Increase DHCP Timeout?

[Allyn Crowe]

I have a situation where I am working on a network with a very slow DHCP response (~1min).

We're seeing that connman autoassigns an IP on boot, but never retries to get a new IP. If we restart connman via pkill -HUP connmand && /usr/sbin/connmand -W nl80211,wext then connman picks up the assigned ip (shown via list-services), but does not update ifconfig

I'm wondering if there is a way to increase the dhcp wait/timeout value to account for this very long delay.

Or if there is any particular reason that connman seems to ignore the new IP after it should have received it...

Specific Info:
original connman version: .69 (also atempted connman .78)
meego 1.2.0

Any help would be appreciated...

[Marcel Holtmann]

Hi Allyn,

Post by Allyn Crowe
I have a situation where I am working on a network with a very slow DHCP response (~1min).
We're seeing that connman autoassigns an IP on boot, but never retries to get a new IP. If we restart connman via pkill -HUP connmand && /usr/sbin/connmand -W nl80211,wext then connman picks up the assigned ip (shown via list-services), but does not update ifconfig
I'm wondering if there is a way to increase the dhcp wait/timeout value to account for this very long delay.
Or if there is any particular reason that connman seems to ignore the new IP after it should have received it...

I leave this for the other to dissect in detail, but essentially if a
DHCP response takes too long, then at some point we are just going ahead
with a link-local address. However if some time later we finally get a
DHCP response, we should upgrade our link-local to the DHCP provided
address. This might be a bug here in our handling. Maybe we are not
trying hard enough to continue with DHCP. But as I said, I leave some
others to dissect this.

However, your DHCP server is seriously bad. No idea how in these day and
age, anybody can have such a bad server ;)

Regards

Marcel

[Allyn Crowe]

Hi Marcel

Post by Marcel Holtmann
Hi Allyn,

Post by Allyn Crowe
I have a situation where I am working on a network with a very slow DHCP response (~1min).
We're seeing that connman autoassigns an IP on boot, but never retries to get a new IP. If we restart connman via pkill -HUP connmand && /usr/sbin/connmand -W nl80211,wext then connman picks up the assigned ip (shown via list-services), but does not update ifconfig
I'm wondering if there is a way to increase the dhcp wait/timeout value to account for this very long delay.
Or if there is any particular reason that connman seems to ignore the new IP after it should have received it...

I leave this for the other to dissect in detail, but essentially if a
DHCP response takes too long, then at some point we are just going ahead
with a link-local address.

Correct that is the behavior we see...

Post by Marcel Holtmann
However if some time later we finally get a
DHCP response, we should upgrade our link-local to the DHCP provided
address.

This is the behavior that is not happening. If we restart connman then it will pull in the appropriate DHCP address

Post by Marcel Holtmann
This might be a bug here in our handling. Maybe we are not
trying hard enough to continue with DHCP. But as I said, I leave some
others to dissect this.
However, your DHCP server is seriously bad. No idea how in these day and
age, anybody can have such a bad server ;)

The joys of corporate networks and remote services ;) Many times they inject a DHCP delay to ensure the devices is fully online before receiving the address (spanning tree delays, etc)

Post by Marcel Holtmann
Regards
Marcel