Failed while updating the boot sectors for disk1 partition1
if you are taking about how to pass parameters to a node that will PXE install, you need to edit /tftpboot/pxelinux/pxelinux.cfg/default.- gb From greg.bruno at Thu Sep 13 2007 From: greg.bruno at (Greg Bruno) Date: Thu, -0700 Subject: [Rocks-Discuss] In-Reply-To: /install/...2) Start insert-ethers 3) After the node PXE-boots, if the kickstart fails with "no valid devices found..." error message, then go to step 4.Otherwise, go home :) 4) Go to the compute node and ALT F2 to go to screen #2 (If you only see the graphical installation screen, then use CTRL ALT F1 first to go back into text mode consoles, and then use ALT F2 to switch to console #2). If you don't see .rocks-release, then try deleting your partitions and recreating them and then rebooting.The funny thing is, all this worked fine before and I also reinstalled the nodes a few times before without this problem.Now, ever since this problem started, I can no longer install the nodes. Thanks, Srihari -------------- next part -------------- An HTML attachment was scrubbed...I have a Cluster Rocks 4.2.1-0 on AMD Atlhon (TM) 64 processor 3200 , the motherboard k8V SE Deluxe (ASUS).I want add different machines and its configurations is: AMD Atlhon (TM) 64 processor 5200 dual-core, the motherboard M2NPV-VM (ASUS).
Also, for the benefit of whoever else may try this: - /install/...
-- Bryan Bales Systems Analyst II University of Houston 713-743-0766 From greg.bruno at Thu Sep 13 2007 From: greg.bruno at (Greg Bruno) Date: Thu, -0700 Subject: [Rocks-Discuss] using a RTL8111B based chipset on compute nodes In-Reply-To: /usr/sbin and /sbin/. i'm sorry, but i don't have a good idea what is happening here. also, please send the output of: # rocks list roll - gb From greg.bruno at Thu Sep 13 2007 From: greg.bruno at (Greg Bruno) Date: Thu, -0700 Subject: [Rocks-Discuss] *Really* disable a RPM installed due to dependency In-Reply-To: Is there a way to solve my problem ?
i don't know of a way to really disable a package -- anaconda dynamically calculates the package dependancies and we don't have control over that.
At this point, I went back to the frontend, and deleted compute-1-0 again from the database and restarted insert-ethers. This time it kickstarted again, and proceeded with the normal installation!
So, here is the summary of steps: 1) Remove node from rocks database using insert-ethers and sync the configuration.
By this time, Rocks added the node compute-1-0 to the cluster and showed that it kickstarted. I went back to the node and rebooted it after writing the partition table to the disk. At this point, if you are "lucky", the partitioner will wipe out your partitions and will create the new partitions and start installing. Right after it created the partitions, the installer bailed out again with the same error message. When this happened, I went back again to screen #2, ran fdisk and saw that Rocks created the standard default 5 partitions.