Usman Ahmad <
[hidden email]> wrote:
> > it was also found that the secondary dhcp server partition (/var) in which the lease and logs files were writing, it was 100% full.
> > can it also be a reason for dhcp packet disruption?
...
> Yes both are in normal state now.
>
> Thant means the issue happens because the new offered lease time was the different than the lease time recorded in secondary server lease file which wasn't updating due to space issue?
Having the filesystem holding the leases file fill up is virtually guaranteed to create problems. The server will be unable to create entries on disk for both leases it's granted and leases it's been informed that it's partner has granted. It may be that the server will continue to function as all lease information is held in memory - but if restarted it will lack all history from when teh filesystem filled up to when it was restarted.
I see two options here :
1) Just fix the filesystem full problem, and over time everything should sort itself out - whenever a client is given a (new or renewal) lease, that lease will get updated and any out of date or missing records will get superseded and then deleted when the lease file get rewritten. You may get issues in the meantime though.
2) Put the unaffected server into partner down and shut down the affected server. Remove the corrupted leases file and then start the affected server. It should then pull the correct information from the other server and the pair should recover to normal operations. May have to manually cancel the partner down state - dunno about that as I don't personally use failover.
_______________________________________________
dhcp-users mailing list
[hidden email]
https://lists.isc.org/mailman/listinfo/dhcp-users
Locked
