Ok , so its 2AM and I just got server #1 done for a customer and have been fightin' with it till now , and here's the wierdness.

The system runs great, raid is all good (2x 80gb SATA, mirrored), everythings all good EXCEPT if I shut it down and try and boot it back inside of 30sec, it hangs..........now if i wait for 1min, it boots up fine , every time , the only time it hangs is on restart or if I shut it down and hit power inside of 30sec.

The only thing I can think is the disks are spinning and they need to come to a stop before starting again, I know its not heat , I've had fans blowing dead on it with the case open, It doesnt do it with just one drive attached, it only does it in raid, so my theory is the disks arent able to sync up while spinning / winding down............

Can anyone think of something else ????????????

Spec's

Fedora Core 5 ( Oh and no driver issues whatsoever, loaded pretty as you please )

P4 531 3.0GHz / HT
4gb Kingston
2x SATA WD 80gb 8mb buffer (mirrored raid )
Tyan S5120 Tomcat i915
Dynatron HS
1U Chassis

EDIT: Ok i got three up and running , and all three do the same exact thing ????????????????
I got no clue as to whats causing it other than either the raid controller or the drives themselves ........

Something funky is going on with the raid controller or the disks themselves, doesn't sound like a problem with Fedora.

BTW, why would you run Fedora on a production machine (ie. A Server) ? Redhat has Fedora for the purpose of beta testing basically, so they can get their server product (Redhat Enterprise Linux) stable. Fedora isn't anywhere stable enough for production machines, if you want a free version of Redhat Enterprise Linux, try out CentOS. CentOS is basically a mirrored image of RHEL, but with no licensing fees.

Quote: Originally Posted by Phosphate Something funky is going on with the raid controller or the disks themselves, doesn't sound like a problem with Fedora. BTW, why would you run Fedora on a production machine (ie. A Server) ? Redhat has Fedora for the purpose of beta testing basically, so they can get their server product (Redhat Enterprise Linux) stable. Fedora isn't anywhere stable enough for production machines, if you want a free version of Redhat Enterprise Linux, try out CentOS. CentOS is basically a mirrored image of RHEL, but with no licensing fees.

Well if it was me I wouldnt run this OS on a production machine , but seein' as the customer is ALWAYS RIGHT , and thats what they ordered , thats what they get.

Oh and the RAID controller works fine with 2003, and it wont even work with Core 4, so thats why i'm leaning towards OS , cuz the problem doesnt occure with 2003.

At what point during booting does it hang? P.O.S.T. or some point during loading Fedora?

If it's during loading Fedora are you using the stock (bloated and horrible) kernel?

Quote: Originally Posted by Phosphate Something funky is going on with the raid controller or the disks themselves, doesn't sound like a problem with Fedora. BTW, why would you run Fedora on a production machine (ie. A Server) ? Redhat has Fedora for the purpose of beta testing basically, so they can get their server product (Redhat Enterprise Linux) stable. Fedora isn't anywhere stable enough for production machines, if you want a free version of Redhat Enterprise Linux, try out CentOS. CentOS is basically a mirrored image of RHEL, but with no licensing fees.

Just noticed something. Dude, your sig is seriously not cool by the way. Forking attack in a sig? Grow the hell up.

Quote: Originally Posted by thedemonxanth At what point during booting does it hang? P.O.S.T. or some point during loading Fedora? If it's during loading Fedora are you using the stock (bloated and horrible) kernel?

Yup good point, I've never bothered with Fedora from all the bad things I've heard about it, but does it start in verbose mode or does it use a kernel splash? At what point does it freeze? Not sure how good you are with Linux, but try checking your old dmesg logs if you have any. IIRC the normal dmesg gets reset each time on boot so a command like "grep|dmesg raid" isn't going to do much...

Quote: Originally Posted by thedemonxanth Just noticed something. Dude, your sig is seriously not cool by the way. Forking attack in a sig? Grow the hell up.

I must concur...Get rid of that nonsense please, Phosphate. NOT cool.

#1-

Quote: At what point during booting does it hang? P.O.S.T. or some point during loading Fedora?

It hangs during post, goes through bios then starts posting the FC5 stuff and just stops.

I'm a Linux NooB to the tenth power so , I cant really tell ya more sorry, and the machines are gone now, but I plan on building one in the next few weeks for some R&D due to the buyer being a BIG customer , and wanting to see if the problem is something I can fix or at least figure out 100%.

#2- Phosphate, thanks for the input, but if your sig is in any way malicious , I must agree and ask that you remove it.

#3- thedemonxanth, Thank you also for your input

i have a feeling that it is an issue with the RAID and the HDD's it dosent sounds like any thing else ... but it is odd... never heard or seen that before

I'm willing to bet it is either kernel bloat or a raid controller that is only partially supported by linux.

The Tyan website lists the onboard controller as a Promise PDC20579 SATA RAID controller. I did a Google.com/linux search (google.com/linux is your friend!!) and the only thing I found about this controller was people having DMA issues with it using kernel 2.6.10, which I'm pretty sure FC5 doesn't use. Since the computer is gone now, there's no way to test it for sure, but it could very well be related to the kernel setting the DMA mode either incorrectly, or just causing problems (would point to a buggy driver). Just as a test, on the next machines, I'd check it by using the drives on the SATA controller in non-RAID mode, and see if it gives you the same problem. If they work, you could use software raid, but that'd be a last resort, I'd guess.

Keep us updated, I'd like to see what happens on the next one you build. It's possible it may be fixed in the kernel by then. Basically, eliminate possibilities until you find out what's causing it.

The Tyan board has SATA ports for raid and for non raid ....... and when hooked into the non-raid ports the system ran just fine , this glitch only occurs when using the raid ports and onboard raid controller.

I'll definatly keep ya posted on this , as I'm pretty sure the customer is getting ready to order 8 more in the next month.

check to see if hyper threading is enabled in the bios, i had to shut this off in order for me to boot into ubuntu and fedora

Quote: Originally Posted by c0w_tipp3r check to see if hyper threading is enabled in the bios, i had to shut this off in order for me to boot into ubuntu and fedora

Good idea , when i build the next batch , i'll give it a shot with HT disabled and see what happens ( it was enabled on all the rigs)

HyperThreading support can be compiled into the kernel as well. :-)