
version alongside it, if need be. If you're serious about performance &
security you simply must stay up to date with software releases.
> "Why? RHEL has a built-in updater application doesn't it?" - I am actually
> new to Linux environment and will try to find out more on this, is it
> something like windows automatic software updater?
Similar in concept only, less automatic and more manual.
> "If you can't upgrade HTTPD, how are you going to compile mod_jk?" - I just
> had the thought that I need only the APR bundle and mod_jk src package to
> compile and load it as a dynamic module! Does it need a HTTPD upgrade?
If you can compile software then you can download the HTTPD source and
install the latest version - you only need root access at the end to
complete the installation, all of the build can be done under another user.
> "Tying one hand behind your back isn't a good way to start." - Good though,
> but I wish I had the authority to change decisions :-(
Often the case; be persuasive, explain the problem.
> "1x HTTPD and 5x Tomcat might be enough, but it's not going to depend on
> the server software, it's going to depend on your application and the
> hardware you're running on" - The Tomcat has an web application that
> actually inserts a record into the database for every https request and that
> all it does.
But what is being requested?
> "Your config suggests that you're running this on one server, how much
> memory is available and how many processors does it have?" - We have one
> linux machine with quad core processor and 8GB of RAM
Is it a 64bit machine? (If so, is the OS the 64bit version and if not,
why not?)
> Thanks for your time Ster...
It's Pid. Just Pid.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Apache 2.2.3(mod_proxy_ajp) - Tomcat 6.0.13 Loadbalancing -
error logs in apache
Posted by jkv <j....@gmail.com>.
Thanks for the reply Ster,
"Find someone who does.", sorry when I said privilege I actually meant
option, that is I cannot install a new Apache over the older 2.2.3 and we
have to go with the default configuration. Because they have chosen RHEL 5.4
as the platform and it comes with 2.2.3 version, I will see if I can
convince them!
"Why? RHEL has a built-in updater application doesn't it?" - I am actually
new to Linux environment and will try to find out more on this, is it
something like windows automatic software updater?
"If you can't upgrade HTTPD, how are you going to compile mod_jk?" - I just
had the thought that I need only the APR bundle and mod_jk src package to
compile and load it as a dynamic module! Does it need a HTTPD upgrade?
"Tying one hand behind your back isn't a good way to start." - Good though,
but I wish I had the authority to change decisions :-(
"1x HTTPD and 5x Tomcat might be enough, but it's not going to depend on
the server software, it's going to depend on your application and the
hardware you're running on" - The Tomcat has an web application that
actually inserts a record into the database for every https request and that
all it does.
"Your config suggests that you're running this on one server, how much
memory is available and how many processors does it have?" - We have one
linux machine with quad core processor and 8GB of RAM
Thanks for your time Ster...
--
View this message in context: http://old.nabble.com/Apache-2.2.3%28mod_proxy_ajp%29---Tomcat-6.0.13-Loadbalancing---error-logs-in-apache-tp27714229p27790893.html
Sent from the Tomcat - User mailing list archive at Nabble.com.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org
Re: Apache 2.2.3(mod_proxy_ajp) - Tomcat 6.0.13 Loadbalancing - error
logs in apache
Posted by Pid <pi...@pidster.com>.
On 04/03/2010 05:34, jkv wrote:
>
> Thanks for the reply Ster,
>
> But we don't have the privilege to upgrade Apache,
Find someone who does.
> because we are using Red
> Had Enterprise Linux and we have to go with the default httpd installation
> in it, i.e., 2.2.3,
Why? RHEL has a built-in updater application doesn't it?
> but is there a possibility for us to use mod_jk instead
> of mod_proxy for load balancing? I read mod_proxy was quite new when 2.2.3
> was released and is this issue related to mod_proxy rather than Apache
> Server as a whole? I definitely can think of upgrading Tomcat.
If you want to use mod_proxy, you definitely need to upgrade HTTPD.
If you can't upgrade HTTPD, how are you going to compile mod_jk?
> We have a eye popping requirement to handle 15000 concurrent https users
> simultaneously, an I am not sure a single Apache Server and five Tomcat
> instances (what we now have) can take this?
If that's your requirement then your should think long and hard about
how to tell the people setting the requirement that you need to upgrade
your software.
Tying one hand behind your back isn't a good way to start.
1x HTTPD and 5x Tomcat might be enough, but it's not going to depend on
the server software, it's going to depend on your application and the
hardware you're running on.
Your config suggests that you're running this on one server, how much
memory is available and how many processors does it have?
p
> Regards
>
>
>
> Pid Ster wrote:
>>
>> On 26/02/2010 06:36, jkv wrote:
>>>
>>> We are using the above setup to load balance http and https request, for
>>> https request
>>
>> Apache HTTPD 2.2.3 was released on 28 Jul 2006, you should definitely
>> upgrade to the latest version, there have been *many* important updates
>> since then.
>>
>> Tomcat 6.0.13 was released on 14 May 2007, you should definitely upgrade
>> to the latest version.
>>
>> Please let us know if the upgraded applications still display the same
>> problem.
>>
>>
>> p
>>
>>> Apache is configured to serve the certificates and the request is
>>> actually
>>> being processed
>>> by 3 tomcat instances (TomcatA, TomcatB, TomcatB)running behind. We are
>>> getting a strange log in apache
>>>
>>> [error] ajp_read_header: ajp_ilink_receive failed
>>> [error] (120006)APR does not understand this error code: proxy: read
>>> response failed from (null) (localhost)
>>>
>>> I have another question like, when Apache forwards a http/https request
>>> to
>>> tomcat and suppose that tomcat takes too long to respond! will the same
>>> request be routed to a different tomcat??
>>> As we have not configured sticky sessions in Apache, We are having many
>>> instances where in the java application in TomcatA takes too long (throws
>>> exception because it waits to connect to another host which takes too
>>> long)
>>> in responding back to Apache and I can see logs in other tomcats, say
>>> TomcatB and TomcatC with a session Id xxxxxxxxxxxxxxxxxxxxx.NodeA, with
>>> empty body which throws exceptions in my application level, why does this
>>> occur and can this be eliminated?? I am not sure that this occurs only
>>> for
>>> https request where we get empty body.
>>>
>>> Apache configurations are as follows
>>>
>>> <Proxy balancer://tomcatcluster>
>>> Order deny,allow
>>> Allow from all
>>> BalancerMember ajp://localhost:8109/test route=NodeA
>>> BalancerMember ajp://localhost:8209/test route=NodeB
>>> BalancerMember ajp://localhost:8309/test route=NodeC
>>> </Proxy>
>>>
>>> ProxyPass /test balancer://tomcatcluster lbmethod=byrequests
>>> nofailover=On
>>> ProxyPassReverse /test ajp://localhost:8109/test
>>> ProxyPassReverse /test ajp://localhost:8209/test
>>> ProxyPassReverse /test ajp://localhost:8309/test
>>>
>>> Thanks in advance.
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
>> For additional commands, e-mail: users-help@tomcat.apache.org
>>
>>
>>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org