Pipe no leaking
-
I have seeing "System reached maximum login capacity" multiple times after a few weeks of running. After a bit of investigation, it's seems the error message pointing to exhausted of pipeno. After restart and deleting dn rules file. I seeing the number keep increasing way more than my user. Concurrent user hovering around 1500 users.. But now my pipeno is 15000.
For short term solutions how do I restart the captive portal cleanly? Which clear out everything included logon user/pipe no and so on. Planing to do this once a weeks.
Can I call captiveportal_init_rules?
-
Hi,
You found the $pipeno and the way its used in /etc/inc/captiveportal.inc
Find the function captiveportal_get_next_dn_ruleno(….) - this is where a new rule set is allocated when a user connects.
The array size is 64500 - 2000 = 62500A couple of line lower, you will find the function captiveportal_free_dn_ruleno($ruleno)
This one should be called when a user gets disconnected.You should look why, on your setup, "captiveportal_free_dn_ruleno($ruleno)" isn't called, or doesn't do it's job.
I propose you add a 'trace/log' call in the beginning of each function, so you can see in your log when a 2-rule-set is allocated, and when they are destroyed.
-
I do the trace already but still not find where it's leak. Because I got much big issues which is lighttpd crash. Happen 2-3 times a day.. Will continue this 1 when I found solutions for the crash.
-
For info : the new version (2.2.6) has a new httpd version which includes bug fixes.
