Tool for DHCP pool statistics/monitoring
Trond Hasle Amundsen
t.h.amundsen at usit.uio.no
Tue Apr 15 10:30:49 UTC 2008
Leif Arne Neset <leifa at alfanett.no> writes:
> If you have a lot of pools or a large leasefile, this can generate
> huge load on the server. We did something like you do but, switched
> to a system where we generate a table of of all the pools with usage
> (active leases and pool size) and store it in a cache file. When
> MRTG asks for the numbers for a pool, the server only checks this
> file unless the file is more than 5 minutes old. If the cache file
> is more than 5 minutes old a new file is beeing generated. This
> scales much better. Currently we run this application on a system
> with more than 100.000 leases, 550 pools and about 70 groups of
> pools beeing checked by MRTG every 15 min.
Yep, scalability is a concern to us as well. We will never reach 100k
leases, but we may eventually deal with 12-15k if VOIP really takes on
at the university. I ran my script on our old DHCP server, an antique,
heavily overloaded Solaris box with 5000 leases. It took a while to
finish. On our new DHCP server (really a failover pair), which is a
pretty standard Dell server running RHEL5, my script takes a whopping
0.1 sec to finish, but with less than 300 leases in total.
Scalability can be improved by caching the config, at least. The
leases file is a kind of cache by itself, not much point in caching
that one unless its paired with the pools/ranges from config as you
suggest. It's an interesting idea, I'll look into it.
BTW, thanks for the feedback.
Trond H. Amundsen <t.h.amundsen at usit.uio.no>
Center for Information Technology Services, University of Oslo
More information about the dhcp-users