Sensors Install Symlink behavior change (avoid overwrite)

Installing the gpfs sensor package will result in zimon been
installed under /opt/IBM/zimon. In case /opt already exists this migth
result in problems, in our case it was a symlink to the gpfs
filesystem which was silently deleted. This was seen with debian.
In general it would be better to have the flexibility were to install
the zimon package, since many installations use /opt as a network
share, e.g via nfs or gpfs.Why not put all IBM related stuff under
/usr/lpp? Zimon is in this setup unsuable for us, since all hpc nodes
we use have the software installed on gpfs mounted under /opt, which will conflict with this layout. Install happend via dpkg on CLI. It pas part of a rolling upgrade from 4.1.0.7 to 4.2.0.0 and then to 4.2.0.2. Why is there any need to have this installe dinto a new location and not also under /usr/lpp/ or /usr/lpp/mmfs? This makes system handling very hard. Why not make the location for ZIMON somehow arbitrarily? tested here the behavior and at least the content behind the SymLink is exposed to be deleted or overwritten! This is documented OK but .... had a longer discussion with customer about the implemented and seen behavior. We agree on the workaround .... BUT .... what happens if GPFS dies, the sensoring the won`t work anymore .... AND .... that alerts never will end up in the monitoring as all has been gone. Another bad side affect is if a node-based configurations is needed it can`t be set up. As when having it placed 'hard' under /opt inside the GPFS filesystem the config files are the same for all nodes instead when having them locally. Wondering if there is any chance to start a discussion with the technical lead about this? Seeing more and more pressure on agile, portable and flexible solution a way via hard coded links is not a good approach. If there is anything i can do to trigger that discussion via another channel feel free to get me in the loop. Especially for that customer it means he need to re-work his whole solution and in the next phase when adding more nodes he has more work to do then usual. As heard several times on events like User Groups, .... when development was onsite it was said they will hear to the Community. To be honest this is a very good example to do so and make customers work easier. As everything different will disatisfied customers. This one here in that case is on the best way to this as he told me.