Issue 213 in couchdb-python: Couchdb connection lost without error

[couchdb...@googlecode.com]

Status: New
Owner: ----
Labels: Type-Defect Priority-Medium

New issue 213 by shn...@gmail.com: Couchdb connection lost without error
http://code.google.com/p/couchdb-python/issues/detail?id=213

My code contains lots of query in a remote couchdb database. The code
begins to run. After a while, it is just frozen. Any output or error are
not given. When I check the connection, I observe that the couchdb database
connection has been lost. It still busies the cpu without doing anything(it
seems so). What may be the reason of connection loss? Can I check the
connection status in python so that I can open new one if the current one
is closed? My related code is:

def addNewDnsRecord(self, record):
if self.checkHashValue(record['hashValue']):
.... #some code to generate an id
record['_id'] = tempid
self.bulk.append(record) #add record at the end of the bulk

if len(self.bulk) == 1000: #bulk with 1000 records are inserted
self.dns_db.update(self.bulk)
self.bulk = []

def checkHashValue(self, hashValue):
result = self.dns_db.view('records/hashcheck', None, key=hashValue)
if len(result) == 0:
return True
else:
return False


Couchdb : 1.0.1
Python : 2.6.5
Operating system of server which code runs: Ubuntu 10.04
Operating System of server which couchdb runs: Ubuntu 12.04


Any help is appreciated

[couchdb...@googlecode.com]

Comment #1 on issue 213 by kxepal: Couchdb connection lost without error
http://code.google.com/p/couchdb-python/issues/detail?id=213

Which version of couchdb-python you're using? Is there something in CouchDB
logs
about? Could you reproduce problem on current tip revision from repository?

If you're still got this problem, could you provide full version of test
script that could reproduce it? Actually, your code is quite common - I'm
using same approach to upload millions documents without any problems and
at least now connection is autoopening on request if it got "outdated".
That's why your problem sounds strange for me. Thanks!

[couchdb...@googlecode.com]

Comment #2 on issue 213 by shn...@gmail.com: Couchdb connection lost
without error
http://code.google.com/p/couchdb-python/issues/detail?id=213

In couch.log file, this lines are written:
[Mon, 23 Jul 2012 21:33:01 GMT] [error] [<0.63.0>] {error_report,<0.24.0>,
{<0.63.0>,crash_report,

[[{initial_call,{mochiweb_socket_server,init,['Argument__1']}},
{pid,<0.63.0>},
{registered_name,[]},
{error_info,{exit,eaddrinuse,
[{gen_server,init_it,6},
{proc_lib,init_p_do_apply,3}]}},
{ancestors,[couch_secondary_services,couch_server_sup,
<0.2.0>]},
{messages,[]},
{links,[<0.53.0>]},
{dictionary,[]},
{trap_exit,true},
{status,running},
{heap_size,1597},
{stack_size,24},
{reductions,446}],
[]]}}

I also updated the python version to 2.7.3. Anything hasn't changed.
Couchdb-python version is 0.8.0

[couchdb...@googlecode.com]

Comment #3 on issue 213 by kxepal: Couchdb connection lost without error
http://code.google.com/p/couchdb-python/issues/detail?id=213

> eaddrinuse
It looks like CouchDB couldn't open socket for your ip:port pair because
some another application is already done it or because there are too much
opened connections to CouchDB so you have exhaust all available ports for
new connections.

Could you play with lsof or netstat utility to track down how much
connections are opened from couchdb-python to your server? Also, please try
to reproduce situation on latest tip from repo, because since 0.8 release
there was major changes in http client module that may eventually get fixed
some negative behavior - I've failed to do it by myself for synthetic tests.

[couchdb...@googlecode.com]

Updates:
Status: WorksForMe

Comment #4 on issue 213 by djc.ochtman: Couchdb connection lost without
error
http://code.google.com/p/couchdb-python/issues/detail?id=213

Closing as WORKSFORME.