Messages in this thread |  | | Date | Sun, 5 May 1996 13:53:59 -0400 (EDT) | From | Greg Patterson <> | Subject | Dial-on-demand in kerneld 1.3.69f. |
| |
This is something I've noticed previously but didn't get around to tracing it all out until this morning. I have set-up init to insure pppd is always up and running at runlevel 5 (dedicated dial-up).
When the modem drops, pppd exits and init restarts everything pretty well with one exception. kerneld begins to use up an extensive amount of CPU resources and today I discovered why. When I installed 1.3.69f, I hadn't realized it had this new Dial-on-demand feature. I found 2-3 /sbin/request-route processes running many times per second. This was happening because I don't have an /etc/ppp/chat.xxx file setup. The last time I found kerneld not idling properly, I killed the process and my system completely froze. I didn't try it this morning. However, when I discovered /sbin/request-route, I edited the script and put a 'sleep 1000' in to slow it down since I don't want to bother with this dial-on-demand stuff.
Getting to my questions:
Why does my system freeze up when I kill the kerneld process?
What options are available to completely disable dial-on-demand? Someone on IRC suggested putting an "alias request-route off" in /etc/conf.modules. I've done this but won't know if it works until my modem connection dies. I could not find any mention of this alias in the source, README's, or on the Modules homepage documents.
Greg.
|  |