Fixed upgrade.sh issue with outer-to-inner upgrades with NFS on inner server.
In an environment where /hxdepots (and thus /p4/common) is shared via NFS, an
upgrade on the outer server will confuse the upgrade.sh script into thinking
that the inner server has already been upgraded, and that only a patch update
rather than a major version update (requiring p4d -xu) is needed. So the
'p4d -xu' step is skipped, resulting in a failed service start.
This change fixes that.
In the case of a new binary being in place on disk, and linked to with the
usual SDP symlinks, but having a major version that the differs with the live
running p4d binary, a message is displayed something like:
The live running p4d binary will be restarted to update from 2021.1.2179737 to 2021.2.2410559.
A major version upgrade for the live running p4d is needed. A schema upgrade will be performed.