Best Linux distro for FOG trunk

[Wayne Workman]

@tprice

1. Troubleshooting why the PXE boot menu is not loading? is this due to the TFTP session is actually timing out or something else.

You’re using undionly.pxe ? Can you try undionly.kkpxe ?

2. What is the best combination for installing FOG i.e. is Ubuntu better than Centos?

Depends on who you ask. If you ask me, I think CentOS 7 has less problems with fog. Fedora is also really solid, but it’s more cutting edge.

[tprice]

Hi Uncle Frank, the options for STP & RSTP are enabled at a global level. I will follow your suggestion and post results.

Thanks

[tprice]

Hi there, I am using undionly.kpxe in the Windows DHCP server.

The result from PCAP show now response from FOG server or anything related to port 69.

Regards
,timeout_issue.pcap

[Sebastian Roth]

In the packet dump I can see two DHCP servers answering (IP .54.1 and .54.253). Both offering the same information - so I guess this is configured for redundency. They point the client to another server on a different subnet (IP .50.45). This should be alright as the gateway/router information in the DHCP packet seams fine - if this really is your FOG server. Or is it only serving TFTP??

One thing I wonder about is the filename offered to the client: pxeboot.0

As a quickfix you can create a link to that file on your TFTP server:

cd /tftpboot
ln -s undionly.kpxe pxeboot.0

The result from PCAP show now response from FOG server or anything related to port 69.

If you want to see that traffic you need to capture on .50.45!

[Scott B]

@tprice said:

2. What is the best combination for installing FOG i.e. is Ubuntu better than Centos?

Just my personal preference, but I’ve always been happy with FOG on Ubuntu and formerly Fedora.

[sudburr]

We started off with Ubuntu Desktop, then Server (no gui), then I switched up to Debian Server, and now I’m running CentOS 7 no GUI on a minimal server install for my development server. Whatever I’m using for dev eventually makes it out to our production/live servers.

IMO Debian is head and shoulders better than Ubuntu, and I’ve found CentOS to be the most stable and trouble free of the bunch.

[Sebastian Roth]

@tprice Did you get this solved?

[tprice]

Hi there, thanks for the response. I need to figure out why the tftp response is looping through my the switch.

It appears the IP Helper on the switch’s are enabled with port 0 relaying to our DHCP server.

[Sebastian Roth]

What do you mean by “port 0 relaying”? Sounds a bit like it does not work. But I am pretty sure you would be fine with the quick fix (linking pxeboot.0). Did that work for you?

[tprice]

Hi there, I did not get it solved. But I think the issue within my Switch configuration. I have nine vlans. I have attached the Fog server on the vlan which is used by the workstations (.54) and the DHCP server is on a different vlans (.50).

I can ping the qdn, fqdn and reverser ping the FOG server within the main vlans.

The fog server is sitting on a Hyper-v host. The virtual switch is bridging the physical NIC.

So I need to figure out why the switch is answering the TFTP request. I suspect the ip helper and DHCP relay on the switch is the root cause.

Troy

[KKTwenty101]

Another consideration on hyper-v is integration services. Ubuntu is not very integration services friendly (NIC and SMP issues mainly). Centos is a better choice for hyper-v guests.

[tprice]

@Sebastian-Roth Hi there, I did not get this solved. I need to look at my switch and TFTP communication.

[Wayne Workman]

@tprice You stated that your fog server is on the .54 vlan, but @Sebastian-Roth pointed out that your redundant DHCP servers are both giving out a .50.45 address for option 066 and pxeboot.0 for option 067.

I think you have a DHCP mis-confguration, I don’t think there is anything wrong with your switches or fog server.

I don’t think pxeboot.0 has been used since .32 and older, so I assume you’ve had fog working before.

[tprice]

Celebration. It works. The things that I have changed was a complete new build of Centos 7, FOG (Version: 7306), and followed new documentation. This allowed me to register a machine and perform inventory but I could not start an upload (could not load images directory (/bin/fog. Upload).

Did some check regarding services and found NFS was not running. Once the latter was started, the ability to perform the partclone was successful.

I am very happy, thank you all

[Wayne Workman]

@tprice What new documentation (for future readers)?