I think there's still something awry with gracefully handling disconnects. It isn't unique to stratum, as I've seen it on LPs as well.
Right now have three units running... one with 1 GPU, one with 2 GPUs, and one with 3 GPUs. At the time all were pointed to stratum pools with bitcoind (LP) as a backup.
My internet connection dropped for a short bit. The 1GPU machine recovered properly. The 2GPU crashed, the 3GPU hung.
I've seen this before, where those with multiple GPUs crash or hang when the 1GPU continues working. The 1 and 3 GPU machines are window 7 x64, the 2 GPU is windows 8 x64.
Using 2.9.6.
M
I just started using cgminer after over a year with phoenix and no issues and I am seeing this behavior. If there is a network issue, cgminer crashes in Windows and I get the not responding...searching for a resolution dialog from Windows. I can replicate the behavior like clockwork because I am mining over an SSH tunnel and I have a scheduled task that kills plonk every 30 minutes and restarts it. When plonk gets killed, cgminer crashes. So it does not like losing its network connection at all.
Additionally, I think the issue only happens when connected via the stratum protocol, but I am investigating the details...