Re: Replication crashes in couch 1.0.1

[Steve Foulkes]

Well, that was stupid.  Time to change the password.

Steve

On 1/20/12 3:14 PM, Steve Foulkes wrote:

Hi,

I'm actaully using 1.1.0, not 1.0.1. It seems that the replication task gets stuck. I've tried canceling it, the call returns OK but the task remains running. The source database is large, varying between 50GB and 100GB. We're using filtered replication. Using continuous and non-continuous replication doesn't make a difference. Here is a stack trace from a more recent crash:

[Fri, 20 Jan 2012 21:11:23 GMT] [error] [<0.3971.0>] {error_report,<0.30.0>,

                      {<0.3971.0>,crash_report,
                       [[{initial_call,
                          {couch_rep_changes_feed,init,['Argument__1']}},
                         {pid,<0.3971.0>},
                         {registered_name,[]},
                         {error_info,
                          {exit,
                           {http_request_failed,

<<"failed to replicate http://cmst1:*****@vocms202.cern.ch:5984/workqueue_inbox/";>>},

                           [{gen_server,terminate,6},
                            {proc_lib,init_p_do_apply,3}]}},
                         {ancestors,

[<0.3951.0>,couch_rep_sup,couch_primary_services,

                           couch_server_sup,<0.31.0>]},
                         {messages,
                          [{ibrowse_async_response,
                            {1327,92924,816028},
                            {error,closing_on_request}}]},
                         {links,[]},
                         {dictionary,

[{timeout,{1327093063907963,#Ref<0.0.2.69305>}}]},

                         {trap_exit,true},
                         {status,running},
                         {heap_size,46368},
                         {stack_size,24},
                         {reductions,2419590}],
                        []]}}

Steve

On 1/19/12 3:35 PM, Steve Foulkes wrote:

Hi,

I'm having problems with couch replication crashing the entire database. This has been working fine for a couple months. Typical error I see in the logs is:

** Last message in was {'EXIT',<0.592.0>,
                            {http_request_failed,

<<"failed to replicatehttp://cmst1:********@localcouch:5984/workqueue_inbox/";>>}}

Full error log is at:http://home.fnal.gov/~sfoulkes/error.txt

Any ideas?
Steve