Hi,
We have some time-outs when there connecting more thane 1400 listeners
to Icecast.
We use CentOS Linux release 7.6.1810 on port 80 and 443 for SSL on IPv4
and IPv6.
The stream have many mountpoint 16 active (8 main en 8 fallback and 8
file when faillback is down)
The server have 2 CPU's on 3Ghz and 6000mb memory. And we 1 GB uplink
(we use 130mbit on peak). We use a hardware firewall Mikrotik CCR1036
and extra software firewall (Arno's Iptables Firewall)
We use Cacti to check server mem,cpu and load en traffic. We see 34% CPU
and 0,5 load on peak.
We change the CONNTRACK on the server to 32384 and we change some limits
for application in /etc/security/limits.conf
* soft core 8388608
* soft nofile 65000
* hard nofile 65000
We use on Icecast t:
<queue-size>524288</queue-size>
<client-timeout>20</client-timeout>
<header-timeout>15</header-timeout>
<source-timeout>10</source-timeout>
<burst-size>65536</burst-size>
When we have the problem the disconnect mountpoints and we see 16-20%
packet loss to icmp to the IPv4 address. And when we login on ssh the
system is response slower.
We see no error in the system log or Icecast log.
Do anyone have some tips to check or tune the configs?
We have running other servers the have more listeners but no problems
but the have no SSL or so many mountpoints.
Best regards,
Michel
--
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.xiph.org/pipermail/icecast/attachments/20190221/e8bd3fff/attachment.html>
