I have what may be a possible bug in Slackware network init scripts, the result being multiple dhcp clients for a single interface. I came across this while trying to debug a named issue at boot, and have not resolved either problem. I have 4 machines running Slackware, 3 on 10.2, 1 on 10.1.
2 of the 10.2, and the 10.1 machines show multiple dhcpcd clients running for a single interface. 2 of those boxes have two nics, with eth0 being dhcp, and eth1 being statically assigned. All machines were recently upgraded to 2.6.16 from 2.4.x, and all are kept up to date patch-wise. I don't recall if the multiple dhcp client for one interface issue showed up under 2.4.x, so I don't know if this is a kernel issue, or an rc.inet1 script bug...

Anyone else experience the multiple dhcp client for a single interface issue, or know of a fix?

This machine has two nics, eth1 is static IP:
root@dev:~#cat /etc/slackware-version && ps ax|grep dhcp|grep -v grep
Slackware 10.2.0
1241 ? S<s 0:00 /sbin/dhcpcd -t 30 -R -d eth0
1245 ? Ss 0:00 /sbin/dhcpcd -t 30 -R -d eth0

This machine has one nic:
root@ws:~# cat /etc/slackware-version && ps ax|grep dhcp|grep -v grep
Slackware 10.2.0
1460 ? S<s 3:02 /sbin/dhcpcd -d -t 60 eth0
1464 ? Ss 1:22 /sbin/dhcpcd -d -t 60 eth0

This machine has one nic:
root@daw1:/boot# cat /etc/slackware-version && ps ax|grep dhcp|grep -v grep
Slackware 10.2.0
1344 ? S<s 0:00 /sbin/dhcpcd -d -t 10 eth0
(The only odd one of the 4 that seems to work correctly, diffing rc.inet1 with the other 10.2 boxes revealed nothing untoward)

This machine has two nics, eth1 is static IP:
root@rdb:/etc# cat /etc/slackware-version && ps ax|grep dhcp|grep -v grep
Slackware 10.1.0
1223 ? S<s 0:00 /sbin/dhcpcd -R -d -t 10 eth0
1232 ? Ss 0:00 /sbin/dhcpcd -R -d -t 10 eth0

This last machine I've added lines to rc.inet1 at the bottom of the script to start a dhcp server on eth1, and network init seems to start the dhcp server twice, which made me think there's a problem in the logic or for loop used by rc.inet1, but as I've mentioned above, this all seems to have started when moving from 2.4.x kernel to 2.6.x

Thanks

you could look at the logger message in /var/log/messages
to see if they are started at the same time.
( you'll need the log from the time the pc booted)

if so, the script might start it twice.
( don't know if that's possible).

egag

I am running Slackware current with kernel 2.6.12, and I only see one DHCP process for my NIC. I don't have any machines running straight 10.2 to test on, though.

Though this topic does remind me of an apparent DHCP bug I saw in Slackware current on my laptop. At shutdown, I would sometimes get a few errors saying something like "Cannot execute /etc/dhcpc/dhcpcd.exe: permission denied".

Thanks for responding, but none of this helps, perhaps I'll look elsewhere....

I get thesame thing, on occasion. Not always, but about 80% of the time.
Looking into the rc.inet1 ans rc.inet1.conf files to resolve a different issue, also. Get a "eth_up" error too.

I have a working set of files that you can dump into the /etc/dhcpc/ directory at http://www.slackware.com/~alien/dhcpc/

That sounds like a messed-up upgrade from Slackware-10.1 ... or an installed 3rd party package that overwrote rc.inet1 perhaps? The "eth_up" function call was present in rc.inet1 until slackware 10.2 where it got replaced with "if_up".

Eric

No, Eric, straight from your rc.inet1 ,rc.inet1.conf, rc.wireless, and rc.wireless.conf files....

I'm gonna look into it, see what happened, but this is a 'clean' install from 'current', never had 10.1 even waving nearby....

Just looked at rc.inet1, and lo and behold, something put "eth1 up" in there....
A mystery.....Cleaned it out. Now hunky dory for that....Now for dhcpcd weirdness at init 6....

If only I could remember where I saw it mentioned before... there is a 3rd party package that edits rc.inet1 when you install it, and writes eth_up in there. Can you check to see what you installed after slackware-current? Must be a network related package. A wireless network manager perhaps?

Eric