Quote from: tuaris on December 25, 2013, 01:52:01 AM

At exactly 8:30PM EST my bitcoind service crashed (out of nowhere) after running for several months:

Code:

Dec 24 20:30:39 bitcoin kernel: pid 3664 (bitcoind), uid 0: exited on signal 6 (core dumped)

Now it will no longer start, it dies with this message:

Code:

Assertion failed: (internal_key.size() >= 8), function ExtractUserKey, file ./db /dbformat.h, line 96

The debug log only shows:

Code:

2013-12-25 01:49:00 Bitcoin version v0.8.5.0-gef14a26-beta ()
2013-12-25 01:49:00 Using OpenSSL version OpenSSL 0.9.8q 2 Dec 2010
2013-12-25 01:49:00 Default data directory /.bitcoin
2013-12-25 01:49:00 Using data directory /var/db/bitcoin/
2013-12-25 01:49:00 Using at most 125 connections (11095 file descriptors available)
2013-12-25 01:49:00 Using 2 threads for script verification
2013-12-25 01:49:00 init message: Verifying wallet...
2013-12-25 01:49:00 dbenv.open LogDir=/var/db/bitcoin/database ErrorFile=/var/db/bitcoin/db.log
2013-12-25 01:49:00 Bound to [::]:8333
2013-12-25 01:49:00 Bound to 0.0.0.0:8333
2013-12-25 01:49:00 init message: Loading block index...
2013-12-25 01:49:00 Opening LevelDB in /var/db/bitcoin/blocks/index
2013-12-25 01:49:00 Opened LevelDB successfully
2013-12-25 01:49:00 Opening LevelDB in /var/db/bitcoin/chainstate
2013-12-25 01:49:00 Opened LevelDB successfully

This is bitcoin 0.8.5 on FreeBSD 9.0 32-bit. Disk space and RAM are okay.

Exactly the same here (except openssl 1.0.1e and 64 bit 11.0). I tried a reindex but found I got the same errors during reindex that the Mac OSX folks typically had. eg: Invalid argument: not an sstable (bad magic number)

I looked at the 0.8.5 -> 0.8.6 patch to env_posix.cc that disabled mmap() for writing in LevelDB for OSX. When I enabled that for FreeBSD as well, the reindex problem was solved.