Block propagation issue has not been completely fixed. I am running 10 qt clients in total and at least 20 times a day i need to restart one because they stop syncing.
If i leave them long enough (20 min - 1hour) they sync by themselves.
I might be noticing the same issue. I'm running a custom front-end that interfaces 2 daemons on separate machines at highest priority, over 2 different ISPs through 2 different continents (Asia and US, via SOCKS proxy) and feeds the miners the highest block count work, and the front-end regularly reports de-synchronisations (> 1 block difference) that resolves only after many blocks.
In spite of this, I still occasionally get consecutive chunks of orphan chains, that seem somewhat dependant on network load (I.e. I don't see any orphans when network hash is ~ 3 mhps, but now I'm seeing like 20% orphan rate, and they come in chunks: i.e. t+0 accepted, t+1306 orphan, t+1418 orphan, t+1530 accepted, where t is about 3 hours before this post) so I assume even these two daemons don't seem to be working well.
However, the mining pools don't seem to be orphaning as much, so it might just be an localised latency issue for me, so I'll be grateful for any advice.