Description of problem: It confuses me that the hotplug system and not the pcmcia system is responsible for the configuration of these cards (for one thing, /var/lib/pcmcia/stab claims the sockets are both empty even when I have a net connection, and for another `cardctl suspend' doesn't seem to be supported. But I digress). The interface `eth1' is a 3Com OfficeConnect Cardbus card configured using `neat' to obtain its address via DHCP. It does this successfully, and `dhclient' thus runs in the background. If I eject this card, the message net.agent[24107]: remove event not handled is logged and the dhclient is not killed. So if I reinsert it then there will be two dhclients running, both thinking they control this interface. The script /etc/hotplug/net.agent contains the claim: # right now it looks like only Gentoo wants to care about # unregistering network devices... This would be a good place to insert something which executes `ifdown' on Red Hat flavoured systems. Unfortunately, ifdown won't run if the HWADDR is specified in the interface configuration because at this point the interface may already have been removed from the system so the address can't be checked. So maybe something more explicit to kill the DHCP client may be in order. (As it happens, `neat' apparently placed a completely bogus HWADDR in my config file, so I removed it and `ifdown' now works.) Version-Release number of selected component (if applicable): hotplug-2004_04_01-1 kernel-2.6.6-1.435 pcmcia-cs-3.2.7-1.5 How reproducible: Always Steps to Reproduce: 1. Configure network Cardbus card to obtain address via DHCP 2. Insert card 3. Remove card Actual results: dhclient left running. Expected results: dhclient killed when network card removed.
Also present in fc3: hotplug doesn't do anything on device removal for network devices, so the interface is never downed. The problem with not doing a proper down on the interface is that things like dhcp or ipsec get left in limbo.
*** Bug 84453 has been marked as a duplicate of this bug. ***
A call to ifdown added in 2004_09_23-1. Note that this may very well not do anything useful without modifying ifdown.
*** Bug 150382 has been marked as a duplicate of this bug. ***
Just bumped into the same issue. Calling ifdown in a remove is not good enough although obviously at least that should be done. If you have dhclient process running on such interface, which is very likely, it lingers and fills logs with messages in this style: .... dhclient: DHCPREQUEST on ath0 to 255.255.255.255 port 67 .... dhclient: send_packet: No such device .... dhclient: DHCPDISCOVER on ath0 to 255.255.255.255 port 67 interval 6 .... dhclient: send_packet: No such device Doh! We know that there is "no such device". Hotplug was already informed. about it.