Remember that the DHCP in xCAT is quite substantially different from "normal" enterprise DHCP, since it needs to only serve a very specific, well-defined, set of computers (all of which normally will have static reservations), except during discovery.
I would simply use whatever the default value is. Or if you do want to tweak it, go for a very long duration (to reduce traffic, even if it is only by a tiny amount). The other big difference between Enterprise DHCP and xCAT DHCP is that, because your xCAT environment is (or should be) very well-defined, it's very unlikely that the DHCP server just crashes out of the blue. If the DHCP server crashes, you probably have a bigger problem. BTW, a 3 day lease time wouldn't be enough to protect against a weekend outage. The lease time needs to be at least twice the longest expected outage. If you do want to protect against a DHCP failure anyway, you can do a few things: - Configure an automatic restart with a cron script or the like. - Configure the nodes to use a static IP address after bootup. There is a node attribute to do that, but I don't recall what it is off the top of my head. This is actually done in a postscript. _______________________________________________________________________ Kevin Keane | Systems Architect | University of San Diego ITS | kke...@sandiego.edu Maher Hall, 192 |5998 Alcalá Park | San Diego, CA 92110-2492 | 619.260.6859 | Text: 760-721-8339 *REMEMBER! **No one from IT at USD will ever ask to confirm or supply your password*. These messages are an attempt to steal your username and password. Please do not reply to, click the links within, or open the attachments of these messages. Delete them! On Mon, Aug 19, 2019 at 11:06 PM Heckes Frank (CI/OSB4) via xCAT-user < xcat-user@lists.sourceforge.net> wrote: > Hello all, > > > > I’ve a couple of questions concerning the dhcpd configuration of a cluster > runining completely on stateless images. > > > > -1- What is ‘best’ choice for the value of the DHCPLEASE parameter to run > a stateless cluster? > > In case of dhcpd failures is seems to be helpful assigning a higher > value > then the default for at least 3 days seems to reasonable in case the > dhcpd crashes over the weekend and an automated check/restarted > fails.(?) > -2- Does my understanding (dhcpd.leases(5)) and experience is right that > the leases stored > in /var/lib/dhcpd/dhcpd.leases guarantee a persistent state during > dhcpd restarts? > > > > Many thanks in advance. > > > > Cheers, > > -Frank Heckes > _______________________________________________ > xCAT-user mailing list > xCAT-user@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/xcat-user >
_______________________________________________ xCAT-user mailing list xCAT-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xcat-user
