Umm, bear with me for not posting a host we rented our dedicated box from. Yet.
We have a Virtuozzo-based VPS with them also, and support is correct.
However, little less than a month ago we rented a dedicated box from them. They ordinarily support RedHat 9, CentOS 3.3 etc. mostly pre-imaged and ready to use.
When we asked if they supported CentOS 4, they said yes no problem.
So we rented the box. Then wanted to use the new 2.6.11 kernel patched for Xen instead of 2.6.9.
We compiled the kernel, forgot to configure it with RAM disk support and forgot to make an initrd img and config Grub to use it. Naturally, reboot failed.
Ticketed support and in some 5 days of messaging with them we couldn't get a straight information on the exact boot problem. All was muffled and unclear.
Then we gave up on Xen because they told us it had problems installing and we just couldn't get someone there to tell us what messages the system outputted on screen while rebooting. We were ignored politely, but in spirit of us making an unusual, hard to implement demand.
It was really strange for us because we were porting the CentOS 4.0/2.6.10-xen kernel on local testing box without a problem.
Heh, then I asked them (cos they insisted on themselves compiling kernels) to update our kernel to new 188.8.131.52. That was six days ago.
After two days, I submitted a ticket asking what was wrong, why no kernel. The answer was they didn't know what the problem was and they compiled it twice without success and they were investigating. So I asked what was the exact problem, and got a muffled answer on 'initial console not able to start'.
So I connected this problem and the previous one and advised them that 2.6.11 kernel boots up from RAM disk and has to be configured with RAM disk support, and that it was the problem the first time also.
The answer was 'Yes, we also figured. We will port it on our testing box and if it works, we will put it on your box.'
My answer was 'Our box isn't in production yet, so you can do whatever with it no need for testing. Make an initrd properly and reboot at will', although I was thinking 'Oh, for Christ's sake do this and reboot the freakin box already!'.
Since then 4 days passed, no reply, ticket is on hold... since these people are dealing Virtuozzo VPS-es my belief is we are being held down because of Xen. Like I don't have the right to deploy what I want on the box I am cashing $$$ for.
So now we are going to do it ourselves and reboot, not wait for them.
My question is: at version 184.108.40.206 already, is asking for a working compile of 2.6.1x kernel a special demand or is 2.6 kernel already deployed widely enough for it to be a normal demand?
Because THIS IS OUTRAGEOUS! I mean waiting for 5-6 days two times in a row for someone who claims to support 2.6 kernels to learn about initrd is way off! Two times! First time they didn't ever bother to look it up! Not to mention the muffled answers, ignorance, ignoring the polite demand to make a LOCAL call when the machine is booting to read us the boot screen issues. If we had an info on 'initial console not being able to...' and ext3 and undetectable volumes, we could tell them about initrd almost 10 days ago and run Xen for some time already.
Would this be considered as a valid reason for refunding? I mean we are building our box, we are nearly done, but once we reboot to Xen we need to replan and reconfigure because the 'game' changes then.