Re: [leaf-user] dhcp (pump) fails or acquire address after network (cable) outage
On Sun, 06 Mar, 2005 at 21:41:17 -0500, Gene Smith wrote: Jon Clausen wrote, On 02/13/2005 03:27 AM: On Sat, 12 Feb, 2005 at 18:00:05 -0500, Gene Smith wrote: snip AFAICT there's a bug which makes pump exit, when no dhcp-server can be reached after N retries. I'm not absolutely sure this is what actually happens, but some googling turned up links to that effect. (Sorry I can't reproduce the search ATM) Working on the *assumption* that pump indeed dies, I threw this together: snip (Un)fortunately the ISP seems to have gotten their act together, at about the same time as I did the above. Hence I don't know whether or not it works as desired... :P On my system I have verified that pump *does not* die. It just seem to quit doing its thing. Yes. I can confirm that this is indeed what happens. I have to kill it and restart it to get my ip addr back. Also, I see no indication in /var/log/syslog that there was a problem other than the lack of the typical slew of messages pump generates when it does a periodic renew. Not even sure ISP was down since syslog indicated that shorewall was rejecting stuff during the time my lease was expired and pump did not run (if that is possible?). Well... Maybe your ISPs dhcp server was down, but your connection was not? Question: How is pump normally started on boot? I am unable to figure out how it starts up after looking through the various files. I see indications that it is somehow tied in with ifup or possibly shorewall startup. Honestly, I don't know. For the problem at hand, however, I now have something which might provide a (different) workaround. It's still just a workaround, and I think we should probably consider getting a newer/better pump, or alternatively switch to dhclient. What I have is two scripts. /sbin/repump which checks; 1; (by way of ping) the reachability of the gateway 2; if the interface has an ip if 1 fails a warning is produced, if 2 fails an error is produced. When more than 3 warnings occur, an error is produced. When the errors reach 3, the interface is restarted (forcing pump to renegotiate). I run it every two minutes from cron, and it seems to be doing it's job. Now, in order for /sbin/repump to know the (correct) gateway address (automatically) I'm making use of a script which was orginally intended to support something else. This script is placed in /etc/network/if-up.d/store_ifinfo and is symlinked from /etc/network/if-down.d/store_ifinfo which means that the script is called when an interface goes up, or down. What is does (going up) is 'intercept' the interface ip parameters which (I discovered) exist as environment vars at this point, and stores them in a file /var/run/ifinfo/$IFACE The scripts are still a bit too rough around the edges for me to 'go public' with them, but if you (Gene) want to, I'll send them. They basically work (for me anyway)... /Jon -- YMMV --- SF email is sponsored by - The IT Product Guide Read honest candid reviews on hundreds of IT Products from real users. Discover which products truly live up to the hype. Start reading now. http://ads.osdn.com/?ad_id=6595alloc_id=14396op=click leaf-user mailing list: leaf-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-user SR FAQ: http://leaf-project.org/pub/doc/docmanager/docid_1891.html
[leaf-user] Re: dhcp (pump) fails or acquire address
Good day all, I had a similar problem with Leaf bering 1.1 and 1.2. I have since upgraded to uClibC 2.2.2. I did find a fix for the older system when I was using it, I replaced PUMP with DHCPD which solved the problem. Darcy Parker Message: 2 Date: Sun, 06 Mar 2005 21:41:17 -0500 From: Gene Smith [EMAIL PROTECTED] Reply-To: [EMAIL PROTECTED] To: leaf-user@lists.sourceforge.net Subject: Re: [leaf-user] dhcp (pump) fails or acquire address after network (cable) outage Jon Clausen wrote, On 02/13/2005 03:27 AM: On Sat, 12 Feb, 2005 at 18:00:05 -0500, Gene Smith wrote: I am running a bering-leaf system with 2.4.18 kernel that I setup about two years ago (not sure of exact version). It has been working fine except for one problem. When the cable goes down and eventually comes back up the bering-leaf system never recovers (clients can't access internet). I tried restarting services (shorewall, networking, ifup/down) to no avail. Usually I just reboot. However I discovered that if I kill and re-run pump (/sbin/pump -i eth1) it then recovers and acquires its IP address. Could I have something configured wrong that prevents a automatic recovery? I have been having similar (if not identical) problems lately: Link goes down, lease expires, link comes up again, pump fails to renew. AFAICT there's a bug which makes pump exit, when no dhcp-server can be reached after N retries. I'm not absolutely sure this is what actually happens, but some googling turned up links to that effect. (Sorry I can't reproduce the search ATM) Working on the *assumption* that pump indeed dies, I threw this together: # cat /sbin/repump #!/bin/sh if [ -z `/sbin/pidof pump` ] ;then /usr/bin/logger Repump: pump looks dead, attempting resurrection; /sbin/pump #else #/usr/bin/logger Repump: pump lives, pid `pidof pump` fi and added: # keeping pump alive: */10 * * * * root/sbin/repump to /etc/crontab. Basically a crude workaround. (Un)fortunately the ISP seems to have gotten their act together, at about the same time as I did the above. Hence I don't know whether or not it works as desired... :P HTH /Jon On my system I have verified that pump *does not* die. It just seem to quit doing its thing. I have to kill it and restart it to get my ip addr back. Also, I see no indication in /var/log/syslog that there was a problem other than the lack of the typical slew of messages pump generates when it does a periodic renew. Not even sure ISP was down since syslog indicated that shorewall was rejecting stuff during the time my lease was expired and pump did not run (if that is possible?). Question: How is pump normally started on boot? I am unable to figure out how it starts up after looking through the various files. I see indications that it is somehow tied in with ifup or possibly shorewall startup. -gene --__--__-- ___ leaf-user mailing list leaf-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-user End of leaf-user Digest --- SF email is sponsored by - The IT Product Guide Read honest candid reviews on hundreds of IT Products from real users. Discover which products truly live up to the hype. Start reading now. http://ads.osdn.com/?ad_id=6595alloc_id=14396op=click leaf-user mailing list: leaf-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-user SR FAQ: http://leaf-project.org/pub/doc/docmanager/docid_1891.html