|
Hi,
So just to be clear you are the only miner which actually does gets instead of posts on purpose.
I already fixed my code to deal with your special case but If you truly believe thats the correct way to do it can you please write it down somewhere? If you made a post which is something along the lines of How LP works and how I believe it should work it would make it a lot easier to develop.
|
|
|
|
|
If you are not adverse to source just add 'info':'' to the info for the pool in servers.
I'll add a command line option later tonight.
Is there a specific pool you don't want to use because its become hopper proof? Or is it just a case of a lot of pools and you don't want to make all of the accounts.
|
|
|
|
|
Google App Engine Eh? Maybe. Stats are first. Then a webpage to show them.
|
|
|
|
|
There is actually another bug which could have caused this. It looks like I screwed up the delag call and didn't notice. So calls could fail slowly one after each other lagging out each pool in turn. And then you'd get the massive failover which happened to you.
Its fixed now. I think I should attach timestamps to the non debug calls also.
|
|
|
|
|
Is there somewhere which actually standardizes long polling? Or an alternate spec I missed?
|
|
|
|
Hello, It appears that you use post instead of get when doing Long Polling. That is contradictory to the unofficial spec and really annoying when trying to debug software LP failures. See https://deepbit.net/longpolling.php-C00w |
|
|
|
Hello, It appears that you use post instead of get when doing Long Polling. That is contradictory to the unofficial spec and really annoying when trying to debug software LP failures. See https://deepbit.net/longpolling.php-C00w |
|
|
|
In case you haven't noticed I got Server Side LP working. And now my rejection rate is 0%. Which is pretty sweet.
I just did a pull and check out this hotness I understand that you feel quite frustrated as you throw another layer in between you and your pool and it spews oodles of output and stops giving work. I try and make this software do everything it can to get your work submitted. In this case however it spewed loads of output for two reasons. 1) It literally tried every server in the book in order to get a getwork and failed. 2) I left a little bit too much LP debugging in because I had just released it and was worrying that it would die. And after spewing output it did come back up which is the point of bitHopper. -c00w Answers: 1) Pastebin? Yes I like pastbin for large files. I like opening an issue on github even better. 2) More details? Your internet probably died for a brief second. It then chewed through every server before defaulting to an extreme corner case I never expected. It will select mineco if literally all the other servers fail. I fixed that now though. However the LP shouldn't have been retried more than once for each server. Let me doublecheck that code. Edit: 3) More details? On looking through the logs I realized that some of my debug messages don't make any sense. What happened in your case is most of that spewing was your miner submitting incorrect LP request. Probably because it always submits them and it got extra desperate when it had no work. But I'm rewriting most of my debug output to make this clearer and the normal output to hide it. Nothing went wrong with the server side LP code. How many miners are you running though? Thats a lot of clients being triggered on a server change. |
|
|
|
|
In case you haven't noticed I got Server Side LP working. And now my rejection rate is 0%. Which is pretty sweet.
EDIT: Actually Its not completely working. But it is a lot less broken.
|
|
|
|
Hi, I am getting the following errors each time I start bithopper, or when it switches pools: Unhandled error in Deferred:
Unhandled Error
Traceback (most recent call last):
File "C:\path\to\bithopper\pool.py", line 383, in <module>
main()
File "C:\path\to\bithopper\pool.py", line 377, in main
update_call.start(117)
File "C:\Python27\lib\site-packages\twisted\internet\task.py", line 163, in start
self()
File "C:\Python27\lib\site-packages\twisted\internet\task.py", line 194, in __
call__
d = defer.maybeDeferred(self.f, *self.a, **self.kw)
--- <exception caught here> ---
File "C:\Python27\lib\site-packages\twisted\internet\defer.py", line 133, in m
aybeDeferred
result = f(*args, **kw)
File "C:\path\to\bithopper\pool.py", line 253, in update_servers
d = getPage(info['api_address'])
File "C:\Python27\lib\site-packages\twisted\web\client.py", line 547, in getPage
*args, **kwargs).deferred
File "C:\Python27\lib\site-packages\twisted\web\client.py", line 525, in _make
GetterFactory
from twisted.internet import ssl
File "C:\Python27\lib\site-packages\twisted\internet\ssl.py", line 42, in <mod
ule>
from OpenSSL import SSL
exceptions.ImportError: No module named OpenSSL
Trying to delag I'm running bithopper on windows xp, with the latest phoenix miner. I'm not exactly sure if these errors are effecting the performance of bithopper, as it continues to rum seemingly unaffected. Is this bad? Yes. You need to install the python OpenSSL libraries on windows. https://launchpad.net/pyopenssl. I'm adding it to the readme. |
|
|
|
Not client disconnects, the error I posted gets spammed several times per second, so quickly that it fills the buffer before I'm able to pause and see what occurred immediately prior.
The None error should be fixed. The other errors will still appear but should be properly recovered. They still print the full error text but I should probably throw in a debug flag for that output. |
|
|
|
Actually, you could probably still hop mine.co but will have to change the effective share range.
Maybe. I need to think about the math more and there are a million other things that need to be done for bitHopper. If you increased the effective share range I think you would still not make as much money hopping it as the other pools. The initial ramp up means you wouldn't get paid very much and then you could never quit or you would have to ramp up again... |
|
|
|
|
1) Lots of crashes?
Are you saying the server process died? Or your client got disconnected... Most of the crashes you guys reported were in your bitHopperLP and as of about 2 am last night had a lot of try catch blocks to deal with the bizarre behavior of miners. I added some more.
2) Spinning out the authentication info?
Yeah I'll do it.
3) That one with server_name being none?
I fixed it. I think if every server lags out it would do that. Now it never will.
4) Adding more pools and mine.co is now not poolhoppable?
I'll set it as a info pool. I'll add the other pools when I have the time. I'm not going to add slush's as a poolhopper pool because pool hopping seems to be mostly random chance.
5) How to put a 10-15 second delay before pulling servers after LP is triggered?
Well LP should't be triggered server side from pools. It triggers miners from the pool changing servers. There is a .addDelayedCall or something similiar in twisted that you could use.
|
|
|
|
|
1) Stats?
Yeah. I'm trying to make it rock solid and get LP to work. If I get that the reject rate may actually be 0% which is cool. A lot of miners just drop shares after a connection failure or are just written oddly/ ignore standards. Then stats. Then scraper for all the pools that aren't deepbit. Then deepBit.
*rant*
Miners I have found issues with:
poclbm: Drops connection way to quickly.
phoenix: Drops shares on a network failure. *I haven't confirmed this but it looks like it*
Diablominer + cgminer: Uses get instead of post for longpolling. Completely ignoring the standard. And it was an obscure bug to find.
Edit: I bet they all have the same issue as phoenix. But I don't use them enough to notice.
*end rant*
2)Deepbit support?
Yeah that sounds like the way to do it. I also want to look at how the multipool guy did it and build in a share estimator given scraped hash information to know when to switch off.
|
|
|
|
|
1) Json decoding error with eligius looping?
It looks like eligius doesn't like us hammering it with getwork requests in the case of one failure. I think its some sort of DDOS proctection. I'll throw in another backup pool for it to swap between.
2) IO Operation on a closed file
It looks like someone connected with LP and then died before it could read. Not a huge issue. I'll wrap it in a try except.
3) Why it probably died?
Same issue as the JSON decoding error with eligius looping. I need to add another backup pool so that if one dies/decides we are DDOSing it we can still do stuff.
Edit: I couldn't find another pool with pool hopper protection that works except btcmine. And well... registration is down. I did redo the networking code to hopefully be a little nicer to eligius and convince it to not DDOS flake out on us as much.
|
|
|
|
|
1) Burf good catch with the error call back. I'll fix it soon.
2) Defaulting to eligius at over difficulty instead of .40*difficulty?
Hmmm. Let me think about it. We are using eligius as our 'solo' mining pool because the payouts are roughly consistent mathematically with what solo mining should be. And the sweet spot for switching from a single pool is .435%
3) I figured out why diablo and cgminer are broken!
They completely ignore how long polling is supposed to work and use a post instead of a get. I'm fixing it now.
|
|
|
|
|
Good catch. I'm fixing it.
|
|
|
|
|
Its an LP issue. Diablominer doesn't actually read from the LP channel. Once it gets a packet it triggers and asks for more work from the normal channel. There is a way to rewrite the clientside LP and I'll add it to the list but for now just run with pool.py --noLP.
|
|
|
|
I actually encountered a new problem with DiabloMiner where it loops RPC Request in BitHopper continuously. This only happens IF DiabloMiner doesn't error out as described earlier.
Hmm. I ran diablo miner and didn't get the LP problem. I did see an abnormally high number of RPC requests. I am seeing an issue where my code is submitting 2-4 times as many RPC request as it should. I'm debugging. EDIT: Turns out my logging function double printed certain things. Fixed now. It looks like Diablominer just pulls a lot of work. I'll keep playing with it and see if I can find the LP issue though. EDIT AGAIN: Actually I remember causing that error in one of my commits. I fixed it once I tested it on my server and it didn't work roughly 2 minutes later. Git Pull and see if it works? |
|
|
|
|
Show Posts
