Hello there,
i tested Sequoia version 2 one year ago and i was very happy that this
solution was stable and very comfortable to administrate.
Since a programming project was/is a never ending story we never used
this cluster-solution for a productive environment.
In the near future we want to use sequoia as a productive solution. To
look ahead to have a "update to date" solution we plan to use Sequoia 3.
When i was testing Sequoia 2 i used jgroups for group-communication and
it works fine. For version 3 we use the new implemented
appia-group-communication-protocol. The configuration of the cluster was
easy and works fine.
But now we've some trouble with appia.
At first a little overview of our environment:
Raidb-1
Data Center 1 -> DB-Node 1 -> Controller 1
Data Center 2 -> DB-Node 2 -> Controller 2
If we startup Controller 1, load the virtualdatabase configuration-file
and enable the backend everything is working fine. If we startup
Controller 2 and load the configuration file for virtualdatabase the
controller join the appia-group and they can communicate each other. In
the administration-Console of the virtualdatabase the command "show
controllers" shows two controllers which information is right. But if i
create a dump of "DB-Node 1" after few minutes the first Controller
throw this error/info-messages (i replaced the ip with the nicknames above):
2007-02-19 09:43:47,593 INFO continuent.hedera.gms
Member(address=/DB-Node-2:32790, uid=DB-Node-2:32790) failed in
Group(gid=xstamm)
2007-02-19 09:43:47,593 INFO continuent.hedera.gms
Member(address=/DB-Node-2:32790, uid=DB-Node-2:32790) failed in
Group(gid=xstamm)
2007-02-19 09:43:47,607 WARN controller.virtualdatabase.xstamm Unable
to find remote controller name for member
Member(address=/DB-Node-2:32790, uid=DB-Node-2:32790) in list
{Member(address=/DB-Node-1:32790, uid=DB-Node-1:32790)=DB-Node-1:1090}
2007-02-19 09:43:47,608 INFO controller.virtualdatabase.xstamm 1
requests were waiting responses from Member(address=/DB-Node-2:32790,
uid=DB-Node-2:32790)
So for my opinion the seconde DB-Node is leaving the Cluster. When the
Dump-creation is finished the DB-Node 2 isn't joinig the cluster again.
In the Apppia-Configuration we used
TCP SEQ Channel
TCP TOKEN Channel
UDP p2p SEQ Channel
UDP p2p TOKEN Channel
with every configuration we get the same problems!
Do you know this problem? Did you know a solution? Maybe i can post my
config-data to find my problem! :)
Thanks for your answers and your time :)
greetings
Chris
_______________________________________________
Sequoia mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/sequoia
