7:04 a.m.
Hi Manik,
These two issue seem to be duplicate, right? I do not understand how is
it possible to set node name from transport element when we have to set
a different name for each channel created in order for this feature to
have any sense?
If you want this feature maybe we need channelNamingSchemaClass
attribute in transport element. Instance of this class will conform to
certain interface and will be consulted to generate the next in sequence
channel name for the created channel. Is this what you meant?
Let me know.
https://jira.jboss.org/jira/browse/ISPN-182
https://jira.jboss.org/jira/browse/ISPN-181
9:57 a.m.
On Sep 10, 2009, at 3:04 PM, Vladimir Blagojevic wrote:
Hi Manik,
These two issue seem to be duplicate, right?
Not that it's me being asked, but
they seem to be exactly the same
thing :)
I do not understand how is
it possible to set node name from transport element when we have to
set
a different name for each channel created in order for this feature to
have any sense?
infinispan creates one channel only for all cache caches associated
to
a CacheManager.
it would be nice to be able pass it a name through channel.setName,
for debugging reasons only.
If you want this feature maybe we need channelNamingSchemaClass
attribute in transport element. Instance of this class will conform to
certain interface and will be consulted to generate the next in
sequence
channel name for the created channel. Is this what you meant?
I'm not sure what you mean with "next in sequence channel name".
Let me know.
https://jira.jboss.org/jira/browse/ISPN-182
https://jira.jboss.org/jira/browse/ISPN-181
_______________________________________________
infinispan-dev mailing list
infinispan-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/infinispan-dev
10:21 a.m.
On 09-09-10 10:57 AM, Mircea Markus wrote:
infinispan creates one channel only for all cache caches associated
to
a CacheManager.
it would be nice to be able pass it a name through channel.setName,
for debugging reasons only.
I see. So two CacheManagers might use two different configuration files
where transport names are different but they usually use the same
configuration files, don't they? So what's the point of the name in that
case?
I mentioned generating next in sequence name as a way to differentiate
each channel instance that belongs to the same cluster. Each channel
would have a unique name (A, B, C...)
I admit - I am confused!
12:41 p.m.
On Sep 10, 2009, at 6:21 PM, Vladimir Blagojevic wrote:
On 09-09-10 10:57 AM, Mircea Markus wrote:
> infinispan creates one channel only for all cache caches associated
> to
> a CacheManager.
> it would be nice to be able pass it a name through channel.setName,
> for debugging reasons only.
>
I see. So two CacheManagers might use two different configuration
files
where transport names are different but they usually use the same
configuration files, don't they?
Good point. What we can do is use a
placeholder ${nodeName:default}
and if one specifies it through -DnodeName will be taken into account.
It's much easier going through logs and looking for nodes named:
"aaaa", "bbb" and "ccc" rather than host:port1, host2:port2
etc.
Actually we even have a script that does that, so that we can debug
logs more easily :)
So what's the point of the name in that
case?
I mentioned generating next in sequence name as a way to differentiate
each channel instance that belongs to the same cluster. Each channel
would have a unique name (A, B, C...)
I admit - I am confused!
lol!
_______________________________________________
infinispan-dev mailing list
infinispan-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/infinispan-dev
2:29 p.m.
On 09-09-10 1:41 PM, Mircea Markus wrote:
Good point. What we can do is use a placeholder ${nodeName:default}
and if one specifies it through -DnodeName will be taken into account.
It's much easier going through logs and looking for nodes named:
"aaaa", "bbb" and "ccc" rather than host:port1, host2:port2
etc.
Actually we even have a script that does that, so that we can debug
logs more easily :)
Yes, but as soon as you have two cache managers in the same VM you are
screwed, right? We would end up with two channels with the same name.
Everything would work as far as JGroups is concerned but logs would be
useless.
Don't forget that channel names need to be unique per JVM and across JVMs!
The only solution I can think of is to add API that allows naming of
CacheManagers. That name could be reused to name channel as well.
1:53 a.m.
If you don't define a logical name, JGroups will pick one for you. The
name will look like {hostname}-{randomShort}, e.g. mybox-12345.
I recommend a good logical name should include
- the function (e.g. "SessionReplication")
- the location (e.g. "mybox")
- if you have multiple instances running on the same box, maybe an ID
Example: SessionCache-mybox-33
Vladimir Blagojevic wrote:
On 09-09-10 1:41 PM, Mircea Markus wrote:
> Good point. What we can do is use a placeholder ${nodeName:default}
> and if one specifies it through -DnodeName will be taken into account.
> It's much easier going through logs and looking for nodes named:
> "aaaa", "bbb" and "ccc" rather than host:port1,
host2:port2 etc.
> Actually we even have a script that does that, so that we can debug
> logs more easily :)
>
>
Yes, but as soon as you have two cache managers in the same VM you are
screwed, right? We would end up with two channels with the same name.
Everything would work as far as JGroups is concerned but logs would be
useless.
Don't forget that channel names need to be unique per JVM and across JVMs!
The only solution I can think of is to add API that allows naming of
CacheManagers. That name could be reused to name channel as well.
--
Bela Ban
Lead JGroups / Clustering Team
JBoss
12:41 p.m.
On 17 Sep 2009, at 07:53, Bela Ban wrote:
If you don't define a logical name, JGroups will pick one for
you. The
name will look like {hostname}-{randomShort}, e.g. mybox-12345.
I recommend a good logical name should include
- the function (e.g. "SessionReplication")
- the location (e.g. "mybox")
- if you have multiple instances running on the same box, maybe an ID
Example: SessionCache-mybox-33
+1, this would be perfect!
--
Manik Surtani
manik(a)jboss.org
Lead, Infinispan
Lead, JBoss Cache
http://www.infinispan.org
http://www.jbosscache.org
3:40 a.m.
On Sep 16, 2009, at 10:29 PM, Vladimir Blagojevic wrote:
On 09-09-10 1:41 PM, Mircea Markus wrote:
> Good point. What we can do is use a placeholder ${nodeName:default}
> and if one specifies it through -DnodeName will be taken into
> account.
> It's much easier going through logs and looking for nodes named:
> "aaaa", "bbb" and "ccc" rather than host:port1,
host2:port2 etc.
> Actually we even have a script that does that, so that we can debug
> logs more easily :)
>
>
Yes, but as soon as you have two cache managers in the same VM you are
screwed, right?
Two cache managers means two different global configs, i.e. the
possibility to give different names.
If for whatever reasons, the user needs to use exactly the same config
file for two CacheManagers within the same JVM than he won't be able
to use this feature, and and channel instance will default to
host:port, as per jgroup's default. It's up to the user to make sure
the node names are unique. Even more, if he doesn't want to use this
feature he doesn't need to, as the nodeName (I think is rather about
the name of the node rather than name of the channel) attribute (or
however will it be named) would be optional.
We would end up with two channels with the same name.
Everything would work as far as JGroups is concerned but logs would be
useless.
Don't forget that channel names need to be unique per JVM and across
JVMs!
The only solution I can think of is to add API that allows naming of
CacheManagers. That name could be reused to name channel as well.
_______________________________________________
infinispan-dev mailing list
infinispan-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/infinispan-dev
12:39 p.m.
Yeah they're dupes. I believe JIRA crashed on my when I was creating
these. I have closed one now.
On 10 Sep 2009, at 13:04, Vladimir Blagojevic wrote:
Hi Manik,
These two issue seem to be duplicate, right? I do not understand how
is
it possible to set node name from transport element when we have to
set
a different name for each channel created in order for this feature to
have any sense?
If you want this feature maybe we need channelNamingSchemaClass
attribute in transport element. Instance of this class will conform to
certain interface and will be consulted to generate the next in
sequence
channel name for the created channel. Is this what you meant?
Not really a sequence, but, for example, rather than using UUIDs in
the logs, etc. (which is what happens now), I should be able to name
my nodes. E.g.,
<transport nodeName="ByBigFatSolarisBox">
<transport nodeName="Fedora11OnMyLaptop">
<transport nodeName="TheOldUbuntuBoxUnderMySofa">
and I believe calling channel.setName() with this string (if present)
will work?
Let me know.
https://jira.jboss.org/jira/browse/ISPN-182
https://jira.jboss.org/jira/browse/ISPN-181
_______________________________________________
infinispan-dev mailing list
infinispan-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/infinispan-dev
--
Manik Surtani
manik(a)jboss.org
Lead, Infinispan
Lead, JBoss Cache
http://www.infinispan.org
http://www.jbosscache.org
5855
days inactive
5868
days old
infinispan-dev@lists.jboss.org
8 comments
4 participants
participants (4)
-
Bela Ban -
Manik Surtani -
Mircea Markus -
Vladimir Blagojevic