I've noticed a problem that has occurred twice over the past week or so. I run two GPU miners (with GUIMiner) at approximately 160Mh/s each. Now GUIMiner reports accepted shares, accepted shares for the last hour and network connection problems in its log. These are the only indicators of miner health directly available to me and are not evidence of anything.

What I have seen is that my miners report normal activity with no connection problems and yet one of the miners has no accepted shares reported on my account log on Slush's pool. The first occasion showed over six hours since the last share had been accepted; the second occasion showed over nine minutes without a share accepted.  On both occasions, the time since the last share accepted matched the time since a block was found on Slush's pool but not necessarily the last block.  It appears as if one of my miner's shares were no longer be credited after the pool's 'block found' event processing.

On both occasions, I stopped and restarted GUIMiner for the affected miner and share accepted activity on my pool account log resumed normally. The shares lost were not recovered.

On one other occasion, I saw that all shares from both miners were reported on my account log as if they had all come from one of my miners with the other showing no accepted shares.  As best as I could determine, this also was related to 'block found' processing on the pool.  That is, after finding a block, Slush's pool merged all accepted shares from both miners into one of the miner's accounts.

I hope this information might prove useful to Slush.