8:44 a.m.
If I do a
cd jboss-5.0.0.GA/server/base/deploy
and
mv web.war web.war.bak
or
rm -fr web.war
Then httpd still shows /web as ENABLED. I assume that /web should get
DISABLED as soon as I move or remove it , correct ?
--
Bela Ban
Lead JGroups / Clustering Team
JBoss - a division of Red Hat
9:58 a.m.
Bela Ban wrote:
If I do a
cd jboss-5.0.0.GA/server/base/deploy
and
mv web.war web.war.bak
or
rm -fr web.war
Then httpd still shows /web as ENABLED. I assume that /web should get
DISABLED as soon as I move or remove it , correct ?
It should become STOPPED. Per your other private question to me, I'm
guessing the reason it's not is because the customized AS config you're
testing with doesn't have the hot deployment scanner running; I replied
privately about that. If that's not it, let us know.
--
Brian Stansberry
Lead, AS Clustering
JBoss, a division of Red Hat
brian.stansberry(a)redhat.com
10:11 a.m.
Thanks that was it...
Now to different issues:
#1
If I start cluster nodes 1 and 2, then create a session on (say) node2
and kill node2 (CTRL-C), I can see that the contents of JBossCache on
node1 have the session in memory.
However, when I do a reload of the web page, I don't fail over to node1,
but instead get an empty web page. If I hit reload long enough, there is
a failover ca 60 seconds later. The contents of the session are still
okay, so they were replicated fine.
So I assume httpd redirects me to node1 because its timeout (60 secs I
guess) removed node2.
However, when node2 goes down, it should tell httpd to remove node2 from
the cluster and not direct requests to it !
This is with Beta2.
#2 "localhost:8000/status" doesn't work anymore, I get 404s. This works
with the 'all' config, modified for mod-lcuster ! The weird thing is
that I don't even restart httpd, but simply start a different JBoss
config ! Is the 'status' application dependent on the JBoss AS side ?
If I do a telnet localhost 8000, then info and <return>, I do get the
status information...
#3 When I restart httpd, all hell breaks loose and the JBossAS instances
don't seem to be able to reconnect to httpd.
Should I create JIRA issues for these problems, with steps to reproduce ?
Brian Stansberry wrote:
Bela Ban wrote:
> If I do a
>
> cd jboss-5.0.0.GA/server/base/deploy
>
> and
>
> mv web.war web.war.bak
>
> or
>
> rm -fr web.war
>
> Then httpd still shows /web as ENABLED. I assume that /web should get
> DISABLED as soon as I move or remove it , correct ?
>
It should become STOPPED. Per your other private question to me, I'm
guessing the reason it's not is because the customized AS config
you're testing with doesn't have the hot deployment scanner running; I
replied privately about that. If that's not it, let us know.
--
Bela Ban
Lead JGroups / Clustering Team
JBoss - a division of Red Hat
11:40 a.m.
Bela Ban wrote:
Thanks that was it...
Now to different issues:
#1
If I start cluster nodes 1 and 2, then create a session on (say) node2
and kill node2 (CTRL-C), I can see that the contents of JBossCache on
node1 have the session in memory.
However, when I do a reload of the web page, I don't fail over to node1,
but instead get an empty web page.
Is this empty web page fully loaded? Or is it still loading while you
hit refresh?
If I hit reload long enough, there is
a failover ca 60 seconds later. The contents of the session are still
okay, so they were replicated fine.
So I assume httpd redirects me to node1 because its timeout (60 secs I
guess) removed node2.
Maybe. Clearly something changed after 60 secs. But if you actually got
a fully loaded empty web page in the meantime, something more
complex/strange is going on, since something had to return that page to you.
However, when node2 goes down, it should tell httpd to remove node2
from
the cluster and not direct requests to it !
This is with Beta2.
#2 "localhost:8000/status" doesn't work anymore, I get 404s. This works
with the 'all' config, modified for mod-lcuster ! The weird thing is
that I don't even restart httpd, but simply start a different JBoss
config ! Is the 'status' application dependent on the JBoss AS side ?
Does http://localhost:8000/status/ work? Note the trailing slash.
Remembering that detail is an old bit of mod_jk black art that might
have carried over.
If I do a telnet localhost 8000, then info and <return>, I do
get the
status information...
#3 When I restart httpd, all hell breaks loose and the JBossAS instances
don't seem to be able to reconnect to httpd.
Should I create JIRA issues for these problems, with steps to reproduce ?
Just checked with Paul and he says go for it.
Brian Stansberry wrote:
> Bela Ban wrote:
>> If I do a
>>
>> cd jboss-5.0.0.GA/server/base/deploy
>>
>> and
>>
>> mv web.war web.war.bak
>>
>> or
>>
>> rm -fr web.war
>>
>> Then httpd still shows /web as ENABLED. I assume that /web should get
>> DISABLED as soon as I move or remove it , correct ?
>>
>
> It should become STOPPED. Per your other private question to me, I'm
> guessing the reason it's not is because the customized AS config
> you're testing with doesn't have the hot deployment scanner running; I
> replied privately about that. If that's not it, let us know.
>
--
Brian Stansberry
Lead, AS Clustering
JBoss, a division of Red Hat
brian.stansberry(a)redhat.com
3:38 p.m.
Brian Stansberry wrote:
Bela Ban wrote:
> Thanks that was it...
>
> Now to different issues:
>
>
> #1
> If I start cluster nodes 1 and 2, then create a session on (say)
> node2 and kill node2 (CTRL-C), I can see that the contents of
> JBossCache on node1 have the session in memory.
>
> However, when I do a reload of the web page, I don't fail over to
> node1, but instead get an empty web page.
Is this empty web page fully loaded? Or is it still loading while you
hit refresh?
Fully loaded. When I do a telnet localhost 8000 and send a GET for /web
HTTP/1.0
I get a 'close' response...
> #2 "localhost:8000/status" doesn't work anymore, I
get 404s. This
> works with the 'all' config, modified for mod-lcuster ! The weird
> thing is that I don't even restart httpd, but simply start a
> different JBoss config ! Is the 'status' application dependent on the
> JBoss AS side ?
>
Does http://localhost:8000/status/ work? Note the trailing slash.
I tried both, same thing: nada
> Should I create JIRA issues for these problems, with steps to
> reproduce ?
>
Just checked with Paul and he says go for it.
I downloaded all components again, with fresh configs and will try to
reproduce these issues tomorrow.
Cheers,
--
Bela Ban
Lead JGroups / Clustering Team
JBoss - a division of Red Hat
5843
days inactive
5843
days old
mod_cluster-dev@lists.jboss.org
4 comments
2 participants
participants (2)
-
Bela Ban -
Brian Stansberry