This was an issue with the pool that stopped/delayed responding to network requests. Miner probably had no explicit timeout check and decided that some execution thread was abnormally stopped. So it had to restart. On Windows some network errors detected much sooner (and as a side effect, Windows loses TCP connections instead of trying to wait a bit and recover). On Linux such connections still trying to recover for some time that exceed miner thread execution watchdog timeout. So there was no issue on Windows machines (miners just reconnected), but Linux miners aborted and restarted until pool recovered.
This definitely needs a fix because any network issue with the secondary coin pool should not abort mining the primary coin.
Yeah, as always, same code works in windows but fails in linux under some circumstances.
I will update linux version as soon as I can build it. My linux system decided not to load for some reason and now I have to reinstall it and all software, it takes some time. Nice OS.