I have a client with a lot of customers using their UMP 7.1 via communities. They want to upgrade to NMS/UMP 7.5.
In a test systems the upgrade from 7.1 to 7.5 caused wasp to hang at startup. Support suggested clearing liferay's tables. After this wasp started, but of course the communities were all gone. This is no where near an acceptable solution in production. Note that lifray is upgraded as part of this update, and the lar files from 7.1 are not compatible with 7.5.
In an attempt to come up with plan B, we built a fresh NMS7.1/UMP 7.1 build. Exported a community as a lar file from production, imported it to the fresh system and confirmed it was OK. We then updated the test system to 7.5, and the community was converted to a Site. We exported the LAR from this 7.5 portal and thought all was good in the world. This exported LAR would not import into any other 7.5 UMP we had. Import failed with the helpful message that an unexpected error had occured.
My understanding is that liferay upgrades its tables when it first starts after an update, this is failing rather often going from 7.1 -> 7.5 and thus causing wasp to hang (This client left it about a week in test, which is hopefuly enough time even for wasp to start). I suspect there may be something left in the tables from versions prior to 7.1, which is causing more historic systems to fail, or it may just be random crud collected over time. I was hoping we could use a clean 7.1 to migrate the communities, then export the LAR's in case prod needed a table clear. Now the only approach for this client is to backup the DB ~1Tb, snapshot UMP, upgrade it, pray wasp starts, and rollback if not. As we have little faith that the upgrade will work without clearing all that lovely data, they are relucatant to take that leap.
We are now stuck with no safe way forwards. Any suggestions other than wait for 7.6? Any ideas why a LAR from our fresh 7.5 would not be accepted by other 7.5 UMP? The LAR failing was an unexpected surprise, perhaps fixing that is our best hope.