RE: [JBoss-dev] RE: Clustering tasks - TreeCache
added ejb cache policy and ha stateful session EJB as a specific example. --- Scott M Stark <[EMAIL PROTECTED]> wrote: > The ejb container will potentially use TreeCache as > its cache policy implementation. This should be the > default cache policy in the near future. > > > Scott Stark > Chief Technology Officer > JBoss Group, LLC > > > > > From: [EMAIL PROTECTED] > on behalf of Ivelin Ivanov > Sent: Mon 12/22/2003 9:43 AM > To: [EMAIL PROTECTED] > Subject: [JBoss-dev] RE: Clustering tasks - > TreeCache > > > > > Can we enumerate the services that are expected to > leverage TreeCache in 4.0? > > > - TreeCache integration for all services (we > should > keep the old implementation though) > > So far I know of: > - Hibernate > - JMS > - HA HttpSession > - HA JNDI > - HA Scheduler > > What else? > > I assume EJB CMP 2 and JDO will not directly use > TreeCache. They will be based on Hibernate, which in > turn will use TreeCache. > > > Ivelin > > > --- Sacha Labourey <[EMAIL PROTECTED]> wrote: > > With more details than my previous e-mail, > > clustering todo list for 4.0: > > - integration with the InvocationResponse from > 4.0 > > (no more HAResponse) > > - HA-Invoker should be an interceptor independent > > from the transport layer > > - invoker downloaded in the HA interceptor should > > be wrapped in objects > > that contain at least the logical name of the > target > > (to be able to keep > > using a target when we receive a new node for > > example) and may contain > > arbitrary info (to have payload-based > load-balancing > > for example) > > - we should have pluggable "sub-partitioning" > > detection algorithm that > > would suspend (stand-by) a node if it is in a > > minority sub-partition > > - add metadata management information such as > what > > "should be" the topology > > of a cluster so that we can say things such as "3 > > nodes are missing", today > > we have a fully dynamic approach and don't know > what > > should be the > > ideal/expected cluster topology > > - TreeCache integration for all services (we > should > > keep the old > > implementation though) > > - Cold-backup stand-by node implementation > > - fully clustered JBossMQ > > - sexy httpsession replication schemes > > - improved Jgroups debugging message/info > (strange > > situations should be > > logged with some kind of warnings) > > > > Note that the current InvocationResponse > > implementation does not allow to > > send back information if the result of the > > invocation is an exception, this > > is a bad limitation IMHO, but it makes code much > > easier to read though. > > > > Cheers, > > > > > > sacha > > > > > > > > > -Original Message- > > > From: Ivelin Ivanov [mailto:[EMAIL PROTECTED] > > > Sent: jeudi, 11. décembre 2003 21:59 > > > To: [EMAIL PROTECTED] > > > Cc: Bela Ban > > > Subject: Re: Clustering tasks > > > > > > > > > Sacha, > > > > > > do you mind if I delete the old Clustering > tasks? > > > > > > They have not been addressed for a year now. > > > I would like to collect more information from > you > > and > > > Bela, so that we can put together a detailed > > > specification. We can publish that and break it > > into > > > small tasks that can be tracked easier. > > > > > > I am already working with Scott and Bill on the > > 4.0 > > > roadmap document and with Adrian on the JMS > > functional > > > spec. > > > > > > Bill just gave me edit rights for the web site > and > > > tasks, so I can try to clean up the projects > > planning > > > pages and the pending tasks. > > > > > > Best, > > > > > > Ivelin > > > > > > > > > --- Bela Ban <[EMAIL PROTECTED]> wrote: > > > > sure, but ask Sacha; he's in charge of > > Clustering > > > > > > > > Ivelin Ivanov wrote: > > > > > > > > >Bela, > > > > > > > > > >I am trying to help with the project > > management. > > > > > > > > > >There are tasks on SF Clustering sub-project, > > which > > > > >were open in January and have no follow-up, > nor > > > > >reference to relevant documentation. > > > > > > > > > >If you don't mind I would like to delete them > > to > > > > >reduce the noise on the tasks list. > > > > > > > > > >Ivelin > > > > > > > > > > > > > > > > > > > > > > > -- > > > > Bela Ban > > > > http://www.jgroups.org > > > > Cell: (408) 316-4459 > > > > > > > > > > > > > > > > > > > > --- > This SF.net email is sponsored by: IBM Linux > Tutorials. > Become an expert in LINUX or just sharpen your > skills. Sign up for IBM's > Free Linux Tutorials. Learn everything from the > bash shell to sys admin. > Click now! > http://ads.osdn.com/?ad_id=1278&alloc_id=3371&op=click > ___ > JBoss-Development mailing list > [EMAIL PROTECTED] > https://lists.sourceforge.net/lists/listinfo/jboss-development > > > > ATTACHMENT part 2 application/ms-tnef
RE: [JBoss-dev] RE: Clustering tasks - TreeCache
HAJNDI could use it rather than the current mechanism of subclassing the non-clustered naming service with replication via direct use of the cluster rpc mechanism. Its something to consider. Scott Stark Chief Technology Officer JBoss Group, LLC From: [EMAIL PROTECTED] on behalf of Adrian Brock Sent: Mon 12/22/2003 10:04 AM To: [EMAIL PROTECTED] Subject: Re: [JBoss-dev] RE: Clustering tasks - TreeCache On Mon, 2003-12-22 at 17:43, Ivelin Ivanov wrote: > Can we enumerate the services that are expected to > leverage TreeCache in 4.0? > > > - TreeCache integration for all services (we should > keep the old implementation though) > > So far I know of: > - Hibernate > - JMS > - HA HttpSession > - HA JNDI > - HA Scheduler > > What else? > HA StatefulSessionBean I don't think HAJNDI will use it? It does not replicate entries, only the participant nodes. Regards, Adrian <>
Re: [JBoss-dev] RE: Clustering tasks - TreeCache
On Mon, 2003-12-22 at 17:43, Ivelin Ivanov wrote: > Can we enumerate the services that are expected to > leverage TreeCache in 4.0? > > > - TreeCache integration for all services (we should > keep the old implementation though) > > So far I know of: > - Hibernate > - JMS > - HA HttpSession > - HA JNDI > - HA Scheduler > > What else? > HA StatefulSessionBean I don't think HAJNDI will use it? It does not replicate entries, only the participant nodes. Regards, Adrian > I assume EJB CMP 2 and JDO will not directly use > TreeCache. They will be based on Hibernate, which in > turn will use TreeCache. > > > Ivelin > > > --- Sacha Labourey <[EMAIL PROTECTED]> wrote: > > With more details than my previous e-mail, > > clustering todo list for 4.0: > > - integration with the InvocationResponse from 4.0 > > (no more HAResponse) > > - HA-Invoker should be an interceptor independent > > from the transport layer > > - invoker downloaded in the HA interceptor should > > be wrapped in objects > > that contain at least the logical name of the target > > (to be able to keep > > using a target when we receive a new node for > > example) and may contain > > arbitrary info (to have payload-based load-balancing > > for example) > > - we should have pluggable "sub-partitioning" > > detection algorithm that > > would suspend (stand-by) a node if it is in a > > minority sub-partition > > - add metadata management information such as what > > "should be" the topology > > of a cluster so that we can say things such as "3 > > nodes are missing", today > > we have a fully dynamic approach and don't know what > > should be the > > ideal/expected cluster topology > > - TreeCache integration for all services (we should > > keep the old > > implementation though) > > - Cold-backup stand-by node implementation > > - fully clustered JBossMQ > > - sexy httpsession replication schemes > > - improved Jgroups debugging message/info (strange > > situations should be > > logged with some kind of warnings) > > > > Note that the current InvocationResponse > > implementation does not allow to > > send back information if the result of the > > invocation is an exception, this > > is a bad limitation IMHO, but it makes code much > > easier to read though. > > > > Cheers, > > > > > > sacha > > > > > > > > > -Original Message- > > > From: Ivelin Ivanov [mailto:[EMAIL PROTECTED] > > > Sent: jeudi, 11. dÃcembre 2003 21:59 > > > To: [EMAIL PROTECTED] > > > Cc: Bela Ban > > > Subject: Re: Clustering tasks > > > > > > > > > Sacha, > > > > > > do you mind if I delete the old Clustering tasks? > > > > > > They have not been addressed for a year now. > > > I would like to collect more information from you > > and > > > Bela, so that we can put together a detailed > > > specification. We can publish that and break it > > into > > > small tasks that can be tracked easier. > > > > > > I am already working with Scott and Bill on the > > 4.0 > > > roadmap document and with Adrian on the JMS > > functional > > > spec. > > > > > > Bill just gave me edit rights for the web site and > > > tasks, so I can try to clean up the projects > > planning > > > pages and the pending tasks. > > > > > > Best, > > > > > > Ivelin > > > > > > > > > --- Bela Ban <[EMAIL PROTECTED]> wrote: > > > > sure, but ask Sacha; he's in charge of > > Clustering > > > > > > > > Ivelin Ivanov wrote: > > > > > > > > >Bela, > > > > > > > > > >I am trying to help with the project > > management. > > > > > > > > > >There are tasks on SF Clustering sub-project, > > which > > > > >were open in January and have no follow-up, nor > > > > >reference to relevant documentation. > > > > > > > > > >If you don't mind I would like to delete them > > to > > > > >reduce the noise on the tasks list. > > > > > > > > > >Ivelin > > > > > > > > > > > > > > > > > > > > > > > -- > > > > Bela Ban > > > > http://www.jgroups.org > > > > Cell: (408) 316-4459 > > > > > > > > > > > > > > > > > > > > --- > This SF.net email is sponsored by: IBM Linux Tutorials. > Become an expert in LINUX or just sharpen your skills. Sign up for IBM's > Free Linux Tutorials. Learn everything from the bash shell to sys admin. > Click now! http://ads.osdn.com/?ad_id=1278&alloc_id=3371&op=click > ___ > JBoss-Development mailing list > [EMAIL PROTECTED] > https://lists.sourceforge.net/lists/listinfo/jboss-development -- Adrian Brock Director of Support Back Office JBoss Group, LLC --- This SF.net email is sponsored by: IBM Linux Tutorials. Become an expert in LINUX or just sharpen your skills. Sign up for IBM's Free Linux Tutorials. Learn everything from the bash shell to sys admin. Click now! http://ads.osdn.com/?ad_id78&alloc_id371&op=click __
RE: [JBoss-dev] RE: Clustering tasks - TreeCache
The ejb container will potentially use TreeCache as its cache policy implementation. This should be the default cache policy in the near future. Scott Stark Chief Technology Officer JBoss Group, LLC From: [EMAIL PROTECTED] on behalf of Ivelin Ivanov Sent: Mon 12/22/2003 9:43 AM To: [EMAIL PROTECTED] Subject: [JBoss-dev] RE: Clustering tasks - TreeCache Can we enumerate the services that are expected to leverage TreeCache in 4.0? > - TreeCache integration for all services (we should keep the old implementation though) So far I know of: - Hibernate - JMS - HA HttpSession - HA JNDI - HA Scheduler What else? I assume EJB CMP 2 and JDO will not directly use TreeCache. They will be based on Hibernate, which in turn will use TreeCache. Ivelin --- Sacha Labourey <[EMAIL PROTECTED]> wrote: > With more details than my previous e-mail, > clustering todo list for 4.0: > - integration with the InvocationResponse from 4.0 > (no more HAResponse) > - HA-Invoker should be an interceptor independent > from the transport layer > - invoker downloaded in the HA interceptor should > be wrapped in objects > that contain at least the logical name of the target > (to be able to keep > using a target when we receive a new node for > example) and may contain > arbitrary info (to have payload-based load-balancing > for example) > - we should have pluggable "sub-partitioning" > detection algorithm that > would suspend (stand-by) a node if it is in a > minority sub-partition > - add metadata management information such as what > "should be" the topology > of a cluster so that we can say things such as "3 > nodes are missing", today > we have a fully dynamic approach and don't know what > should be the > ideal/expected cluster topology > - TreeCache integration for all services (we should > keep the old > implementation though) > - Cold-backup stand-by node implementation > - fully clustered JBossMQ > - sexy httpsession replication schemes > - improved Jgroups debugging message/info (strange > situations should be > logged with some kind of warnings) > > Note that the current InvocationResponse > implementation does not allow to > send back information if the result of the > invocation is an exception, this > is a bad limitation IMHO, but it makes code much > easier to read though. > > Cheers, > > > sacha > > > > > -Original Message- > > From: Ivelin Ivanov [mailto:[EMAIL PROTECTED] > > Sent: jeudi, 11. décembre 2003 21:59 > > To: [EMAIL PROTECTED] > > Cc: Bela Ban > > Subject: Re: Clustering tasks > > > > > > Sacha, > > > > do you mind if I delete the old Clustering tasks? > > > > They have not been addressed for a year now. > > I would like to collect more information from you > and > > Bela, so that we can put together a detailed > > specification. We can publish that and break it > into > > small tasks that can be tracked easier. > > > > I am already working with Scott and Bill on the > 4.0 > > roadmap document and with Adrian on the JMS > functional > > spec. > > > > Bill just gave me edit rights for the web site and > > tasks, so I can try to clean up the projects > planning > > pages and the pending tasks. > > > > Best, > > > > Ivelin > > > > > > --- Bela Ban <[EMAIL PROTECTED]> wrote: > > > sure, but ask Sacha; he's in charge of > Clustering > > > > > > Ivelin Ivanov wrote: > > > > > > >Bela, > > > > > > > >I am trying to help with the project > management. > > > > > > > >There are tasks on SF Clustering sub-project, > which > > > >were open in January and have no follow-up, nor > > > >reference to relevant documentation. > > > > > > > >If you don't mind I would like to delete them > to > > > >reduce the noise on the tasks list. > > > > > > > >Ivelin > > > > > > > > > > > > > > > > > > -- > > > Bela Ban > > > http://www.jgroups.org > > > Cell: (408) 316-4459 > > > > > > > > > > > --- This SF.net email is sponsored by: IBM Linux Tutorials. Become an expert in LINUX or just sharpen your skills. Sign up for IBM's Free Linux Tutorials. Learn everything from the bash shell to sys admin. Click now! http://ads.osdn.com/?ad_id=1278&alloc_id=3371&op=click ___ JBoss-Development mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/jboss-development <>
RE: [JBoss-dev] RE: Clustering tasks
Some of them are already there, some not, but in 4.0 they should all be there with easy configuration syntax: - asych vs synch - immediate vs time-based - granularity: session, attribute, AOP-liked (field-based) - transparent (or pre-defined) sub-cluster (brother-nodes) replication (when you have 5 nodes, you don't want each node to replicate to the four others) - base code is there, integration is missing - state startup: when a new node starts, should it pre-replicate the state of its brother-node or not - cluster-wide, http-concurrent-requests-for-same-session handling scheme. > -Original Message- > From: [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] On > Behalf Of Ivelin Ivanov > Sent: lundi, 22. décembre 2003 05:22 > To: [EMAIL PROTECTED]; [EMAIL PROTECTED] > Subject: [JBoss-dev] RE: Clustering tasks > > > Can Sacha or somebody explain what the various http > replication schemes are ? > > >Sacha: - sexy httpsession replication schemes > > Ivelin > > > > --- > This SF.net email is sponsored by: IBM Linux Tutorials. > Become an expert in LINUX or just sharpen your skills. Sign > up for IBM's > Free Linux Tutorials. Learn everything from the bash shell > to sys admin. > Click now! http://ads.osdn.com/?ad_id=1278&alloc_id=3371&op=click > ___ > JBoss-Development mailing list > [EMAIL PROTECTED] > https://lists.sourceforge.net/lists/listinfo/jboss-development > > --- This SF.net email is sponsored by: IBM Linux Tutorials. Become an expert in LINUX or just sharpen your skills. Sign up for IBM's Free Linux Tutorials. Learn everything from the bash shell to sys admin. Click now! http://ads.osdn.com/?ad_id78&alloc_id371&op=click ___ JBoss-Development mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/jboss-development