UCCX Failover problem

Unanswered Question
Sep 15th, 2009

Hi all,

UCCX 5.0 Premium.

I have 2 server in High Availability.

When I reboot the first server (primary), the secondary server becomes Master, which is fine.

After the primary server is rebooted and is up again, all the services became slaves, except the following services, which will take the Master status:

"CRS SQL Agent"

"CRS SQL Historical"

"CRS SQL Repository"

So basically all services are failing over correctly except those stated above will go to the first server instead of the second.

Can please someone explain to me how i can fix it?

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 4.3 (6 ratings)
godspeedbin Tue, 09/15/2009 - 02:18

what is the status for such three services on secondary server?

at this stage, you may try:

1. power off the primary server

2. do data check and resync on secondary server

3. start up the primary server

Jayd000007 Tue, 09/15/2009 - 02:27

the services I mentioned above are master on the primary and slaves on the secondary, which shouldn't be the case.

When I power down the primary, they do became Master on the secondary. Once the primary is up again they became salves on the secondary and Masters on the Primary.

Any idea how I can solve this issue?

RSwartzmiller Tue, 09/15/2009 - 08:03

The CRS Cluster uses the publisher/subscriber database model for data replication across the system. Under normal circumstances, the publisher acts as the source of data and the subscriber acts as the target for the data. The publisher/subscriber database model enables CRS to provide high-availability and failover support. To support this on the database level, the data must be available on multiple nodes of the cluster. To have such data availability, replication is used for the Agent, Historical, and Repository datastore.

The Configuration datastore does not use replication; instead, it uses atomic transaction to commit data changes to all active Configuration datastore in the cluster.

The publisher is the main database. All data is written to this database, with the other databases (subscribers) synchronizing with the publisher. If the publisher fails, then data can be written to the subscriber(s). When the publisher is back online, it returns to accepting writes. It also synchronizes with the subscriber(s) by performing the following functions:

• Adding any files or records that are new.

• Deleting any files or records that have been removed.

• Updating any files or records that have a later modification time stamp on the subscriber database.

When the publisher is fully synchronized, then all subscribers return to synchronizing with the publisher.

So what way it change when you boot your primary backup

Jayd000007 Tue, 09/15/2009 - 09:01

Just find out that the primary is not updating the RtCSQsSummary and RtICDStatistics. Even after rebooting it several times. The services I mentioned above are marked "Master" on the primary.

Can someone help?

Jayd000007 Wed, 09/16/2009 - 03:01

Thanks that was very helpful, i was starting to worry about the database.

Now can you tell me why those services dont failover correctly?

Thank you

stephane.bordenave Wed, 09/16/2009 - 04:58

Those services failover correctly !

During the initial setup, you have define a publisher for the datastore.

"CRS SQL Agent"

"CRS SQL Historical"

"CRS SQL Repository"

are master on the publisher because this server manage the database replication.

Jayd000007 Wed, 09/16/2009 - 05:42

That's the answer I was looking for.

So even if those services are Master on the primary, it means the replication with the secondary is still fine.

Thanks you made my day.


This Discussion