Rt cluster using postgres

I had an idea to cluster RT, using postgres to synchronize the different
RT servers.
The only issue I have, is if the DB’s stop talking to eachother, then
there would be a problem with the ticket sequencing. (you would get
duplicate tickets for the same id).

I’ve tried to look around for something like modifying the ticket id. I
know it’s numeric only, but haven’t found much to do something like have
one server do odd numbers, the other even or some such…
Any ideas of a way to do this, or is it simply not possible at this
time.

Also, any other reasons or possible problems anyone could see with this?

My secondary step, is just to run 1 instance, but use the postgres
replication and probably rsync for a hot standby.

I think your solution is going to come from postgres, rather than RT.

I believe Ticket ID’s come from tickets_id_seq. You could set them to
increment by 2, with one starting at an even number and another starting
at an odd number.

See: PostgreSQL: Documentation: 9.4: CREATE SEQUENCE

However, I don’t have enough experience with clustering to know if
that’s a good idea.

Here’s some info on postgres that could be useful to your situation:
https://wiki.postgresql.org/wiki/Replication,_Clustering,_and_Connection_Pooling

Joseph WagnerOn 08/28/2015 03:35 PM, Jeff Mundine wrote:

I had an idea to cluster RT, using postgres to synchronize the different
RT servers.
The only issue I have, is if the DB’s stop talking to eachother, then
there would be a problem with the ticket sequencing. (you would get
duplicate tickets for the same id).

I’ve tried to look around for something like modifying the ticket id. I
know it’s numeric only, but haven’t found much to do something like have
one server do odd numbers, the other even or some such…
Any ideas of a way to do this, or is it simply not possible at this
time.

Also, any other reasons or possible problems anyone could see with this?

My secondary step, is just to run 1 instance, but use the postgres
replication and probably rsync for a hot standby.