[erlang-questions] exec-port crashing

[Alberto Rosso]

In our current architecture the use of standard IO is limited and the heavy
traffic goes through c-node sockets. I'm going to investigate in detail the
memory corruption.
Having a comparison between TCP and stdin/err pipes would be interesting
indeed.

I'm also interested in finding a more fault tolerant way to manage external
processes: Michael the solution you implemented is interesting although
difficult to integrate within our system which is not intended to supply
services.

Thank you all



On 3 December 2015 at 08:19, Michael Truog <mjtruog@REDACTED> wrote:

> On 12/02/2015 10:20 PM, Max Lapshin wrote:
>
>>
>> Sergey, have you compared TCP localhost vs stdin/stdout pipe?
>> Throughput and latency.
>>
>> TCP on localhost should be better due to the atomic send
> being limited by the MTU on the localhost interface
> which is normally 16436 on Linux, when compared to the
> atomic send on a pipe being limited by PIPE_BUF which
> is normally 4096.  There is a definite improvement
> when using UNIX domain sockets instead of INET domain
> sockets, so hopefully
> https://github.com/erlang/otp/pull/612 will make its
> way in sometime.
>



-- 
Alberto Rosso
save a tree: please don't print this email unless strictly necessary

L'esperienza e la filosofia che non generano l'indulgenza e la carità sono
due acquisti che non valgono quello che costano. (Alexandre Dumas figlio)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://erlang.org/pipermail/erlang-questions/attachments/20151203/f7e07911/attachment.htm>