fisheye cold failover

[Kumaran]

Has anyone implemented a cold failover instance for fisheye ? If so, are there any recommendations ?

We have a fisheye installation and a instance directory. Copying both the directories to another instance will work. But, is there a specific directory that needs to be synced between the instances ?

[Piotr Swiecicki]

FishEye/Crucible internal state is kept in

• filesystem in $FISHEYE_INST folder
• configured database

In order to keep cold failover system you need to ensure both of them are copied consistently to failover location.  Note copying live system is very dangerous as you risk having inconsistent data.  

The recommended way of backing up a system is using built in tools, documented in https://confluence.atlassian.com/fisheye/backing-up-and-restoring-fisheye-data-298976928.html.  If you want to customise it and take benefit of filesystem snapshot mechanisms for instance, I would suggest the following approach:

• take FishEye/Crucible instance down (e.g. run bin/stop.sh)
• copy database to failover location
• copy filesystem to failover location (e.g. with the use of rsync, FS snapshots etc.)
• take FishEye/Crucible instance up

it would require short downtime but should ensure consistency of the data.