Re: 5.2-RELEASE TODO

From: Tom <tom_at_sdf.com>
Date: Thu, 4 Dec 2003 00:04:10 -0800 (PST)
On Wed, 3 Dec 2003, Eric Anderson wrote:

...
> >  Well, there are so many kinds of PERC cards.  Some are just Mylex cards.
> >Others are MegaRAID.  I think they use some Adaptec now.
> >
> My troubles were with the MegaRAID variant, I'm not sure about the others..
>
> >  Either way, I tried booting the install CD with all cards disconnected,
> >just to see if I could get the installer up to the main menu.  No go.  The
> >display switches off as soon as sysinstall starts probing devices.  It
> >appears to panic, but the display is dead.  Once I was able to use
> >scrolllock at the critical moment just before the kernel starts
> >sysinstall, and prevent the display from switch off, but when I release
> >the scroll lock, I was at the DDB prompt.  Not good.
> >
> >
> First - I hope you mean the cards are physically OUT of the machine - it
> seemed to me that if they had a logical disk configuration on the card,
> it would hang.  Although, it sounds like your problem is different.
> Actually, it kind of sounds like it thinks you are doing the "-p" boot
> thing to check for keyboard and roll over to the serial port.  Other
> than that, I'm out of ideas (doesn't take long for that to happen
> though!).. :(

  Yes, the cards are physically out of the machine.

  "-p" seems unlikely, since the display is literally shutoff.  The
monitor goes into a power-save mode, so it appears that as soon as
sysinstall touches whatever devices it touches, the onboard video stops
generating output.  I've also tried a serial console, which seems to fail
the same way.  There is simply no output to the serial console after
sysinstall starts, and the machine appears to be hung.

  So, this was with 5.0 and 5.1.  I will have to try 5.2 and see if it has
the same problem.  No one has contacted me about my PR, or even given me
any suggestions about more debugging info that I can gather, so unless
this gets fixed by chance, I suspect that 5.x is just going to orphan a
significant number of machines.  Which is a shame, since 5.x would be
clearly superior on a quad Xeon machine than 4.9-RELEASE.

> Eric


Tom
Received on Wed Dec 03 2003 - 22:56:20 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:32 UTC