After that, I was able to fetch 1000 mtree records by issuing mtree.list
command over xmlrpc.

it works ofcause. But if response will become greater it wilt failed again.  For example 30K rows htable:

root@sw5 src]# curl -X GET -H "Content-Type: application/json" -d '{"jsonrpc": "2.0", "method": "htable.dump", "params":["big_htable"],  "id": 1 }' http://192.168.10.190:5071/jsonrpc/ > /tmp/big_htable.data 
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
 59 4181k   59 2479k  100    79  13.4M    437 --:--:-- --:--:-- --:--:-- 13.5M
curl: (18) transfer closed with 1743335 bytes remaining to read

Now I have to set tcp_conn_wq_max > 1 743 335.  
for 40K rows htable  tcp_conn_wq_max > 3 153 335  

So i am afraid change default value (32k) so vast according to possible side effects for SIP tcp connections


On 16 August 2018 at 09:13, Juha Heinanen <jh@tutpro.com> wrote:
Henning Westerholt writes:

> The correct spelling should be tcp_buf_write (new name tcp_async). You can
> find this in the core docs.

Thanks.  tcp_async is enabled by default.  I then went and set

tcp_conn_wq_max=256000

After that, I was able to fetch 1000 mtree records by issuing mtree.list
command over xmlrpc.

-- Juha

_______________________________________________
Kamailio (SER) - Users Mailing List
sr-users@lists.kamailio.org
https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users



--
Savolainen Dmitri