Benoit Chesneau
Benoit Chesneau
yes :) On Fri, 16 Oct 2015 at 08:38, Silviu Caragea [email protected] wrote: > Hello, > > The new pool management is part of this release ? > > Silviu...
main difficulty with it is that it's introducing a breaking change for those who're using them , so it will have to be in the next major release I guess....
@svileng can you eventually provide a trace of your connections. It seems weird to return connect_timeout the first time. If the pool is empty or no host is already connected...
@svileng you can start to trace with `hackney_trace:enable/2`: ```erlang hackney_trace:enable(max, io). ``` Put that line before starting all your requests and it should log to outpu. `io` can be replaced...
@svileng @desoulter thanks, looking at it. I will revisit that issue sometimes tomorrow and provide more feedback @svileng this looks standard. if you see anything weird related to the timeout...
@svileng sorry i missed the initial notifications. I will have a look today. Sorry for the delay
As a follow-up to our discussion on slack, just wanted to add more infos. Current pool is checking if the socket is still active before returning it to the client....
i think i'm agree with such solution, i aml also thinking a socket shouldn't be open too long in the pool to not exhaust the server resources or make it...
@gutiago btw sorry for the delay to answer I was side-tracked too much...
tasks are sill valid let keep it open