I have a question regarding clustering in Request Tracker (RT). Does RT have any built-in mechanisms that support clustering or high-availability configurations?
Here’s my scenario:
I have two locations – a primary location and a secondary (local) one. Ideally, if the connection between these two sites is interrupted, the secondary location should operate as an independent instance, allowing users to continue working without disruption.
Is there any way to achieve this setup with RT? Or perhaps some alternative solution or approach that others have found effective for similar needs?
Dealing with a link failure and allowing both sites to continue writing to the database is a very hard problem to solve. Very few applications can do it, and RT is not one of them.
If you truly need both sites to be active when the interlink is down, then you should install RT on a third site that both sides can talk to.
To handle the much simpler problem of one site going away entirely, there are three things to consider:
Application
Database
Attachment Storage
As far as I’m aware the application can be run on multiple web servers behind a load balancer, all state & session information is contained in the database.
Resilience of the database will depend on which you choose but both MySQL/MariaDB and PostpreSQL have mirroring/clustering solutions.
If you have externalised your attachments you will need to make sure that the storage is available to all the web servers. S3 is the obvious choice but an NFS share, or rsync after externalisation would also be options.