> hi,guys
> i encounter an segfault with uwsgi.1.4.4 + python 2.7.3  , backtrace as
> follow:
> #0  0x0000003549079a14 in strncpy ()
> #1  0x00000000004622ad in init_uwsgi_app (loader=1, arg1=0xb43b730,
> wsgi_req=0x2ab1752bf060, interpreter=0xb43f120,  app_type=0)
> #2  0x000000000046009f in uwsgi_python_init_apps ()
> #3  0x0000000000452ca3 in uwsgi_init_all_apps ()
> #4  0x0000000000451f2d in uwsgi_start (v_argv=0xb42e460)
> #5  0x0000000000450809 in main (argc=2, argv=0x7fffc0bbe9a8,
> envp=0x7fffc0bbe9c0)
>
>
>
> After detailed analyzing, this might arised under "lazy" + "vhost"
> +"max_requests".
>
> While run as vhost, each procedure in uwsgi_request_wsgi()  will free
> wsgi_req -> appid, that will be visited in an other process, if it exceeds
> maxrequests,  and in the later, this appid is neither reset nor checked
> before strncpy!
>
> (Is appid_len reset to zero needed, after free(wsgi_req -> appid)? )
>
>
> here is my config.xml
>
>
>    <uwsgi>
>        <id>app001</id>
>       <socket>127.0.0.1:9999</socket>
>        <module>webapp:default_action</module>
>       <master />
>       <vhost />
>       <stats>:5050</stats>
>       <pidfile2>/home/yuanfu.zzy/tmp/uwsgixxx.pid</pidfile2>
>       <cache>3000</cache>
>      <!--
>  <daemonize2>/home/yuanfu.zzy/tmp/wxupgrade/uwsgi.log</daemonize2> -->
>       <max-requests>100</max-requests>
>       <processes>4</processes>
>        <threads>10</threads>
>           <enable-threads/>
>        <reload-mercy>4</reload-mercy>
>        <lazy>true</lazy>
>    <!--no-default-app/-->
>   </uwsgi>
>
> again, i've tried higher max-requests, 100 just to boost the crash.
> and here is webapp.py
>
> def default_action(env, start_response):
>   status_code = '200 OK'
>   response_headers = [("Content-Type", 'text/html; charset=utf-8')]
>   response_body = 'ok'
>   start_response(status_code, response_headers)
>   return [response_body]
>
>
> sorry for my poor writing, not my native language though...
> and any advice is appreciated!


Can you retry disabling threads ?

sadly i cannot reproduce the problem, so it could be some kind of race
condition

-- 
Roberto De Ioris
http://unbit.it
_______________________________________________
uWSGI mailing list
[email protected]
http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi

Reply via email to