I have CCR-1036 and running around 800 pppoe sessions and 1200 hotspot users (hosts 2500 in hotspot). After 3-4 hrs the CCR starts giving issues like no new users can ping to gateway (hotspot server) and get login page in hotspot. If we stop the hotspot server for 2-3 minutes and start it again then everything is normal. I suspect there is some limitation in terms of active users or hosts in hotspot. BW consumption of both pppoe and hotspot users is around 450 meg (rx+tx).
CPU utilisation is normal at 15-20%.
All cores of CPU being utilised almost equally.
There are 6-7 firewall rules apart from the ones automatically created while setting up hotspot server. Tried disabling these rules but the issue persist.
Connection tracking is on.
There is no issue to pppoe connections. Issue is with hotspot users.
Maybe your conn. tracking table is filling up or something similar? Try turning the values in ip/firewall/connections/tracking down, especially the tcp established timeout.
Total entries in connections is about 125000. And at the bottom it says max entries 475264. That means its only 25% being used. Will this cause any issue? Also CPE load is around 20. There may be some limitations in hotspot (active and hosts) entries.
This is happening frequently now. Is there any bug in 6.18. We have other CCR-1036 with 6.15 but never had problem of restarting hotspot service. In 6.18 we have to restart hotspot service every few hours. Configuration is same except IP addresses.
One issue noticed - when hotspot stops responding I can’t see user’s MAC address in ARP and hosts (in hotspot). The moment I restart hotspot service the MAC is shown in ARP and Hosts. Also user can login and surf internet.
