Error thrown after 88 hour import about bitcoin-abe

[jtobey]

I've committed a workaround that may let you load the home page, but there is still a bug in how Abe computes chain statistics. Please paste the output of this query:

SELECT COUNT(1), COUNT(block_height), COUNT(block_satoshi_seconds), COUNT(block_total_ss)
  FROM block;

And this one:

SELECT c.chain_name, b.block_height, b.block_nTime,
       b.block_total_seconds, b.block_total_satoshis,
       b.block_satoshi_seconds, b.block_hash,
       b.block_total_ss, c.chain_id, c.chain_code3,
       c.chain_address_version, c.chain_last_block_id
  FROM chain c
  JOIN block b ON (c.chain_last_block_id = b.block_id)
 ORDER BY c.chain_name;

If you'd like me to try to reproduce the problem, I obviously want to get it right the first time, so let me know the exact version you used (as shown by "git rev-parse HEAD" if cloned from github) and the dbtype (MySQL, etc.). I could also use your full list of block hashes in the order loaded:

SELECT block_hash FROM block ORDER BY block_id;

This will be long, so send it by pastebin or email. If you can not provide it, please give the exact version of bitcoind you used.

[grimd34th]

emailed you the files

[grimd34th]

did you ever figure out what it was?

[jtobey]

I have not had time yet to look.

On Sat, Nov 17, 2012 at 10:01 PM, grimd34th [email protected]:

did you ever figure out what it was?

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/7#issuecomment-10481986.

[grimd34th]

i've rented a seperate box, threw the latest bitcoin and abe on it
i'll report back with results

[jtobey]

If you need a working instance quickly, I suggest you go back to v0.6 or
Sept. 15 (fc10484). I think those
versions predate this bug.

-John

On Sun, Nov 18, 2012 at 4:35 PM, grimd34th [email protected] wrote:

i've rented a seperate box, threw the latest bitcoin and abe on it
i'll report back with results

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/7#issuecomment-10491994.

[grimd34th]

its more or less for research not necessity

[grimd34th]

loaded absolutely fine on the other server

[jtobey]

Thanks for the update. I still suspect you have found a bug, unless the first 91844 block hashes are the same on both servers. The bug relates to orphaned blocks and blocks stored out of sequence. These arise from network latency during bitcoind operation, so they are not generally reproducible across bitcoind databases. The hash list you sent me will let me test this idea.

[jtobey]

I believe this is fixed in 3d5a832.

I was wrong about network latency being involved. The bug affects blocks that contain a previously loaded coinbase transaction, and the first such block is 91842. That and all later blocks would display incorrectly.

I am not sure why it would succeed on your other server. Did you use an older version of Abe software as I suggested?

[grimd34th]

i gave up on it, and i no longer have that server

[jtobey]

OK, thanks for the reply. I haven't tested the fix with BTC, but Namecoin has a similar block that now works, and your query output points to a problem around the first such Bitcoin block. Closing.