There is a PR[1] open for hawkular-parent that upgrades Cassandra from 2.2.0 to 2.2.2 and the driver from 2.2.0-rc2 to 2.2.0-rc3, so now is a good time to discuss this. Earlier this week, Cassandra 3.0 was released[2]. There are some substantial changes including a rewrite of the storage engine to be better optimized for CQL, materialized views, and range deletes to name a few. The driver repo has been maintaining separate branches for each Cassandra branch which has been causing a lot of maintenance overhead. It was announced earlier today on the driver mailing list that the 2.2 branch, which we are currently using, has been merged into the 3.0 branch[3]. There will be no further development on the 2.2 branch. The driver is designed to be backwards compatible such that newer versions of the driver can be used with older versions of Cassandra. This means we could upgrade the driver without upgrading Cassandra; however, I think it makes sense to upgrade Cassandra as well. Are there any questions, comments, concerns, objections, etc.? [1] https://github.com/hawkular/hawkular-parent-pom/pull/55 [2] http://www.mail-archive.com/user@cassandra.apache.org/msg44740.html [3] https://groups.google.com/a/lists.datastax.com/forum/#!topic/java-driver-... _______________________________________________ hawkular-dev mailing list hawkular-dev(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/hawkular-dev

On Nov 17, 2015, at 10:45 AM, Thomas Segismont <tsegismo(a)redhat.com&gt; wrote: Le 17/11/2015 16:41, Peter Palaga a écrit : > A related topic: Juca has recently created C+ driver module in > hawkular-common [1]. The motivation is that ATM, pretty much every HK > component is using the c* driver and pretty much every HK component is > packaging the driver into its war/ear thus unnecessarily increasing the > size of the HK main distro. With this new module the components can > assume the driver is provided and do not need to include it in their > wars/ears. Does anybody have a problem with this approach? I'm perfectly fine with this.

On Nov 17, 2015, at 11:15 AM, Juraci Paixão Kröhling <jpkroehling(a)redhat.com&gt; wrote: On 17.11.2015 17:07, John Sanda wrote: > I’m fine with this as well. It is a step in the right direction; however, I think ultimately we should be using the same physical session object since we are communicating with the same physical cluster. To achieve this, I think a resource adapter is likely the best solution, but I don’t yet have any experience with writing resource adapters. While I also have no experience with that, I think I might have some free cycles for the next MS for it, so, I could handle that together with the kettle+accounts dist that was discussed on IRC some time ago.

On 17.11.2015 17:07, John Sanda wrote: > >I’m fine with this as well. It is a step in the right direction; however, I think ultimately we should be using the same physical session object since we are communicating with the same physical cluster. To achieve this, I think a resource adapter is likely the best solution, but I don’t yet have any experience with writing resource adapters. While I also have no experience with that, I think I might have some free cycles for the next MS for it, so, I could handle that together with the kettle+accounts dist that was discussed on IRC some time ago.

I was looking *exactly* at their code when I read your email :-) I'll ping them, to see what's the best way to consume their module and to check that it's even supported. - Juca. On 17.11.2015 17:53, Thomas Segismont wrote: > Le 17/11/2015 17:15, Juraci Paixão Kröhling a écrit : >> On 17.11.2015 17:07, John Sanda wrote: >>>> I’m fine with this as well. It is a step in the right direction; however, I think ultimately we should be using the same physical session object since we are communicating with the same physical cluster. To achieve this, I think a resource adapter is likely the best solution, but I don’t yet have any experience with writing resource adapters. >> While I also have no experience with that, I think I might have some >> free cycles for the next MS for it, so, I could handle that together >> with the kettle+accounts dist that was discussed on IRC some time ago. > > https://docs.jboss.org/author/display/TEIID/Cassandra+Data+Sources > > It would be awesome to have CLI management of the session and to reuse > (if possible) what's been done next door :) > > People who you might want to ping: > Jesper Pedersen > Emmanuel Bernard > Emmanuel Hugonnet > _______________________________________________ > hawkular-dev mailing list > hawkular-dev(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/hawkular-dev > _______________________________________________ hawkular-dev mailing list hawkular-dev(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/hawkular-dev

On Nov 17, 2015, at 11:02 AM, Juraci Paixão Kröhling <jpkroehling(a)redhat.com&gt; wrote: On 13.11.2015 20:46, John Sanda wrote: > Are there any questions, comments, concerns, objections, etc.? I'd ask to do this change in at least a week. I use cassandra-unit for Accounts unit tests and it seems there were already some breaking changes between 2.2 and 2.3, so, I'm concerned that there would be even bigger changes in 3.0. - Juca.