Processing is going to be down for maintenance for between 2 and 6 hours.  We are migrating some servers around.

Web site should be up mostly but will have a little downtime as well.

All the nodes are running the same code that was in test.

My guess is that your miner hadn't been restarted since before the pushpool instance was advertising the stratum url.  Now when you connect to the old url, your miner probably automatically tries to flip to stratum.

Yeah, there was a fix for the stratum implementation that I haven't pushed to production yet:
https://github.com/fireduck64/SockThing/pull/1

If you are working on test but not the others that is probably why.

As the pool is very much in the red, I am increasing the fee by 1% across the board.  Sorry.  Updated fee schedule is on the web page: http://hhtt.1209k.com/

It looks like you've never been on stratum on this pool.  You probably first connected to the pushpool instance back when it wasn't advertising that people should switch to stratum.

Can you try connecting to my test server:
stratum+tcp://stratum-test.hhtt.1209k.com:3333/

I've removed the get client call, it might be confusing your client somehow.  Anyways, this way I'll get a complete log of all messages and should be able to figure it out from there.

We are up on the proper chain now.

I just realized that I am not very smart.  The block submit certainly wouldn't work with what I have rigged up.  Shut back down for now.

It is kinda back up now.  I am editing the blocktemplate by hand.  No transactions, but should still help advance the 0.7 side chain.

Downloading blockchain on 0.7 bitcoind.  Pool will be back up when that is done.  Sorry for all the chaos.

I understand.  My question should have been phrased:

Is 500k the real limit we should follow for now or was that just a quick conservative number while the issue is being researched?  (I have another source that suggests the real limit is 998k).

Thanks for sharing your amazing knowledge of which number is bigger.  We are all enlightened.

My guess is that it is fine:
http://blockchain.info/block-index/356435/00000000000002707802d1a7ed26e04f720578ada6cc8f2f0d2a1d040dfc2491

Anyone know if 750k is ok?

I don't have anything like that yet.  There are a few issues.  One technical one is deciding when a new difficulty applies on an open session without losing work or crediting things wrong.  Another is different users have different desires for variance.  For example, on my miners I don't care as long as I have enough data points for my alarms on metrics to work correctly.  Others might want that tighter so I think for now I'll leave it up to miners to set as they see fit.
 

Yeah, pays most of my hosting costs this month.

Another high fee block, this time 6 BTC total fees:
http://blockchain.info/block-index/356955/00000000000002704f682ac6ab29f51864c082c600b96c0b7abec6eb952f155c

Settings are all default except '-blockmaxsize=750000'.

My experiment with 750k block size in pdx worked well.

We got a higher transaction fee for including more:
http://blockchain.info/block-index/356435/00000000000002707802d1a7ed26e04f720578ada6cc8f2f0d2a1d040dfc2491

I'm switching the other nodes now, which involves a restart of bitcoind.

Me too.  It is less negative at least.

I generally do it based on last week or last day.  Last hour seems a little too new.