You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@wicket.apache.org by Ravi <ra...@intellinet.de> on 2018/03/02 12:43:41 UTC
Wicket in a wildfly cluster
Hi all,
we have the following scenario;
Our single-page application is a 100% used for application forms and makes
heavy use of ajax calls (validation, visibility, etc).
There might even be an ajax call for every blur on an input field.
We're trying to put this application in a wildfly cluster that replicates
the session & pagestore through infinispan.
It seemes that the pagestore is not replicated fast enough(?), so that we
get a ComponentNotFoundException.
After playing around with the wicket/infinispan configuration our setup
looks like this:
Wicket:
- HttpSessionDataStore
- getStoreSettings().setAsynchronous(false);
- getStoreSettings().setInmemoryCacheSize(0);
-
getRequestCycleSettings().setRenderStrategy(RenderStrategy.ONE_PASS_RENDER);
JBoss:
- cluster mode (ha profile)
Infinispan
- Distributed cache with default config
Although this configuration seems to work now and then, it still breaks
quite often and is very fragile.
Since no one here has a lot experience with clustering, maybe someone here
has any theoretical/practical experience and is willing to share :)
Needles to say that we googled the web and forum ;)
Thank you,
Ravi
--
Sent from: http://apache-wicket.1842946.n4.nabble.com/Users-forum-f1842947.html
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
For additional commands, e-mail: users-help@wicket.apache.org
Re: Wicket in a wildfly cluster
Posted by sh...@gmail.com.
Dear Ravi
Please let me know the loading test result, cause I built identity provider with this cluster mode, and I have to check whether it exists the same problems.
Thanks
> On 8 Mar 2018, at 20:15, Ravi <ra...@intellinet.de> wrote:
>
> Hi Shengche,
>
> sorry for the late reply. The replication as such is working. We do run into
> what seems to be a timing factor.
>
> On a very small scale it works now with the pagestore on a shared
> filesystem. We'll do some loadtesting within the next days.
>
> -Ravi
>
> --
> Sent from: http://apache-wicket.1842946.n4.nabble.com/Users-forum-f1842947.html
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
> For additional commands, e-mail: users-help@wicket.apache.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
For additional commands, e-mail: users-help@wicket.apache.org
Re: Wicket in a wildfly cluster
Posted by Ravi <ra...@intellinet.de>.
Hi Shengche,
sorry for the late reply. The replication as such is working. We do run into
what seems to be a timing factor.
On a very small scale it works now with the pagestore on a shared
filesystem. We'll do some loadtesting within the next days.
-Ravi
--
Sent from: http://apache-wicket.1842946.n4.nabble.com/Users-forum-f1842947.html
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
For additional commands, e-mail: users-help@wicket.apache.org
Re: Wicket in a wildfly cluster
Posted by sh...@gmail.com.
Ravi
Another question, does your apache https using mod_cluster package from jboss ?
Or have you check the replication packets working status via port 6666?
> On 5 Mar 2018, at 19:09, Ravi <ra...@intellinet.de> wrote:
>
> Hi,
>
> wildfly runs in domain mode with default settings for the distributed cache.
>
> It seems that the problem is not the jboss settings but somehow the
> pagestore is either not written fast enough or concurrently.
>
>
>
> --
> Sent from: http://apache-wicket.1842946.n4.nabble.com/Users-forum-f1842947.html
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
> For additional commands, e-mail: users-help@wicket.apache.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
For additional commands, e-mail: users-help@wicket.apache.org
Re: Wicket in a wildfly cluster
Posted by Ravi <ra...@intellinet.de>.
Hi,
wildfly runs in domain mode with default settings for the distributed cache.
It seems that the problem is not the jboss settings but somehow the
pagestore is either not written fast enough or concurrently.
--
Sent from: http://apache-wicket.1842946.n4.nabble.com/Users-forum-f1842947.html
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
For additional commands, e-mail: users-help@wicket.apache.org
Re: Wicket in a wildfly cluster
Posted by sh...@gmail.com.
Hello
Dose your wildfly runs at standalone mode ? or domain mode ? any adjustment of session replicating?
I add boss-web.xml as flowing
<?xml version="1.0" encoding="UTF-8"?>
<jboss-web xmlns="http://www.jboss.com/xml/ns/javaee"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="
http://www.jboss.com/xml/ns/javaee
http://www.jboss.org/j2ee/schema/jboss-web_5_1.xsd">
<context-root>/</context-root>
<replication-config>
<replication-trigger>SET_AND_GET</replication-trigger>
<replication-granularity>SESSION</replication-granularity>
</replication-config>
</jboss-web>
> On 2 Mar 2018, at 22:05, Ravi <ra...@intellinet.de> wrote:
>
> We also tried to put the pagestore on a shared filesystem, with a similar
> result.
> The session itself is only a few kb.
>
> Thanks for taking the time!
>
> --
> Sent from: http://apache-wicket.1842946.n4.nabble.com/Users-forum-f1842947.html
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
> For additional commands, e-mail: users-help@wicket.apache.org
>
Re: Wicket in a wildfly cluster
Posted by Ravi <ra...@intellinet.de>.
We also tried to put the pagestore on a shared filesystem, with a similar
result.
The session itself is only a few kb.
Thanks for taking the time!
--
Sent from: http://apache-wicket.1842946.n4.nabble.com/Users-forum-f1842947.html
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
For additional commands, e-mail: users-help@wicket.apache.org
Re: Wicket in a wildfly cluster
Posted by Bas Gooren <ba...@iswd.nl>.
Hi Ravi,
Sorry - I have no experience with "real-time" session replication.
Personally I would never attempt it, as wicket sessions can grow relatively
large and thus replication may not be instantaneous (as you are
experiencing)…
Met vriendelijke groet,
Kind regards,
Bas Gooren
Op 2 maart 2018 bij 14:40:50, Ravi (ravi.knox@intellinet.de) schreef:
Hi Bas,
we have an apache that balances by using a round robin (byrequest).
The application runs in an infrastructure that does not support
sticky-sessions.
So far from what we've seen both servers getting pinged withing
milliseconds. One server responds correctly. The other one throws a
ComponentNotFoundException.
But it is not deterministic, meaning it does work at times.
Ravi
--
Sent from:
http://apache-wicket.1842946.n4.nabble.com/Users-forum-f1842947.html
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
For additional commands, e-mail: users-help@wicket.apache.org
Re: Wicket in a wildfly cluster
Posted by Ravi <ra...@intellinet.de>.
Hi Bas,
we have an apache that balances by using a round robin (byrequest).
The application runs in an infrastructure that does not support
sticky-sessions.
So far from what we've seen both servers getting pinged withing
milliseconds. One server responds correctly. The other one throws a
ComponentNotFoundException.
But it is not deterministic, meaning it does work at times.
Ravi
--
Sent from: http://apache-wicket.1842946.n4.nabble.com/Users-forum-f1842947.html
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
For additional commands, e-mail: users-help@wicket.apache.org
Re: Wicket in a wildfly cluster
Posted by Bas Gooren <ba...@iswd.nl>.
Hi Ravi,
You don’t mention what’s in front of your application servers.
Are you doing non-sticky load balancing?
Or are you having issues when the user hits the same application server?
(Which would be a different issue)
Met vriendelijke groet,
Kind regards,
Bas Gooren
Op 2 maart 2018 bij 13:43:49, Ravi (ravi.knox@intellinet.de) schreef:
Hi all,
we have the following scenario;
Our single-page application is a 100% used for application forms and makes
heavy use of ajax calls (validation, visibility, etc).
There might even be an ajax call for every blur on an input field.
We're trying to put this application in a wildfly cluster that replicates
the session & pagestore through infinispan.
It seemes that the pagestore is not replicated fast enough(?), so that we
get a ComponentNotFoundException.
After playing around with the wicket/infinispan configuration our setup
looks like this:
Wicket:
- HttpSessionDataStore
- getStoreSettings().setAsynchronous(false);
- getStoreSettings().setInmemoryCacheSize(0);
-
getRequestCycleSettings().setRenderStrategy(RenderStrategy.ONE_PASS_RENDER);
JBoss:
- cluster mode (ha profile)
Infinispan
- Distributed cache with default config
Although this configuration seems to work now and then, it still breaks
quite often and is very fragile.
Since no one here has a lot experience with clustering, maybe someone here
has any theoretical/practical experience and is willing to share :)
Needles to say that we googled the web and forum ;)
Thank you,
Ravi
--
Sent from:
http://apache-wicket.1842946.n4.nabble.com/Users-forum-f1842947.html
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@wicket.apache.org
For additional commands, e-mail: users-help@wicket.apache.org