I like the idea but i'm HORRIFIED about people misusing that and actually try to read internal states within their programs :D<br><br><div class="gmail_quote">On Mon, Nov 28, 2011 at 12:02 PM, Michal Ptaszek <span dir="ltr"><<a href="mailto:michal.ptaszek@erlang-solutions.com">michal.ptaszek@erlang-solutions.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">Hey,<br>
<br>
good point, I knew I forgot about something - will add support for that<br>
really soon.<br>
<br>
Kind regards,<br>
<span class="HOEnZb"><font color="#888888">Michal Ptaszek<br>
</font></span><div class="HOEnZb"><div class="h5"><br>
On Nov 28, 2011, at 11:02 AM, Attila Rajmund Nohl wrote:<br>
<br>
> Hello!<br>
><br>
> I really like the idea. But shouldn't this list include the message queue too?<br>
><br>
> 2011/11/28 Michal Ptaszek <<a href="mailto:michal.ptaszek@erlang-solutions.com">michal.ptaszek@erlang-solutions.com</a>>:<br>
>> Hi everyone,<br>
>><br>
>> This idea was born in my mind when debugging some complex, live system<br>
>> and trying to figure out where did all my memory go.<br>
>><br>
>> So, when debugging live system/investigating suspicious memory consumption patterns<br>
>> or simply trying to understand better what's going on with our processes, it might be useful<br>
>> to take a peep at the data given process operates on.<br>
>><br>
>> Right now it is possible to fetch internal gen_* processes state via sys:get_status, we can do<br>
>> some tracing (even using DTrace), we can also check erlang:process_info output and analyze<br>
>> it to become more or less familiar with what is the heap size of our suspect. Still, not all processes<br>
>> are OTP-compatible, and even if: we are going to get only "alive" data coming from process' state<br>
>> (not counting the outdated, not yet garbage collected terms). Also, process_info informs us only<br>
>> about allocated size of the heap, not about the actual usage (although the pre-allocated chunks<br>
>> are not available to the system, yet we might see how far we are from growing/shrinking it).<br>
>><br>
>> Enough with introduction, let's focus on the actual meat: my idea was to create a new BIF,<br>
>> namely erlang:inspect_heap(Pid) that allows us to take a look at any process' heap, fetch the<br>
>> terms residing there and check their actual size. So, for instance:<br>
>><br>
>>> (ejabberd@localhost)12> S = erlang:inspect_heap(pid(0, 358, 0)).<br>
>>> [{[[<<"5">>]|<br>
>>> 284735200226724471091958640173737944785062822211005333957298336375301959844499896296764925551414319236776784],<br>
>>> 20},<br>
>>> {{'$internal_queue_len',0},3},<br>
>>> {{random_seed,{8236,26623,17360}},7},<br>
>>> {{'$ancestors',[ejabberd_c2s_sup,ejabberd_sup,<0.40.0>]},9},<br>
>>> {{'$initial_call',{gen,init_it,6}},7},<br>
>>> {{state,{socket_state,tls,<br>
>>> {tlssock,#Port<0.3936>,#Port<0.3938>},<br>
>>> <0.357.0>},<br>
>>> ejabberd_socket,#Ref<0.0.0.10120>,false,<<"2855118401">>,<br>
>>> {sasl_state,"jabber",<<"<a href="http://pvp.net" target="_blank">pvp.net</a>">>,[],<br>
>>> #Fun<ejabberd_c2s.0.67315917>,#Fun<ejabberd_c2s.1.67315917>,<br>
>>> #Fun<ejabberd_c2s.2.67315917>,cyrsasl_digest,<br>
>>> {state,5,<<"3598825873">>,<br>
>>> {<<"dupa">>,<<...>>},<br>
>>> <<>>,#Fun<ejabberd_c2s.0.67315917>,...}},<br>
>>> true,<br>
>>> {jid,<<"dupa">>,<<"<a href="http://pvp.net" target="_blank">pvp.net</a>">>,<<"hubbard">>,<<"dupa">>,<br>
>>> <<"<a href="http://pvp.net" target="_blank">pvp.net</a>">>,<<"hubbard">>},<br>
>>> <<"Nicknamedupa">>,<br>
>>> {{1322,217197,749816},<0.358.0>},<br>
>>> {1,{<<"dupa">>,nil,nil}},<br>
>>> {1,{<<"dupa">>,nil,nil}},<br>
>>> {1,{<<"dupa">>,nil,nil}},<br>
>>> {xmlelement,<<"presence">>,[],<br>
>>> [{xmlcdata,<<...>>},{xmlelement,...},{...}|...]},<br>
>>> {userlist,none,[],false}},<br>
>>> 564},<br>
>>> {{limits,undefined},3},<br>
>>> {{[],[]},3}]<br>
>><br>
>> gives us a pretty good knowledge on <0.358.0>:<br>
>> • '$_' - OTP + gen_fsm2 process dictionary stuff, 3, 9, 7 words each<br>
>> • random_seed - obvious<br>
>> • {{limits,undefined},3} - internal limits for gen_fsm2 message queue, 3 words<br>
>> • {[], []} - most probably leftovers after fetching user's presence lists<br>
>> • {state, _} - gen_fsm2 state record - 564 words<br>
>> • {[[<<"5">>]|, ...} - sequential tracing tokens? (I'm not very familiar with that, I would<br>
>> say that's something from our rootset<br>
>><br>
>> The implementation is rather simple: if the process we probe is not the caller one (we are not doing<br>
>> erlang:inspect_heap(self()), the data is copied from the callee heap to caller heap (to prevent from having<br>
>> cross-process references in variables), then we compute flat size of the each term we moved. Also, rootset<br>
>> is also included in the summary (i.e. process dict, seq tokens, etc.).<br>
>><br>
>> Code is included in my inspect_heap OTP branch on:<br>
>> github: <a href="https://github.com/paulgray/otp/tree/inspect_heap" target="_blank">https://github.com/paulgray/otp/tree/inspect_heap</a><br>
>><br>
>> I am still a little bit hesitant about suspending process we probe: can anyone tell<br>
>> me if acquiring main process lock would be enough to keep its heap untouched during<br>
>> the call?<br>
>><br>
>> Please, do point any bugs and tell me what do you think about the idea.<br>
>><br>
>> Best regards,<br>
>> Michal Ptaszek<br>
>> _______________________________________________<br>
>> erlang-questions mailing list<br>
>> <a href="mailto:erlang-questions@erlang.org">erlang-questions@erlang.org</a><br>
>> <a href="http://erlang.org/mailman/listinfo/erlang-questions" target="_blank">http://erlang.org/mailman/listinfo/erlang-questions</a><br>
>><br>
<br>
_______________________________________________<br>
erlang-questions mailing list<br>
<a href="mailto:erlang-questions@erlang.org">erlang-questions@erlang.org</a><br>
<a href="http://erlang.org/mailman/listinfo/erlang-questions" target="_blank">http://erlang.org/mailman/listinfo/erlang-questions</a><br>
</div></div></blockquote></div><br><br clear="all"><div><br></div>-- <br>Best Regards,<br>- Ahmed Omar<div><a href="http://nl.linkedin.com/in/adiaa" target="_blank">http://nl.linkedin.com/in/adiaa</a></div><div>Follow me on twitter</div>
<div><a href="http://twitter.com/#!/spawn_think" target="_blank">@spawn_think</a></div><br>