Olivier Fambon wrote:
Alex [19/10/07 12:08]:
I would like to have two controllers, on different machines, hosting
a single virtual database. So I've set up the two controllers, they
are both hosting a virtual database called MyDB with the same
settings, but they aren't "joined" - they don't know about each other.
...
What is the glue that tells them they're both hosting the same
virtual database?
Hi Alex,
part of the answer is as Mr Cecchet said on your other post: you
should have a <Distribution> element in both your vdb.xml files.
This is the glue, talking configuration.
Note that the names you specify in there for your backends and the
associated jdbc urls should all be different on each vdb part.
Right, I've looked and found out I am missing a <Distribution> element
in my virtual database XML file.
So I've added:
<Distribution>
<MessageTimeouts/>
</Distribution>
However, then it seems that using some sort of RAID is expected, because
controller startup results in:
14:28:00,868 ERROR controller.xml.DatabasesParser SingleDB configuration
cannot be used in a distributed virtual database.
14:28:00,871 ERROR controller.xml.DatabasesParser Failed to create the
request manager.
org.xml.sax.SAXException: SingleDB configuration cannot be used in a
distributed virtual database.
at
org.continuent.sequoia.controller.xml.DatabasesParser.endElement(DatabasesParser.java:775)
at org.apache.crimson.parser.Parser2.maybeElement(Parser2.java:1528)
at org.apache.crimson.parser.Parser2.content(Parser2.java:1779)
at org.apache.crimson.parser.Parser2.maybeElement(Parser2.java:1507)
The options in my virtualdatabase file regarding SingleDB are:
<RequestScheduler>
<SingleDBScheduler level="passThrough" />
</RequestScheduler>
<LoadBalancer>
<SingleDB/> <!-- no children allowed! -->
</LoadBalancer>
As far as I can see, this is correct - I do want just a single backend
behind both controllers.
The key is getting the two controllers talking to each other (hence me
adding <Distribution> tag)...
Maybe I'm barking up the wrong tree here somehow.
The desired outcome for this setup is two machines, neither of which is
a single point of failure, hence wanting each machine having a
controller with a single backend behind it...
thanks,
Alex
_______________________________________________
Sequoia mailing list
[email protected]
https://forge.continuent.org/mailman/listinfo/sequoia
