On 26 September 2010 10:11, Peter Somers <petsom...@gmail.com> wrote:
> Hi Dave,
>
> No luck. Couchdb crashed again :-(
>
> I really hope someone can help me on this issue...
>
> Best regards,
>
> Peter
Hi Peter
There may be 2 things going on here.
- your current crash appears to be due to file corruption;
- {{badmatch,{error,file_corruption}},
this is after couch reads an md5 hash in read_raw_iolist_int - maybe
due to going over length, I can't tell.
- your first crash is the one I can replicate - the whole VM dies with
a memory allocation issue. I can replicate this on my win7 enterprise
couch running 1.01, but not 0.11.0.
I can load the DB up to this size; 0xffd9a0f6d, couchdb dies:
{{badmatch,{error,enomem}}, [{couch_file,read_raw_iolist_int,3},
{couch_file,handle_call,3},
Restarting windows/couchdb & then curl http://localhost:5984/fatso yields
{"error":"kill","reason":"{gen_server,call,
[couch_server,
{open,<<\"fatso\">>,
[{user_ctx,
{user_ctx,null,
[<<\"_admin\">>],
<<\"{couch_httpd_auth,
default_authentication_handler}\">>}}]},
infinity]}"}
Anyway some more work ahead to identify exactly where this comes from
- thanks for raising the jira.
cheers
Dave
