Server doesn't accept new connections after a while

[bauerj]

This is the last SSL-connection message in the log:

[27/10/2015-09:36:47] SSL     XX.XXX.88.65:49605    1 2.4.4

It's Tue 27 Oct 21:28:38 CET 2015 right now which means the server didn't receive a new connection for 12 hours.

The server is still connected to irc and thus visible in the peer list.

When I try to connect from the client with SSL I get a "Not connected" message.

When I try to connect with TCP, the server accepts the connection but the client doesn't seem to sync.

electrum-server getinfo still works and reports 171 open sessions.

Connecting with openssl_client fails too:

root@bauerj ~ # openssl s_client -connect electrum.bauerj.eu:50002
connect: Connection refused
connect:errno=111

I can connect to the TCP port but I get no response:

telnet electrum.bauerj.eu 50001
Trying 192.198.88.125...
Connected to electrum.bauerj.eu.
Escape character is '^]'.
{"id": 0, "method": "server.version", "params": []}
{"id": 0, "method": "server.version", "params": []}

The server imports blocks as expected. Restarting makes the server accept connections again.

[ecdsa]

this probably means that the ssl thread died.
please try to find a traceback in your log and post it here.

[bauerj]

The last traceback was 5 days ago (and unrelated).

[bauerj]

[22/10/2015-15:20:34] irc
Traceback (most recent call last):
  File "build/bdist.linux-x86_64/egg/electrumserver/ircthread.py", line 137, in run
    client.process_forever()
  File "build/bdist.linux-x86_64/egg/irc/client.py", line 278, in process_forever
    self.process_once(timeout)
  File "build/bdist.linux-x86_64/egg/irc/client.py", line 259, in process_once
    self.process_data(i)
  File "build/bdist.linux-x86_64/egg/irc/client.py", line 216, in process_data
    c.process_data()
  File "build/bdist.linux-x86_64/egg/irc/client.py", line 573, in process_data

[bauerj]

electrum.bauerj.eu shows this behaviour again. Is there anything I can do to debug this?

I'm not restarting it for now.

[EagleTM]

How recent is the codebase you're running and how much RAM do you have in the server and free to use for electrum server?
It might be related to #126 just with a different symptom

[ecdsa]

@EagleTM no, it's different. his ssl thread died.

[EagleTM]

However I have at least one other server operator who runs with 16 gig and 512 MB swap and runs into this issue. I think it's one possible side effect of memory starvation. The others being std::bad_alloc crash or OOM killer

[EagleTM]

OK it's independent of the RAM thing after all. It seems under some circumstances the ssl thread dies - pretty early after starting up it seems - while tcp and irc are still working. No idea how to debug for the time being. I'm suspecting openssl is acting up (some security bugfix possbily)

[EagleTM]

Ignore my last comment. In this particular case it was an expired SSL certificate.

[danny91]

Having the same issue. TCP or SSL stops accepting connections. SSL happened last night and TCP tonight. Checked traceback and no entries about anything stopping.

[luggs-co]

I always seem to notice this issue, though personally I have looked in awhile, has there been a patch for this yet?

[valesi]

Today I found my TCP unresponsive (although it appears to have been dead for about 23 hours). Here's what I've found so far. "command" will be short for run_electrum_server.py debug "command". transports[0] is my TcpServer thread.

"transports[0]" shows the TcpServer thread is stopped/dead. Server wasn't listening on 50001 at all. Trying to start the thread with "transports[0].run()" failed: File "build/bdist.linux-x86_64/egg/electrumserver/stratum_tcp.py", line 220, in run | IOError: [Errno 2] No such file or directory. Not sure what that's about.

electrum-server sessions | grep ^TCP still shows old sessions, that netstat shows as CLOSE_WAIT. It seems the dead thread doesn't release resources.

"transports.remove(transports[0])" and "transports.append(TcpServer(dispatcher, 'btc.smsys.me', 50001, False, None, None))" recreated the thread. "transports[1].start()" got it to accept TCP sessions again.

I'll try to make some time to debug this. I don't quite understand a bit of what's going on in stratum_tcp.py yet.

[shsmith]

I have seen the same problem.
There is an unhandled exception that is output to the console but not the log files.
Here is an example:

Exception in thread Thread-9:
Traceback (most recent call last):
  File "/usr/lib/python2.7/threading.py", line 810, in __bootstrap_inner
    self.run()
  File "build/bdist.linux-x86_64/egg/electrumserver/stratum_tcp.py", line 287, in run
    stop_session(fd)
  File "build/bdist.linux-x86_64/egg/electrumserver/stratum_tcp.py", line 164, in stop_session
    session.stop()
  File "build/bdist.linux-x86_64/egg/electrumserver/processor.py", line 237, in stop
    self.dispatcher.remove_session(self)
  File "build/bdist.linux-x86_64/egg/electrumserver/processor.py", line 189, in remove_session
    del self.sessions[key]
KeyError: 'x.x.x.x:51335'

Nothing catches this exception so the whole thread dies.

[ecdsa]

@shsmith good catch, I will have a look

[ecdsa]

that's fixed in 7949a93

[shsmith]

Found another unhandled exception that kills a thread:

Exception in thread Thread-4:
Traceback (most recent call last):
  File "/usr/lib/python2.7/threading.py", line 810, in __bootstrap_inner
    self.run()
  File "/usr/lib/python2.7/threading.py", line 763, in run
    self.__target(*self.__args, **self.__kwargs)
  File "build/bdist.linux-x86_64/egg/electrumserver/blockchain_processor.py", line 79, in do_catch_up
    self.header = self.block2header(self.bitcoind('getblock', (self.storage.last_hash,)))
  File "build/bdist.linux-x86_64/egg/electrumserver/blockchain_processor.py", line 146, in bitcoind
    raise BaseException(r['error'])
BaseException: {u'message': u"Can't read block from disk", u'code': -32603}

Same as before, this one did not appear in the log but was sent to the console output.

[shsmith]

Yet another unhandled exception that kills the TCP thread:

Exception in thread Thread-8:

Traceback (most recent call last):
  File "/usr/lib/python2.7/threading.py", line 810, in __bootstrap_inner
    self.run()
  File "build/bdist.linux-x86_64/egg/electrumserver/stratum_tcp.py", line 243, in run
    poller.modify(session.raw_connection, mode)
  File "/usr/lib/python2.7/socket.py", line 224, in meth
    return getattr(self._sock,name)(*args)
  File "/usr/lib/python2.7/socket.py", line 170, in _dummy
    raise error(EBADF, 'Bad file descriptor')
error: [Errno 9] Bad file descriptor