FW: xCAT Cluster Management SW?
egan at sense.net
Wed Jan 15 14:58:48 PST 2003
> Is this still going to happen? I have spent the last 3 days trying to
> get the latest release of xcat working. Here are the problems we ran
> into ( BTW -- we have installed 2 eariler versions of xcat
> with minimal
> problems. ) It would seem that the latest xcat.org debauchle
> has really
> screwed the pooch.
Since I had to break out the OSS, IBM, and xCAT parts in to different
tarballs it has lead to some confusion and undocumented missing
commands. xCAT 1.2.0 fixes that.
> 1) there is no updated Howto to be found. Everythink included in the
> tarballs seems to point to xcat.org. Some updated documentation can
> really help.
xCAT 1.2.0 has many changes and a new mini HOWTO with links to the other
detail updated documentation. xCAT 1.2.0 HOWTO will be written by Matt
after 1.2.0 releases.
> 2) There are several files/programs missing, with no
> instructions on how
> to find them or build them. Examples: fping, mpcli, etc.
xCAT is 3 tarballs now. You need all three and need to extract all 3 in
the same directory. IBM will not host the OSS parts. Get parts 1 and 2
from http://www.alphaworks.ibm.com/tech/xCAT/ and the 3 from
> 3) VERY lacking documentation on any of the hardware shipped with the
> IBM clusters. Can you _please_ include information on the ESP/ELP
> terminal servers? We had to spend 2 days on the phone with Equinox to
> just get the IP address reset on an ESP. It is not that tough
> to do, but
> again some documentation would be nice.
That is fully documented in the xCAT 1.1.0 redbook and there is a new
terminal server HOWTO in the xCAT 1.2.0 release.
> 4) The IBM RSA/ASMA adapter sucks. Sorry guys, but a piece of hardware
> that is supposed to provide remote administration should
> _not_ fail 25%
> of the time. It is really tiresome to have to dump power on and RSA
> adapter just to get it to work.
Many of the problems with the RSA firmware have been resolved, please
upgrade the firmware. If you have a specific list of issues please
send them directly to me. The most common problems are related to
improper cabling. That is fully documented on
1.1.7 is stable and has been in production for 3 months. Upgrading to
1.2.0 is strongly recommmended and if you can wait for 1.2.0, please do.
xcat.org and the mailing list with archives should be back up this week.
More information about the Beowulf