Install ESX3 with HP RDP/Altiris

[EMcCloud]

Hello All,

I cannot get a scripted install of ESX3 going in RDP. This is killing me. It bombs out at a failure to logon to FTP (Failed to log into xxx.xxx.xxx.xxx: Failed to connect to the FTP server). If I check the IIS logs, I don't see the install session connecting, but I do see connection logs if I use my desktop browser to the install files. Seems not to be on the network when at the "Create boot environment" portion of the job.

I've tried it on two different Altiris servers, same result.

Portfast is enabled on the network port.

I've seen this type of issue a few times on this board but no resolution. Anyone get past this issue?

Hardware: DL360, this is a test box.

Kickstart file: (IP addresses have been altered to protect the wicked)

\# Auto-Generated Scripted Install Configuration file.

\# This file is used for VMware ESX Server Scripted Install Deployment

\# Installation Method

url --url ftp://anonymous:rdp@%#*"select tcp_addr from aclient_prop where computer_id=0"%/dslib/osdist/vmesx301/

\# root Password

rootpw --iscrypted $1$uAzNhSFK$KNsygGXrQC3giODj78HLJ0

\# Authconfig

auth --enableshadow --enablemd5

\# BootLoader ( The user has to use grub by default )

bootloader --location=mbr

\# Timezone

timezone America/Detroit

\# X windowing System

skipx

\# Install or Upgrade

install

\# Text Mode

text

\# Network install type

\# The hostname string below is replaced with the displayed

\# computer name in the Altiris DS console during deployment.

network --bootproto static --ip 10.200.42.xxx --netmask 255.255.255.0 --gateway 10.200.42.xxx --nameserver 10.200.25.xxx --hostname %#*"select replace(\[name],' ','') from computer where computer_id=\{ID}"% --addvmportgroup=0 --vlanid=0

\# Language

lang en_US

\# Langauge Support

langsupport --default en_US

\# Keyboard

keyboard us

\# Mouse

mouse none

\# Reboot after install ?

reboot

\# Firewall settings

firewall --disabled

\# Clear Partitions

clearpart --all --initlabel --drives=sda

\# Partitioning

\# %hddevice% is replaced with the detected storage device name by the

\# vmesx.sh script executed on the target server.

\# To specify specific or custom device names simply replace %hddevice%

\# with specific device names ( Ex cciss/c0d0 ).

part /boot --fstype ext3 --size 250 --ondisk %hddevice% --asprimary

part / --fstype ext3 --size 5110 --ondisk %hddevice% --asprimary

part swap --size 1600 --ondisk %hddevice% --asprimary

part /var --fstype ext3 --size 3072 --ondisk %hddevice%

part /home --fstype ext3 --size 2048 --ondisk %hddevice%

part /tmp --fstype ext3 --size 2048 --ondisk %hddevice%

part /opt --fstype ext3 --size 4096 --ondisk %hddevice%

part None --fstype vmkcore --size 100 --ondisk %hddevice%

part None --fstype vmfs3 --size 10000 --grow --ondisk %hddevice%

\# VMware Specific Commands

vmaccepteula

vmlicense --mode=server --server=27000@10.200.98.xxx --edition=esxFull

%packages

@base

#Post install tasks

%post

\# Transfer the Altiris agent along with its config files

mkdir /tmp/altiris

cd /tmp/altiris

ftp -n <<EOF2
open %#*"select tcp_addr from aclient_prop where computer_id=0"%
user anonymous rdp
cd /dslib/osoem/altiris
binary
prompt
mget altiris*.i386.bin
mget adlagent.conf.custom
mget adlagent.conf.default
exit
EOF2

AltirisConfDir=/opt/altiris/deployment/adlagent/conf
\# Create script to configure ESX and install adlagent (called by rc.local)
\# Using echos to thwart post section script variable and command substitution
echo '#!/bin/bash' >> ./hpinstall.sh
echo '# Script to configure ESX and install adlagent. Called from rc.local.' >> ./hpinstall.sh

echo '# RDP install log file' >> ./hpinstall.sh
echo 'logfile=/root/install.rdp.log' >> ./hpinstall.sh
echo '# Create vmfs filesystem' >> ./hpinstall.sh
echo 'vmfsqueuedir="/etc/vmware/vmfs3queue"' >> ./hpinstall.sh
echo 'filecount=$(ls -1A /vmfs/volumes | wc -l)' >> ./hpinstall.sh
echo '# Check for existing vmfs volumes' >> ./hpinstall.sh
echo 'if \[ $filecount -eq 0 ]; then' >> ./hpinstall.sh
echo ' # No current vmfs volumes' >> ./hpinstall.sh
echo ' # Check vmfs fs creation queue in case ESX is waiting to build on next boot' >> ./hpinstall.sh
echo ' if [\[ -s $vmfsqueuedir ]]; then' >> ./hpinstall.sh
echo ' # Items in queue' >> ./hpinstall.sh
echo ' echo vmfs fs queue contains data - no vmfs created >>$logfile' >> ./hpinstall.sh
echo ' else' >> ./hpinstall.sh
echo ' # Nothing in queue' >> ./hpinstall.sh
echo ' # All clear to go ahead and create vmfs fs' >> ./hpinstall.sh
echo ' # Create vmfs fs' >> ./hpinstall.sh
echo ' vmfsdevice=`fdisk -l | grep %hddevice% | grep fb | cut -d" " -f1`' >> ./hpinstall.sh
echo ' partnum=$\{vmfsdevice:(-1)}' >> ./hpinstall.sh
echo ' vmfspart=`esxcfg-vmhbadevs | grep %hddevice% | cut -d" " -f1`' >> ./hpinstall.sh
echo ' echo "Creating vmfs fs on $vmfspart:$partnum" >>$logfile' >> ./hpinstall.sh
echo ' vmkfstools -C vmfs3 -S localvmfs $vmfspart:$partnum' >> ./hpinstall.sh
echo ' fi' >> ./hpinstall.sh
echo 'else' >> ./hpinstall.sh
echo ' # vmfs volumes exist' >> ./hpinstall.sh
echo ' echo vmfs fs volumes exist - no vmfs created >>$logfile' >> ./hpinstall.sh
echo 'fi' >> ./hpinstall.sh

cat >> ./hpinstall.sh << EOF1

\# Firewall Configuration
\# Enable adlagent and file transfer ports
\# You need to set a static port ("4300" in this example) for file transfer in
\# the deployment console under Tools->Options->Global
esxcfg-firewall --openPort 402,tcp,out,adlagent
esxcfg-firewall --openPort 4300,tcp,out,adlagentFileTransfer

\# Install Altiris Adlagent
cd /tmp/altiris
chmod +x altiris-adlagent*.bin
./altiris-adlagent*.i386.bin 1>>/root/install.rdp.log 2>>/root/install.rdp.log
\# Install adlagent custom configuration
if \[ -e adlagent.conf.custom ]; then
mv $AltirisConfDir/adlagent.conf $AltirisConfDir/adlagnet.conf.bak
cp -f adlagent.conf.custom $AltirisConfDir/adlagent.conf
elif \[ -e adlagent.conf.default ]; then
mv $AltirisConfDir/adlagent.conf $AltirisConfDir/adlagent.conf.bak
sed -e "s/0.0.0.0/%#*"select tcp_addr from aclient_prop where computer_id=0"%/g" adlagent.conf.default > $AltirisConfDir/adlagent.conf
fi
\# Reset adlagent to pick up config if necessary
/etc/init.d/adlagent stop
/etc/init.d/adlagent start

\# Reset rc.local to original
mv -f /etc/rc.d/rc.local.sav /etc/rc.d/rc.local
EOF1

\# make hpinstall.sh executable
chmod +x /tmp/altiris/hpinstall.sh

\# save a copy of rc.local
cp /etc/rc.d/rc.local /etc/rc.d/rc.local.sav

\# add hpinstall.sh to rc.local
cat >> /etc/rc.d/rc.local << EOF

cd /tmp/altiris

/tmp/altiris/hpinstall.sh

EOF

%vmlicense_text

Message was edited by:

EMcCloud

[sbeaver]

Try hardcoding the ftp address in the cfg file

[EMcCloud]

That's how I did it originally and same result.

[DFATAnt]

I don't use Altiris, but I do use HP hardware and I am having similar problems. I get messages saying the ftp server is not available (or something similar) in a dialog box. By clicking on OK many times, the scripted install eventually kicks off. It seems to me that there is an issue with DHCP supplying an address, but when the DHCP address is finally received everything works fine.

I have not been able to figure out what is causing the delay, but another person experiencing the same problem pointed me at the following link which might help you as well

http://fedoraproject.org/wiki/AnacondaNetworkIssues

There is mention of problems with the onboard Broadcom NICS in the HPs as well as mentioning delays with DHCP.

I hope this helps

Ant

[sbeaver]

I actually hard code all the IP's in my configs and do not see this issue. Can you try to hard code the IP?

[DFATAnt]

I could try that, but I'm sure where to begin. Any help with doing that would be greatly appreciated.

Ant

[sbeaver]

network --bootproto static --ip=xx.xx.xx.xx --netmask=255.255.255.0 --gateway=xx.xx.xx.xx --nameserver=xx.xx.xx.xx --hostname=SERVER3.DOMAIN.COM --addvmportgroup=1 --vlanid=0

[DFATAnt]

Thanks. I've already got that part working. I've been testing scripted installs using either FTP or NFS including getting the ks.cfg file from the network and to get access to the network initially the server needs to get a DHCP address. It's this part that is timing out for around 30 seconds before the DHCP address is allocated and is requiring intervention to get the process working.

Ant

[ilatimer]

I have seen this problem many times and it usually comes down to the network port settings not being set to portfast. Even with portfast enabled on the network port I have still had this problem with gigabit switch ports. In one case we ended up connecting the server to a 100MB switch port for the installation and then changed it to gigabit later. Just something to try.

[DFATAnt]

Thanks, I'll give it a go. I know the ports have been set to portfast, and as it is a gigabit switch they are set to auto detect.

Ant

[Matt4]

Here's a GUI Patcher out of Germany that I've had excellent success with!

esxPatcher

http://www.mightycare.de/index.php?option=com_docman&task=cat_view&gid=16&Itemid=69

[CPetro]

Thanks, I'll give it a go. I know the ports have

been set to portfast, and as it is a gigabit switch

they are set to auto detect.

Old versions of SystemImager used to have a 30 second timeout for ARP tables to get set up and stable &&etc.

Some switches just take longer to do this than others.

[mldmld]

Hi,

I get this error one time.

\- RDP is a customization of Altiris by HP. HP give an ISO file of it with everything already installed and the right directories in place

\- IBM, DELL, ... give "setup.Exe" to add their scripts in Altiris.

So if you use Altiris, it's hard to recover all the directories and virtual directories in IIS.

I think, if you look in IIS that the virtual directory is not here.

In my company, we decided to install RDP, then to install altiris on it.

We are still testing, but it seems to work.

ML

[DFATAnt]

The problems I have been experiencing have been on my DEV network. I have done the same on our production network (which uses brand new Nortel switches) and I get the same thing, although the timeout doesn't seem to be as long.

I have also tried using NFS instead of FTP and I am still prompted with a message saying that the NFS mount point cannot be found (even though the ks.cfg file has already been obtained from the same NFS share). If I press OK, I get prompted for the NFS share information. Without changing any of the information, and pressing OK to continue again, the install starts to work.

I am hoping that someone has found a way around the manual intervention for either FTP or NFS.

Thanks

Ant

[J-D]

is it solved now?

Might just be that your ftp is not setup to allow anonymous access?

I had no problems with ftp using HP RDP 3.5.

[EMcCloud]

Hey everyone, Got sidetracked on further VI3 testing.

Anonymous access to the FTP server works fine. I tested that from my laptop without a problem.

Had to come back to this. As a new wrinkle, I decided to take Altiris out of the mix for now. When I pair down the ks.cfg file for a CD/Floppy install, I'm getting an error about the partitioning:

The following errors occurred with your partitioning:

Please specify a mount point for this partition.

Press 'OK' to reboot your system.

I found this: http://www.vmware.com/community/message.jspa?messageID=464989

and changed all the --ondisk sda to --ondisk=sda

No luck.

Not sure if this is the real root of the problem, or if this is something different.

Message was edited by:

EMcCloud

[DFATAnt]

I use HP servers, so hopefully the information I provide you will be the answer to this one. Try using the following:

--ondisk cciss/c0d0

HP's seem to use different labeling to address their disks.

Also, if you want to clear the partitions before using them, try the following:

clearpart --all --initlabel --drives=cciss/c0d0

Ant

[EMcCloud]

I'll give that a shot. I'm using the sda because that is what the kickstart script builder used when I created the file from an existing server.

[EMcCloud]

Ok, changing my disk references to cciss/c0d0 worked, the install went through. (Thanks DFATAnt!)

However, now I can find what the IP address is assigned to. It will not talk to the network. I can only get to it through iLo.

I tried reinstalling and adding --device eth0 to the networking, same result.

I pinged the ip address (ping 10.x.x.x -t) of my server, then moved the known good cable to every NIC on the server, with no response. ???

Should I be using --device vswif0 instead?

[EMcCloud]

Ok, using the commands:

esxcfg-nics -l

lspci

esxcfg-vswitch -l

I was able to see that the NIC that had the connection I wanted to be the service console was vmnic2, and vmnic0 was automatically assigned to the SC.

Using esxcfg-vswitch -L vmnic2 vSwitch0 and esxcfg-vswitch -U vmnic0 vSwitch0 I was able to shuffle around the NICs so I had what I wanted.

Now working on adding those commands to the script.

I'm just gong to keep putting in my steps here so others can see what is going on. Feel free to make suggestions and point out my (many) mistakes. No matter how much Vmware can make me feel like a noob, its nice to learn something new, and its great to have a community like this one to help you learn it.