Usually, it's not necessary. In some transitions the LP connection is not severed neatly, which consequently causes buggy miners to fail doing LP from that point on. They never recover from that one broken LP. So the essence is: If you are missing log messages about LP and are suddenly experiencing an elevated amount of invalid shares, try restarting your miner.

This can actually happen at any pool, even during normal operation, when the connection has a hickup. Therefore it's wise to check for it now and then anyway.

I remember seeing this problem with Phoenix v1, but only because I was using it myself. I have not tested other mining software, or more recent versions of Phoenix.

When we see elevated invalid shares pool-wide after a patch, we always post a message here on the board asking you to restart your miners.