Keep loosing WAN IP Address - dhclient does not seem to do update

[sullrich]

@Selective:

Sorry, no luck =(
Snapshot 02-14-2007:

–---------------------

Feb 18 11:59:18 dhclient[271]: exiting.
Feb 18 11:59:18 dhclient[271]: exiting.
Feb 18 11:59:18 dhclient[271]: short write: wanted 21 got 0 bytes
Feb 18 11:59:18 dhclient[271]: short write: wanted 21 got 0 bytes
Feb 18 11:59:18 dhclient[271]: DHCPACK from 90.224.168.1
Feb 18 11:59:18 dhclient[271]: DHCPREQUEST on rl0 to 255.255.255.255 port 67
Feb 18 11:58:37 last message repeated 5 times
Feb 18 11:54:02 last message repeated 3 times
Feb 18 11:52:00 last message repeated 3 times
Feb 18 11:51:28 dhclient[271]: DHCPREQUEST on rl0 to 213.67.15.234 port 67
Feb 18 11:41:51 php: : FTP proxy disabled for interface opt2 - ignoring.
Feb 18 11:41:51 php: : FTP proxy disabled for interface opt1 - ignoring.
Feb 18 11:41:51 php: : FTP proxy disabled for interface LAN - ignoring.
Feb 18 11:41:50 check_reload_status: reloading filter
Feb 18 11:41:44 check_reload_status: updating dyndns
Feb 18 11:41:41 php: : FTP proxy disabled for interface opt2 - ignoring.
Feb 18 11:41:41 php: : FTP proxy disabled for interface opt1 - ignoring.
Feb 18 11:41:41 php: : FTP proxy disabled for interface LAN - ignoring.

Do you see anything in the filter logs showing blocked packets around this time?

Also, anyone having this problem, please run this from a shell:

fetch -o /etc/inc http://www.pfsense.com/~sullrich/filter.inc

[Selective]

no, i have not seen any blocked packets that could have something to do with this, but i am one of those
who are affected by the "promiscuous mode disabled", and therefore my firwall logs stops after a while.

isn´t there a way to catch the dhclient message and "restart" dhclient when its terminating, just as a workaround until you pinned down the problem ?

[sullrich]

We already have code to detect this but it doesnt appear to be working.

[sullrich]

Good news!

I was able to replicate this problem thanks to sdale and Vescovo.

Vescovo had the final fix and it has been commited.  Please upgrade to a snapshot about an hour from this message.

[yoda715]

@sullrich:

Good news!

I was able to replicate this problem thanks to sdale and Vescovo.

Vescovo had the final fix and it has been commited.  Please upgrade to a snapshot about an hour from this message.

I think you forgot to wave your magic voodoo wand, problem still exists  ;). I think the problem lies in the rc.newwanip not catching the interface being assigned 0.0.0.0 IP address.

[Selective]

i did a workaround for that problem a couple of months ago:

rc.newwanip:
–---------------------------
if($curwanip == "0.0.0.0") {
        log_error("Failed to update WAN IP, restarting dhclient.");
        exec("dhclient {$interface_real}");
        /* die now, dhclient will kick off rc.newwanip again? */
        exit;
}

[freax]

Is not the real deal but if you disconnect the cable and connect it again you won't have link.
I have both ends with fixed ip (the router and firewall interface) and it should have link again like it should.

Like I said, i think it's a good way to debug whenever you want. Tests always available.

Sorry if i'm wrong.

[Krakke]

Tried this  "fetch -o /etc/inc http://www.pfsense.com/~sullrich/filter.inc"

Now i have to reinstall pfsense  :'(

[sullrich]

I have added a workaround and verified that it does indeed seem to work.

Update to the latest snapshot at least an hour from now.  Currently 11:46PM EST.

[Selective]

im sorry to say this but, the new snapshot (02-18) is not working :(
–------------------------------

Feb 20 23:40:20 php: : rc.newwanip working with IP address 0.0.0.0 wan rl0.
Feb 20 23:40:20 php: : Informational: rc.newwanip is starting rl0.
Feb 20 23:40:19 dhclient: exit_with_hooks()
Feb 20 23:40:19 dhclient: PREINIT
Feb 20 23:40:19 dhclient: netstat
Feb 20 23:40:19 dhclient[21934]: DHCPOFFER from 90.224.168.1
Feb 20 23:40:19 dhclient[21934]: DHCPDISCOVER on rl0 to 255.255.255.255 port 67 interval 8
Feb 20 23:40:19 dhclient: exit_with_hooks()
Feb 20 23:40:19 check_reload_status: rc.newwanip starting
Feb 20 23:40:19 dhclient: PREINIT
Feb 20 23:40:19 dhclient: netstat
Feb 20 23:40:19 dhclient: exit_with_hooks()
Feb 20 23:40:18 dhclient: Deleting old routes
Feb 20 23:40:18 dhclient: EXPIRE FAIL
Feb 20 23:40:18 dhclient: netstat
Feb 20 23:40:18 dhclient: exit_with_hooks()
Feb 20 23:40:18 dhclient: PREINIT
Feb 20 23:40:18 dhclient: netstat
Feb 20 23:18:20 dhclient[280]: exiting.
Feb 20 23:18:20 dhclient[280]: exiting.
Feb 20 23:18:20 dhclient[280]: short write: wanted 21 got 0 bytes
Feb 20 23:18:20 dhclient[280]: short write: wanted 21 got 0 bytes
Feb 20 23:18:20 dhclient[280]: DHCPACK from 90.224.168.1
Feb 20 23:18:20 dhclient[280]: DHCPREQUEST on rl0 to 255.255.255.255 port 67
Feb 20 23:16:07 last message repeated 6 times
Feb 20 23:12:37 last message repeated 7 times
Feb 20 23:10:43 last message repeated 4 times
Feb 20 23:10:13 dhclient[280]: DHCPREQUEST on rl0 to 213.67.15.234 port 67
Feb 20 23:00:37 php: : FTP proxy disabled for interface opt2 - ignoring.
Feb 20 23:00:37 php: : FTP proxy disabled for interface opt1 - ignoring.
Feb 20 23:00:37 php: : FTP proxy disabled for interface LAN - ignoring.
Feb 20 23:00:36 check_reload_status: reloading filter
Feb 20 23:00:30 check_reload_status: updating dyndns
Feb 20 23:00:27 php: : FTP proxy disabled for interface opt2 - ignoring.
Feb 20 23:00:27 php: : FTP proxy disabled for interface opt1 - ignoring.
Feb 20 23:00:27 php: : FTP proxy disabled for interface LAN - ignoring.
Feb 20 23:00:23 check_reload_status: reloading filter
Feb 20 23:00:23 php: : Configuring slbd
Feb 20 23:00:23 php: : Resyncing configuration for all packages.
Feb 20 23:00:19 php: : Creating rrd update script

[sullrich]

I really don't see how it could not work.

Run this from a shell:

cat /sbin/dhclient | grep reload

What does the results say?

[Selective]

it does not say anything:
–----------------------------

Enter an option: 8

cat /sbin/dhclient | grep reload

Binary file (standard input) matches

[sullrich]

Can someone else please test?  This absolutely works for me after 20 hours when normally it would break in 30 minutes…

[Krakke]

@sullrich:

Can someone else please test?  This absolutely works for me after 20 hours when normally it would break in 30 minutes…

Works fine for me, only lost my ip 2 times, and that was before the latest snapshot.

[sullrich]

@Krakke:

Works fine for me, only lost my ip 2 times, and that was before the latest snapshot.

Great!  Almost started to go insane, because I know this fixed my issue.  Please, someone else test?

[Krakke]

@sullrich:

@Krakke:

Works fine for me, only lost my ip 2 times, and that was before the latest snapshot.

Great!  Almost started to go insane, because I know this fixed my issue.  Please, someone else test?

Im sorry sullrich, went out for a while, when i came back

Feb 21 20:01:24 dhclient[300]: DHCPREQUEST on fxp0 to 172.21.249.50 port 67
Feb 21 20:01:24 dhclient[300]: DHCPACK from 172.21.249.50
Feb 21 20:01:24 dhclient[300]: short write: wanted 21 got 0 bytes
Feb 21 20:01:24 dhclient[300]: short write: wanted 21 got 0 bytes
Feb 21 20:01:24 dhclient[300]: exiting.
Feb 21 20:01:24 dhclient[300]: exiting.

[hoba]

Atm there is a workaround in pllace that will restart the dhclient when it exits. Does it get restarted?

[Krakke]

@hoba:

Atm there is a workaround in pllace that will restart the dhclient when it exits. Does it get restarted?

Nope at 20:01:24 it exited, at 21:01 i lost my ip, then i had to do a renew.

[sullrich]

It can take up to 5 minutes to restart dhclient depending on if check_reload_status is running.

So if you still have trouble please check that check_reload_status is running from a shell by doing:

ps awux | grep check_reload_status

Also, cat /etc/crontab | grep check

[sullrich]

I found yet another problem.  This WAN DHCP issue has been nothing but fun!

Try out a recent snapshot atleast an hour from now.