[ome-users] Rolling reboot / fail over on OMERO server installation

[Josh Moore]

On Fri, Oct 2, 2015 at 3:02 PM, Julio MATEOS_LANGERAK
<julio.mateos-langerak at igh.cnrs.fr> wrote:
> Dear all,

Hi Julio,


> I remember someone mentioning an OMERO installation with a rolling mechanism
> to do upgrades.
>
> We want to do something similar in our installation to let a second server
> quick in when a failure is predicted (e.g.. a memory usage threshold)
>
> Has someone done something like this before?

The extension mechanism that you might have heard about is available
as a configuration option:

  https://github.com/openmicroscopy/openmicroscopy/blob/v5.1.4/etc/omero.properties#L642

which points to an implementation of this interface:

  https://github.com/openmicroscopy/openmicroscopy/blob/v5.1.4/components/blitz/src/ome/services/blitz/redirect/Redirector.java

No implementation is shipped with a vanilla OMERO but it should be
possible to implement something locally which will read the current
server from either a database or a configuration file. For this to
work, it's necessary to have enough memory locally to temporarily run
two servers until all active sessions are closed.


> Best, Julio

Cheers,
~Josh.