Did we ever find out why squidnet was losing money?

[makomk]

A SolidCoin pool, squidnet, was reporting some truely bizarre problems a couple of days ago in which they were losing masses of money - their blocks appeared to be accepted, got into the blockchain and showed up on the block explorer but for some reason the payouts never showed up in their wallet. I don't remember hearing anything about what happened next or what the problem was, though. It was just bizarre and made no sense. Anyone know or can guess the answer?

16:29 <+TimothyA> RealSolid: in what scenario would we be sending more coins than we actually receive??
16:29 <+TimothyA> we're operating at a daily loss for some reason all of a sudden :|
16:30 < Bitcoinfornewegg> people withdrawing
16:30 <+TimothyA> Bitcoinfornewegg: no, this is before withdrawing
16:30 <+TimothyA> for some reason we're paying out too much
16:30 < Bitcoinfornewegg> thats indeed odd
16:30 < Bitcoinfornewegg> please dont fix it ok
16:30 <+TimothyA> -_-
16:30 <+TimothyA> if we don't fix it, pool = gone
...

16:31 < Bitcoinfornewegg> fees for tx?
16:31 <+TimothyA> Bitcoin: that wouldn't be a 50 SLC loss per hour
16:32 < brukmann> TimothyA: well, didn't one of the newer betas start marking more shares valid?
16:32 <+TimothyA> brukmann: that has nothing to do with this :|
...
16:42 <@RealSolid> TimothyA: are you recording block values correctly
16:42 <@mtrlt> mm thought squidnet uses pplns?
16:43 <+TimothyA> RealSolid: yes
16:43 <@RealSolid> TimothyA: ie youre not saying you won blocks you didnt
16:43 <@RealSolid> also
16:43 <+TimothyA> not as far as I know
16:43 < nickwright> guys what is teh issueh ??
16:43 <@RealSolid> firstly id go back over the last 100 blocks you won
16:44 <+TimothyA> I've upped the confirmations to 20
16:44 <@RealSolid> and make sure they are in the chain
16:44 <+TimothyA> they are
16:44 <@RealSolid> you checked the last 100?
16:44 <@mtrlt> could the payouts be done eligius-style? in the generation tx
16:44 <+TimothyA> sharky did
...
22:01 <+TimothyA> great, now I'm losing 50SLC every 10 minutes
22:01 <+TimothyA> fucking brilliant
22:01 <+TimothyA> for some reason I'm not getting actually paid for the blocks that made it into the chain
22:02 < mush> :S
22:02 <+TimothyA> ... and now suddendly 200SLC vanished
22:02 < forsetifox> Kill it.
22:04 <+TimothyA> ...and another 400SLC
22:05 <+TimothyA> I'm not running a goddamn faucet here
22:05 <@mtrlt> shut down everything
22:05 <+TimothyA> or do I have to put confirms on 1000 or something to get it to work reliably?
22:05 <@mtrlt> how many confirms do you use now?
22:05 <+TimothyA> mtrlt: 20
22:06 <@mtrlt> >_>
22:07 <@AhimothMobile> 20 should be plenty
22:07 <+TimothyA> AhimothMobile: then why is shit still vanishing?
...

22:10 <@mtrlt> TimothyA: are you using data from the blockexplorer to determine whether you got a block?
22:11 <+TimothyA> mtrlt: I shouldn't have to in the first place -_-
22:11 <@mtrlt> so yes
22:11 <+TimothyA> but that's the only place where I can get the goddamn blocknumber now
22:11 <+TimothyA> RELIABLY
22:11 <@mtrlt> AhimothMobile: have you fixed reorgs?
22:11 <@AhimothMobile> blockexplorer might have some bad data in it right now from the occasional re-orgs
22:12 < mush> TimothyA what are you having issues with, a website?
22:12 <+TimothyA> mush: squidnet
22:12 <@AhimothMobile> not yet... testing in dev
22:12 <+TimothyA> we're paying out more than what we're getting somehow
22:12 <+TimothyA> we're paying out more than what we're getting somehow
22:12 <@mtrlt> so that's the problem
22:12 < mush> Ah, you run it right?
22:12 <@mtrlt> problem identified but not solved.
22:12 <+TimothyA> and if it doesn't get fixed quick, we have to shut it down
...

22:22 <+TimothyA> here, wtf is this shit
22:22 <+TimothyA> there are blocks on the website that aren't even in the wallet
22:22 <@mtrlt> cuqa: make sure you're using 0.09
22:22 <+TimothyA> yet they show up on the blockexplorer
22:22 < cuqa> yea, im using 0.09
22:22 <+TimothyA> HOW THE FUCK DOES THAT SHIT WORK?!
22:23 <@mtrlt> TimothyA: does the BE show that squidnet found the blocks?
22:23 < cuqa> is there some dbug mode for reaper why shares are invalid or whatever
22:23 <+TimothyA> mtrlt: yes, but they aren't in the wallet
22:23 <+TimothyA> but yet they are on squidnet's block listy
22:24 <+TimothyA> http://blockexplorer.ahimoth.com/Home/BlockDetails?blockNum=49273 http://blockexplorer.ahimoth.com/Home/BlockDetails?blockNum=49271 http://blockexplorer.ahimoth.com/Home/BlockDetails?blockNum=49265
22:24 <+TimothyA> not in the wallets

[johnj]

Well that explains one strategy:

Fuck with a pools blocks, so the poolop has to buy SC inorder to save face with his users.

Fucking shameful.

[cereal7802]

from what i saw in the chat the issue seemed less to do with solidcoin and more to do with squidnet code. seemed like it was sending out coins to users while also marking those coins as reserved in the pool wallet to be sent out on request of the users. perhaps that is not the case but that is what seemed to be said in SC irc.

[makomk]

Well that explains one strategy:

Fuck with a pools blocks, so the poolop has to buy SC inorder to save face with his users.

Fucking shameful.

That's the odd thing. RealSolid can definitely block their blocks from entering the blockchain, but without a back channel that'd be hard to disguise even in a closed source client he shouldn't be able to muck with them once they're in there in this way. It just doesn't make sense.

from what i saw in the chat the issue seemed less to do with solidcoin and more to do with squidnet code. seemed like it was sending out coins to users while also marking those coins as reserved in the pool wallet to be sent out on request of the users. perhaps that is not the case but that is what seemed to be said in SC irc.

That's not how I understood it. The way I read it, he was marking coins as available to users when the blocks their shares were in reached a certain number of confirmations, but the payouts never actually became available in the pool wallet.

[johnj]

Well that explains one strategy:

Fuck with a pools blocks, so the poolop has to buy SC inorder to save face with his users.

Fucking shameful.

That's the odd thing. RealSolid can definitely block their blocks from entering the blockchain, but without a back channel that'd be hard to disguise even in a closed source client he shouldn't be able to muck with them once they're in there in this way. It just doesn't make sense.

Of note, I understand very little of all that back end protocol stuff.  However, the outcome (poolOP must buy on the market inorder to save face) is still the same, regardless of purposeful interference or 'unintended' bug.
Edit: by 'unintended' bug, I mean releasing closed-source untested software that other people must carry the liability

[LoupGaroux]

Well golly, isn't the dev of sc sitting on 1.2 million sc that was pre-mined to take care of little problems like users that were raped by sc 1.0 or leaky code, or violent assault by cyber ninjas?

Shouldn't squidnet just just file a claim with their neighborhood Community Improvement Fund office, and I'm sure they will be made whole right away through the largesse of those tax and screw folks at sc.

[DeathAndTaxes]

1.2 million.  You wish.   

Try 12 million.  Plus about another 60K coins collected as half of the "king's tax" so far.

[cereal7802]

[18:40] <TimothyA> the issue was with neither squidnet's code or solidcoin (partly solidcoin, due to the send buys prior to b9)                                                           
[18:40] <TimothyA> it was MMC's code which was full of shit code and race conditions                                                           
[20:11] <TimothyA> forward that to the bitcointalk thread about squidnet                                                           
[20:11] <TimothyA> it won't let me post there     

so looks like i was wrong.

[makomk]

[18:40] <TimothyA> the issue was with neither squidnet's code or solidcoin (partly solidcoin, due to the send buys prior to b9)                                                           
[18:40] <TimothyA> it was MMC's code which was full of shit code and race conditions                                                           
[20:11] <TimothyA> forward that to the bitcointalk thread about squidnet                                                           
[20:11] <TimothyA> it won't let me post there     

so looks like i was wrong.

Well, guess that sorts it then, though I'm surprised he didn't spot that sooner. The audit trail required to do so should've been available to him.