DNS Not working after TC_Install to USB

[drendar]

Hi There,

I wonder if someone can assist...  I'm new to TC, but know my way around RHEL and Ubuntu, so not a total noob :-)

Running TC in a virtualbox guest with an 8Gb virtual USB drive inserted.

Basically, I boot from CorePlus CD, choose (for example) the fluxbox option.  Go to TC_Install and follow the instructions.  The only changes I make really are adding "kmap=qwerty/uk vga=789" to the bootvars.  I choose to install everything except the Wireless networking bumpf, as I don't need that.

Upon then booting from the USB, APPs will not work, reporting "Error, check network, mirror or writable extension directory"

A bit of digging shows that DNS lookups are failing, and APPs is throwing "wget: bad address ftp.vim.org"

Note that DNS and installing apps when booting from coreplus CD works fine.

Also networking is up and running.  I can ping my server, SSH, and I can bring up web pages in chromium using the IP address instead of FQDN


I'm guessing there's something not working in the DHCPClient area, but I've no idea where to start looking, and I've been at this for hours now, bed is calling.

Could anyone please provide me with a little nudge in the right direction?

Many Thanks,

(P.S  Holy crap these captchas are hard to read!)

[gerald_clark]

Use bridge mode, not NAT.  Then DHCP and DNS from your router will work.

[curaga]

You don't say if you're still running in vbox, or if the usb install is on real hw but the cd was in vbox?

[drendar]

Use bridge mode, not NAT.  Then DHCP and DNS from your router will work.

Unfortunately that isn't possible due to restrictions on my work network.  However I doubt this is the problem, as it all works perfectly when booting from the TCPlus CD, it only fails after installing to the virtual USB.

You don't say if you're still running in vbox, or if the usb install is on real hw but the cd was in vbox?

Everything is virtual, including the USB stick (VDI attached as removable solid state device to USB controller)


--

Update - The problem appears to be related to there being multiple network interfaces...  eth0 is NAT and eth1 is Host Only.

It seems that after install, the DHCP derived DNS setting from eth1 (Which is blank) is overriding the setting received from DHCP on eth0.

If I disable the second interface then it works as expected.

I'm thinking this is probably a bug?

[gerald_clark]

Why would you configure 2 virtual network interfaces in a VM?

[curaga]

Multiple network interfaces are always a special configuration and need your attention. Use the bootcode nodhcp and configure your networks in bootlocal.sh.

[drendar]

Why would you configure 2 virtual network interfaces in a VM?

Because the system I'm simulating is dual homed.  One NIC facing the corporate network and the other an isolated, unrouted VLAN.

There are (or rather, will be) other VMs that exist only on the Host-Only Network, and the hosts "host-only" interface is usually shutdown, enabled only for troubleshooting (Which is why it's host-only and not 'Internal network'

Multiple network interfaces are always a special configuration and need your attention. Use the bootcode nodhcp and configure your networks in bootlocal.sh.

Forgive my ignorance, but is it possible to still leave DHCP enabled on one interface?


All this though doesn't explain why the system works *perfectly* when booting from CorePlus CD, and only fails once installed.

[curaga]

Yes, you just specify DHCP for this interface, static for that. The default is DHCP for all found interfaces, under the assumption only one is usually operable.
Look in dhcp.sh for an example, or use the control panel network applet.

The result of which interface is first is random, and the CD loads more extensions, which affects other processes at boot.

[drendar]

Righto, Thanks!