Hi Folks,
I have received report of this bug behaviour before, but never seen it
happen myself, or been able to get any diagnostic at all until this
afternoon. Thought I would post in case it is useful info!
This has been reported to us previously as "when customer gets DHCP lease,
no access to anything is possible. When customer enters static lease,
everything is good. Reboot router, everything fine again for a while. Then
stops working"
Today, I discovered the reason. After some number of leases has been issued
by DHCP server, some number in the vicinity of about 200 leases, new leases
are issued with bad subnet mask, no gateway and no DNS. End result, of
course, is no connectivity.
This is apparently just related to number of leases, not actual dhcp server
- i.e. the problem happens after the same number of leases are formed even
when there are 3 DHCP servers on 4 different subnetworks, or just one DHCP
server for all clients.
Today, I had a chance to work with a fellow who had done some much better
diagnostic work on a network experiencing essentially the same behaviour -
he discovered that DHCP becomes broken after some time (or some number of
leases) in that clients no longer receive netmask, gateway or DNS detail.
Thus the client picks up IP address only, and a default mask (assume based
on address 'class') and blank gateway and DNS.
The routerOS version in place omn that network is 6.10 - we have upgraded to
6.11 and see if it happens again.
If so, we'll grab some pcaps of DHCP traffic and report to MT - in the
meantime, if anyone has experience with this issue too, please post it! ;)
Cheers, Mike.
