DokuWiki

User Tools

Site Tools

Trace:
cluster:144

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision 		Previous revision
cluster:144 [2015/12/23 13:46]
hmeij [Warewulf Golden Image] 		cluster:144 [2018/07/26 14:52] (current)
hmeij07 [Deploying]
Line 47: Line 47:
 Set ''rsync'' to work between the two. Unmount any NFS file systems on the node. Adjust the exclusion selections in template file (ie after /home is unmounted on node I do want the mount point...scratch space, etc). Set ''rsync'' to work between the two. Unmount any NFS file systems on the node. Adjust the exclusion selections in template file (ie after /home is unmounted on node I do want the mount point...scratch space, etc).
  
-  * /usr/lib/libexec/warewulf/wwmkchroot/golden-ststem.tmpl+
  
 <code> <code>
  
 # minder: all NFS file systems unmounted? # minder: all NFS file systems unmounted?
 +# or add rsync excludes in
 +# /usr/libexec/warewulf/wwmkchroot/golden-system.tmpl
  
 mkdir /var/chgroots/goldimages; cd /var/chgroots/goldimages mkdir /var/chgroots/goldimages; cd /var/chgroots/goldimages
Line 87: Line 89:
 # switch node to image VNFS # switch node to image VNFS
 wwsh provision set b6 --vnfs=b0.chroot wwsh provision set b6 --vnfs=b0.chroot
 +
 +# Rajil adds for GPU
 +# Keep in mind that nouveau should be disabled,
 +wwsh provision set c038 --kargs=\
 +"acpi_pad.disable=1 console=ttyS0,115200 rd.driver.blacklist=nouveau nouveau.modeset=0" 
  
 # just to be prudent # just to be prudent
Line 118: Line 125:
  
 This is currently not working as expected. In my first attempts I'd specify sda1 (size=500), sda2 (size=2048, type=swap) and sda3 (size=fill) but what I end up with is a standard layout it looks like.  Any sizes are also ignored. So for now I just pick the ones I want (sda1, sda3, sda7). This is currently not working as expected. In my first attempts I'd specify sda1 (size=500), sda2 (size=2048, type=swap) and sda3 (size=fill) but what I end up with is a standard layout it looks like.  Any sizes are also ignored. So for now I just pick the ones I want (sda1, sda3, sda7).
 +
 +Note: this problem turns out to be hardware related, it is not appearing on newer hardware
 + --- //[[hmeij@wesleyan.edu|Henk]] 2017/09/13 08:11//
  
  
Line 155: Line 165:
 ==== Part 2 ==== ==== Part 2 ====
  
-To avoid the problems detailed above I started over with a fresh node.  First I installed CentOS 6.7 vanilla on the hard disk with partitions sda1 (/boot, 500 MB), sd2 (2048 MB, swap) and sda3 +To avoid the problems detailed above I started over with a fresh node.  First I installed CentOS 6.7 vanilla on the hard disk with partitions sda1 (/boot, 500 MB), sd2 (2048 MB, swap) and sda3 (/, size=fill). Since I had a new kernel now we need to make a new bootstrap image and provision it. 
 + 
 +<code> 
 + 
 +  wwbootstrap --chroot=/var/chroots/goldimages/n0.chroot 2.6.32-573.12.1.el6.x86_64 
 +  wwsh provision set n22 --vnfs=n0.chroot --bootstrap=2.6.32-573.12.1.el6.x86_64  
 + 
 +</code> 
 + 
 +Before we create the n0.chroot VNFS image edit the file and comment out these lines, with a golden image approach we do not have a hybridization path but we want these file systems included. 
 + 
 +<code> 
 + 
 +# /etc/warewul/vnfs.conf
  
 # comment out for golden image # comment out for golden image
Line 163: Line 186:
 #hybridize += /usr/include #hybridize += /usr/include
  
 +</code>
  
-  wwbootstrap --chroot=/var/chroots/goldimages/n0.chroot 2.6.32-573.12.1.el6.x86_64 +I also cleaned the following files inside the chroot environment: 
-  wwsh provision set --vnfs=n0.chroot --bootstrap=2.6.32-573.12.1.el6.x86_64 n22 + 
 +<code>
  
 # /etc/fstab, edited (clean also mtab and remove files /dev/disk/by-uuid/* # /etc/fstab, edited (clean also mtab and remove files /dev/disk/by-uuid/*
Line 190: Line 215:
     kernel /vmlinuz-2.6.32-573.12.1.el6.x86_64 ro root=/dev/sda3 rhgb quiet     kernel /vmlinuz-2.6.32-573.12.1.el6.x86_64 ro root=/dev/sda3 rhgb quiet
     initrd /initramfs-2.6.32-573.12.1.el6.x86_64.img     initrd /initramfs-2.6.32-573.12.1.el6.x86_64.img
 +
 +</code>
 +
 +Now when first provisioning happens the right partitions are created and the node imaged. With bootload=EXIT or simply shutting down the warewulf master dhcpd process, the node is now booting from local disk.
 +
 +Yea.
 +
 +==== Deploying ====
 +
 +As part of my deployment I edited out any device information in file ''/etc/udev/rules.d/70-persistent-net.rules'', that information will be filled in on boot. Remade my VNFS. The reason for this is that in migrating from RHEL 5.5 to CentOS 6.7 I observed the order of the NICs to reverse themselves (but not always). Not sure why but will tell me what NIC eth0 will be.
 +
 +Then we build a template file with node specs in it like so:
 +
 +<code>
 +# HP blades reverses HWADDR (don't ask)
 + # use nic port bottom (no changes)
 + # set both ipaddr/hwaddr then PXEboot, wait...ww times out but images
 + # post edit
 + #   /etc/udev/rules.d fix eth? entries per below, then startup scripts
 + #   eth0 down CC 10.10.x.x onboot=no
 + #   eth1 up   CD 192.168.x.x onboot=yes
 + # reboot (with bootload=EXIT)
 + n1  1C:C1:DE:1C:82:D5 192.168.102.10 1C:C1:DE:1C:82:D4 10.10.102.10 10.11.103.10
 + n2  1C:C1:DE:1C:81:01 192.168.102.11 1C:C1:DE:1C:81:00 10.10.102.11 10.11.103.11
 +...
 +</code>
 +
 +My deploy script (give an entire line from above as arguments)
 +
 +<code>
 +
 +#!/bin/bash
 +
 + # deploy a n0.chroot node via PXE golden image transfer
 + # dynamic files are always in stateless CHROOT/centos-6/root/wwtemplates
 + node=$1
 + hwaddr0=$2
 + ipaddr0=$3
 + hwaddr1=$4
 + ipaddr1=$5
 + ipaddri=$6
 +
 + if [ $# != 6 ]; then
 +         echo "missing args: node hwaddr0 ipaddr0 hwaddr1 ipaddr1 ipaddri"
 +         exit
 + fi
 +
 + wwsh object delete $node -y 
 + sleep 3
 +
 + wwsh node new $node --netdev=eth0 \
 + --hwaddr=$hwaddr0 --ipaddr=$ipaddr0 \
 + --netmask=255.255.0.0  --network=255.255.0.0 -y
 +
 + wwsh node set $node --netdev=eth1 \
 + --hwaddr=$hwaddr1 --ipaddr=$ipaddr1 \
 + --netmask=255.255.0.0  --network=255.255.0.0 -y
 +
 + wwsh node set $node --netdev=ib0 \
 + --ipaddr=$ipaddri \
 + --netmask=255.255.0.0  --network=255.255.0.0 -y
 +
 + wwsh provision set $node --fileadd passwd,shadow,group -y
 + wwsh provision set $node --fileadd hosts,bashrc,profile -y
 + wwsh provision set $node --fileadd network.ww,ifcfg-eth1.ww,ifcfg-ib0.ww -y
 +
 +# note: no diskpartition, already exists and fails on this hardware,
 +# otherwise add diskpartion=sda so that "mkbootable" succeeds
 +
 + wwsh object modify -s bootloader=sda $node -y
 + wwsh object modify -s diskformat=sda1,sda4 $node -y
 +
 + if [ "$node" == 'n22' ]; then
 + # golden images with 3 partitions
 + wwsh object modify -s filesystems="mountpoint=/boot:dev=sda1:type=ext4:size=1024,dev=sda2:type=swap:size=6144,mountpoint=/:dev=sda3:type=ext4:size=+" $node -y
 + else
 + # hp blade 4 partitions
 + wwsh object modify -s filesystems="mountpoint=/boot:dev=sda1:type=ext4:size=1024,dev=sda2:type=swap:size=6144,mountpoint=/:dev=sda4:type=ext4:size=+" $node -y
 + fi
 +
 + wwsh provision set $node --vnfs=n0.chroot -y
 + wwsh provision set $node --bootstrap=2.6.32-573.12.1.el6.x86_64 -y
 +
 + wwsh provision set --bootlocal=UNDEF $node -y
 +
 + wwsh pxe update
 + wwsh dhcp update
 + service dhcpd restart
 + service httpd restart
 +
 + echo "now reboot: $node"
 +
 + echo "wwsh provision set --bootlocal=EXIT $node -y"
 +
 +
 +</code>
  
  
 \\ \\
 **[[cluster:0|Back]]** **[[cluster:0|Back]]**
cluster/144.1450896407.txt.gz · Last modified: 2015/12/23 13:46 (external edit)

Page Tools

Except where otherwise noted, content on this wiki is licensed under the following license: CC Attribution-Share Alike 4.0 International
CC Attribution-Share Alike 4.0 International Donate Powered by PHP Valid HTML5 Valid CSS Driven by DokuWiki