I received another reply today from Slush regarding the recent influx of invalid blocks.
The primary reason for this behavior was recent malleability attack on Bitcoin network which caused the Bitcoind overload with transactions and directly affected the ability to propagate blocks into the network.
Not only our nodes were affected but many other nodes across entire BTC network as well.
Which is what I said.
Well, probably that in combination with bitcoinds that haven't been optimised to cope in this transaction spam environment which can cause massive mempool and cpu blowouts and subsequently very slow block changes/getblocktemplate calls.
In other words slush hadn't manage to optimise his bitcoinds to cope under the circumstances and ended up depending on the newer bitcoind version to fix it. While he says many other nodes were affected, there were pools that were not.
