Topic: Out of free connection

Ok, my syslog looks about the same as the guy who posted a similar post, but never received a reply.

http://www.chillispot.info/chilliforum/ … php?id=259

I am running this on DDWRT-v24 sp2

This started fairly suddenly and was not having issues (at least not this frequent). After the router runs for several hours I usually start to see a few error messages saying "Peer protocol is not defined". Shortly after I will start to see MAC address come through that are requesting DHCP leases, but I am greeted by error messages like the following:

2010-06-28 00:00:56    Daemon.Info    10.255.255.3    Jun 28 06:00:56 chillispot[430]: chilli.c: 1088: Rereading configuration file and doing DNS lookup


2010-06-28 00:01:01    Daemon.Notice    10.255.255.3    Jun 28 06:01:01 chillispot[430]: chilli.c: 3513: New DHCP request from MAC=00-02-6F-59-45-F2


2010-06-28 00:01:01    Daemon.Error    10.255.255.3    Jun 28 06:01:01 chillispot[430]: chilli.c: 1190: Out of free connection


2010-06-28 00:01:01    Daemon.Error    10.255.255.3    Jun 28 06:01:01 chillispot[430]: chilli.c: 3532: Failed to allocate connection


2010-06-28 00:01:01    Daemon.Notice    10.255.255.3    Jun 28 06:01:01 chillispot[430]: chilli.c: 3513: New DHCP request from MAC=00-02-6F-59-45-F4


2010-06-28 00:01:01    Daemon.Error    10.255.255.3    Jun 28 06:01:01 chillispot[430]: chilli.c: 1190: Out of free connection


2010-06-28 00:01:01    Daemon.Error    10.255.255.3    Jun 28 06:01:01 chillispot[430]: chilli.c: 3532: Failed to allocate connection


2010-06-28 00:01:01    Daemon.Notice    10.255.255.3    Jun 28 06:01:02 chillispot[430]: chilli.c: 3513: New DHCP request from MAC=00-02-6F-59-45-F8

Restarting the router does seem to fix the problem.


I was wondering if the problem might be from running out of IP addresses to lease but figured I would receive a different error if this was the case. Can chilli only make or keep so many active connections at a time, independent of the amount of DHCP leases?

Any advice would be very helpful, there seems to be precious little information on the subject.


Thanks

Re: Out of free connection

OK, I have changed the network on one of the routers from 192.168.182.0/23 to add some more hosts to see if that addresses the problem, although it would be nice, I have my doubts.

The only real information I have found searching around for these errors comes from the source http://chillispot.sourcearchive.com/doc … ource.html

int static newconn(struct app_conn_t **conn)
{

  if (!firstfreeconn) {
    sys_err(LOG_ERR, __FILE__, __LINE__, 0,
          "Out of free connection");
    return -1;
  }

and

 if (options.debug) printf("New DHCP connection established\n");

  /* Allocate new application connection */
  if (newconn(&appconn)) {
    sys_err(LOG_ERR, __FILE__, __LINE__, 0,
          "Failed to allocate connection");
    return 0;
  }

It looks like, there are error messages pertaining to not being to allocate a free IP, while I am guessing some part of the connection process is what is giving me trouble here.

Re: Out of free connection

Hmm, after watching the logs more it seems that a "Peer protocol not defined" tends to appear more often and the other messages have stopped appearing on the one router running chillispot for some reason. Of course I can find little information into what this actully means. If anyone has any clues or advice, please let me know.

Thanks smile

Re: Out of free connection

Ok, for anyone that was interested or finds this, it looks like I might have the error resolved, also I am running chilli on a WRT54gl running ddwrt.  I found out that I had someone trying to brute force the router from the outside and resolved that issue. I turned off certain services to free more ram and I would still encounter the same problem.  Finally I noticed that around every five minutes a group of 16MAC addresses would try to request a DHCP lease, and then tried to release it.

These MAC addresses were malfunctioning Access Points that were set to static IP addresses. After a firmware update they stopped requesting the leases and the problem has at least delayed itself over 14hours now which is quite an improvement.

Since the other routers have had this error once or twice I am thinking that there is some other issue at work, but if its only every few days I can write a script to just restart the service or reboot them in the middle of the night.