Bitcoin Forum

With recent DDoS's and other pool instability, it would be really useful for poclbm to fallback to another JSON-RPC endpoint/username/password, if the current JSON-RPC endpoint fails for any reason.

It is possible to use the 'flexible mining proxy' or pushpool or other external scripts and solutions, but a fully reliable solution requires running such proxies on each miner, which is decidedly sub-optimal.  Solution must be self-contained within poclbm.

Edit: bounty claimed

A Perl script (http://forum.bitcoin.org/?topic=4202.0) I made does what you're talking about.

https://github.com/kylegibson/poclbm

This is a REALLY quick hack, but it works. If you want additional features let me know.

That was fast lolx

I am working on a change that lets you specify the number of errors before the swap is made, i.e. a fault tolerance sensitivity.

A list of fallbacks would be nice (recently both slush and deepbit were down at the same time), rather than just one.  As you already mentioned, tuning the number of failures would be useful.  And figuring out some way to notice when the primary server is back up (presumably it is primary for a reason).

Regardless...  that meets the criteria in the original post.  10 BTC sent.

Bounty received. I will have the changes committed soon for tolerance and a list of backups.

Changes that enable for multiple backup pools and tolerance have been committed.

This commit does not include periodic checking of the primary pool. I have not yet designed a decent way to accomplish this yet, but will commit and reply when I do.

I committed a change which displays the current pool as part of the display (e.g. hashrate).

hi frozen, great job. Can you explain simply how the commandline should be input? also would poclbm try to connect to main pool from time to time and if it does get connected switch back to the preferred pool?

For unattended miners, let's say the primary crashes, and it mines on the secondary for a few hours, but you don't personally see this happen.

Does this log anywhere that it was using a second pool for some period of time?
Because ultimately, you'll have to go to that pool and collect payout, it would be nice to know if a crash happened while you weren't there.
Otherwise you'd be building up BTC in your secondaries without knowing about it.

It includes a README...

would this be compatible with the windows build of poclbm?

No, the code doesn't currently re-connect to the preferred pool, except when all of the secondaries have gone down.

It doesn't currently log anywhere which pools it has connected to, but it does display the current pool.

I am not sure whether it is compatible with the windows build, but I didn't do anything linux-specific, so I don't see why it would not work.

FYI - Just committed a change which attempts primary pool fail back on a specified get work frequency.