ultra 5 locks up after 2 days uptime

stucky · n00b Joined: 20 Feb 2004 Posts: 16

I successfully installed gentoo sparc (2.4.25 gentoo kernel sources) on software raid1 on my U5.
Everything seems fine but when I let it run for a while I found that after about 2 days the box completely locks up. I cannot ssh anymore and when I serial console to it it doesn't invoke /sbin/ logon it just does nothing at all.
Sending it a BRK signal is the only thing that still works but from here I can only turn the box off or reboot. I have no clue what state the OS is in but its somehow totally non-functional. If I check the logs after I see that it just stopped logging at all at some point and the next line I see is from the reboot I forced.

When I reset it from the OPB it reboots fine but I noticed that the raid is broken everytime after this happens and I have to rebuild it.

Not sure if this has to do with that but I noticed that the older sparc livecd had a big make.conf in it with all kinds of stuff preset whereas the 2004 livecd's make .conf has barely 2 or 3 lines in it and the compiler flags are different too.
The old one used to say -O3 -pipe and the new one says -O2 only.
I changed it back to -O3 -pipe. Should I not have done that cause now I ended up building unstable binaries + kernel ? What happened to the make.conf ?

Alex

patrickbores · Posted: Sun Apr 11, 2004 2:32 pm Post subject:

Might be a hardware problem. What is the CPU speed on the machine? At my previous job, we bought a couple of Ultra 5s with the 360Mhz chips. They were displaying the same kind of behavior you've described.

It turned out that BOTH had bad CPUs on them (and they were only a few serial numbers apart). So we concluded that Sun put out some flawed ones. We ordered new CPUs, and things worked fine after that.

Maybe not the answer you were hoping for, but that's what I know.

HTH,

Patrick
_________________
P.S. - this is what part of the alphabet would look like if Q and R were eliminated.

stucky · n00b Joined: 20 Feb 2004 Posts: 16

indeed it wasn't but thanks for the reply !

My U5 is actually a 270 MHz but that doesn't mean it can't have the same prob.
Not sure if I wanna spend bucks on a new CPU when I'm not even sure that's it cause
I got the U5 for free (from a storage room at SUN..probably bad to begin with but I thought I might be lucky..) and I wanted to keep it free.
Out of curiosity what was the serialnumber on your bad one ?
Do you still know ?

Alex

sterbaya · n00b Joined: 23 Sep 2003 Posts: 63 Location: France

You can run hardware test at OBP with :
setenv auto-boot? false
setenv diag-switch? true
setenv diag-level max

With serial console, hardware test will be display on console.
With graphic console (monitor), you'lll have a blank screen about 5 min. Type : show-post-resulsts for view posts results.

You can upgrade OBP with solaris patch 106121 you can find here :
http://sunsolve6.sun.com/search/document.do?assetkey=1-21-106121-18-1[/url]