If I do xnba the node boot loops, with pxe it doesn't loop but pxe.pm had to be edited to correct the "LOCALBOOT 0" issue. My xnba looks just like yours. I think I have the install working over pxe now, at least the first node appears to be alright thus far. It's looking like I can't use xnba for these nodes, at least not without some tweaking. I'll report back if I get everything working. Editting the pxe.pm to output the right stuff wasn't difficult though.
On Thu, 2016-08-25 at 11:51 -0500, Russell Jones wrote: > I don't get "Localboot 0" in the node's pxe file unless I have the > node > defined as netboot=pxe. In that case it writes out to > /tftpboot/pxelinux.cfg... > > If I have the node defined as "netboot=xnba", then the following is > the > contents of my pxe file for a diskfull node that is told to boot: > > #!gpxe > #boot > exit > > > Maybe double check that the nodes are indeed defined as xnba, and > check > the dhcpd.leases file to make sure they are grabbing the write pxe > file? > > > > On 8/25/2016 7:43 AM, Mark Potter wrote: > > > > I am currently deploying a small cluster, remotely and am having an > > issue where xnba is causing a boot loop. Using RHEL 6.8 and xCAT > > 2.12.2 > > > > The nodes install just fine using xnba, and the node file in > > /tftpboot/xcat/xnba/nodes is updated properly. However when the > > node > > reboots it retrieves the node file during the pxe process > > successfully, > > pauses, shows "No more network devices" and reboots. If I switch > > the > > node to pxe it hangs unless I replace "LOCALBOOT 0" with "kernel > > chain.c32 > > append hd0". Once I do that the node does boot. I could change > > pxe.pm > > to produce what's needed but using standard pxe instead of xnba > > causes > > the node to hang during install. > > > > I have searched the list and have seen similar issues but didn't > > find a > > solution. It doesn't matter whether I am installing via pxe or xnba > > so > > a solution to either would be acceptable at this point. I don't > > have a > > preference but the added work of changing the boot method for each > > node > > and having to document the necessity of changes for node > > redeployment > > make the workaround untenable. > > > > ------------------------------------------------------------------- > ----------- > _______________________________________________ > xCAT-user mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/xcat-user -- Mark L. Potter <[email protected]> PCPC Direct Ltd. ------------------------------------------------------------------------------ _______________________________________________ xCAT-user mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/xcat-user
