[Qt] crashing testnet client with latest master? #4628

My testnet client crashes for no apparent reason and also the debug.log is currently not helpful. Did anyone observe the same thing? The mainnet seems be running just fine!

This is my command line for starting the client: bitcoin-qt.exe -debug -help-debug -logips -testnet -txindex -bind=127.0.0.1 -externalip=XYZ.onion @laanwj Can there be a problem with the new testnet seeds?

I doubt that's it, but is there a message in the log that it is using the seeds?

Can you get a backtrace?

It's no debug build, so currently no. Can you try testnet with my parameters? And I'm going to get a debug build and report back tomorrow.

Also this was with peers.dat and fee_estimates.dat deleted to ensure there was no corruption of these.

After filling in a valid onion address and removing -txindex (don't want to reindex now), no problems. How long does it take to crash? What phase is the client in when it crashes?

Well it happens frequently after a few minutes of uptime. The client was synced and just sat in the system tray.

These are the last entries in the debug.log:

<pre> 2014-08-04 14:33:19 sending: getdata (37 bytes) peer=2 2014-08-04 14:33:20 received: block (3500 bytes) peer=2 2014-08-04 14:33:20 received block 0000000095ccf30194baaf7e988fab6ed2c5f085bacbe6d5ac4f20cf1a8660af peer=2 2014-08-04 14:33:20 - Load block from disk: 1.00ms [0.00s] 2014-08-04 14:33:20 - Connect 15 transactions: 0.00ms (0.000ms/tx, 0.000ms/txin) [0.03s] 2014-08-04 14:33:20 - Verify 15 txins: 1.00ms (0.067ms/txin) [0.06s] 2014-08-04 14:33:20 - Index writing: 2.00ms [0.02s] 2014-08-04 14:33:20 - Callbacks: 0.00ms [0.00s] 2014-08-04 14:33:20 - Connect total: 3.00ms [0.11s] 2014-08-04 14:33:20 - Flush: 0.00ms [0.00s] 2014-08-04 14:33:20 - Writing chainstate: 1.00ms [0.03s] 2014-08-04 14:33:29 ERROR: Proxy error: host unreachable </pre>

I guess it was a random event that the last entry is a proxy message, will start again and take a look at the end of the log.

But is there any message anywhere in that execution about "Adding fixed seed nodes as DNS doesn't seem to be available."? If not, the seeds can be excluded as the problem. Edit: btw another recent testnet crash #4607

Crashing again now this is shown:

<pre> 2014-08-04 14:39:09 received: inv (37 bytes) peer=2 2014-08-04 14:39:09 got inv: block 00000000fadd536bb4d0c6d4ded910115cfd6c3a351d60ac6972bfe6b5b1b247 new peer=2 2014-08-04 14:39:09 Requesting block 00000000fadd536bb4d0c6d4ded910115cfd6c3a351d60ac6972bfe6b5b1b247 peer=2 2014-08-04 14:39:09 sending: getdata (37 bytes) peer=2 2014-08-04 14:39:10 received: block (1493 bytes) peer=2 2014-08-04 14:39:10 received block 00000000fadd536bb4d0c6d4ded910115cfd6c3a351d60ac6972bfe6b5b1b247 peer=2 2014-08-04 14:39:10 - Load block from disk: 0.00ms [0.00s] 2014-08-04 14:39:10 - Connect 2 transactions: 0.00ms (0.000ms/tx, 0.000ms/txin) [0.01s] 2014-08-04 14:39:10 - Verify 1 txins: 0.00ms (0.000ms/txin) [0.07s] 2014-08-04 14:39:10 - Index writing: 0.00ms [0.01s] 2014-08-04 14:39:10 - Callbacks: 0.00ms [0.00s] 2014-08-04 14:39:10 - Connect total: 0.00ms [0.09s] 2014-08-04 14:39:10 - Flush: 0.00ms [0.00s] 2014-08-04 14:39:10 - Writing chainstate: 0.00ms [0.02s] </pre>

@laanwj Only seed message is this: 2014-08-04 14:37:23 P2P peers available. Skipped DNS seeding.

I've just merged #4598. This solves some possible crashes.

I merged that patch and currently it's working fine, will report back later again.

@laanwj No more crash since #4598, looking good... closing this.