after 8 years of continuous running omero.server on an old MacPro we have finally managed to purchase a new DELL rackmount server and a Synology NAS rackmount for a fresh OMERO installation.
We decided to run it on Debian 9.8.
After some fiddling around we got OMERO to install. We have created a modified installation script to perform this fully automatic. This was created on a test PC
Eventually we transferred this script onto the DELL server and - hey - it installed OMERO automatically without any error
Admin start - to start omero - fine. Everything worked smoothly...........until yesterday morning when we had an unexpected power outage on our entire university campus. Server crashed (not yet connected to a UPS)...... I know.
But we simply restarted the Server - and run into all kinds of problems. All of them NOT related to any damage with the system.
Upon OMERO start we always and constantly got the following error messages:
No descriptor given. Using etc/grid/default.xml
icegridnode: failure occurred in daemon:
service caught unhandled std::exception:
Network.cpp:2065: Ice::SocketException:
socket exception: Address already in use
... and scratched our heads. It took us a 6 hours to figure out what happened.
The "challenge" were the two daemons "icegridregistry" and the "glacier2router" occupying the two ports 4063 and 4064. According to the omero documentation they automatically start during system reboot.
Apparently, the omero.server start up process does NOT verify whether these are running already and than stumbles into these issues.
Our solution (not being the deepest LINUX experts) was: we "simply" disabled to autostart on reboot systemwide and - it worked !!!
Now omero.server starts without any error message and we can continue with user installation and data migration (that will be fun as well - I am sure).
Now - here is my question - is this a "normal" behaviour ?? Using another dummy system we could confirm this behaviour on another Debian9 installation, so it appears to be a repeatable "feature".
Has anyone else stumbled into the same problem ? Have we missed some part(s) of the documentation.....?
Anyway - all comments, advice and help is indeed deeply welcome !!
Best
Peter
