[Diy_efi] I think this is working, and admin note

[Jon V]

OK, I think I've got it ironed out and nailed down... one final test....

I have set diy-efi up so that it redirects to diy_efi, but unfortunately
that triggers mailman's moderator-oversight mode. In other words, mail
sent to diy-efi will be held until I get around to freeing it, but then it
will be delivered to diy_efi and distributed to the list. To avoid delays,
just use the diy_efi address directly.

All messages will have "[Diy_efi]" on the subject line... that run of test
messages should be the last you see with "[Diy-efi]".

Sigh.... we're getting there.

It will probably take another day or two before the web pages are fully
updated to point to mailman and not have majordomo instructions. 

There will also be four reboots needed. One I'm going to be doing in about
a half hour (after I've copied all the files off of hektor's drive into a
safe place, I'm going to pull the drive out and set it asside), and the-.


For those that are interested:

An upgrade has been planned for a long time. Unfortunately, it sometimes
takes a failure to shift priorities enough that something actually gets
done.

For example, I actually purchased the hardware for the new server back in
early January..... :) I was going to get everything "just right" and cut
the lists over as transparently as possible. Unfortunately, I misjudged
the amount of spare time I would have available, and the machine has been
sitting around basically "burning in" as I did whatever it is I do to
convince people to pay me.

The old server (hektor), for those who don't know or didn't care, was an
older DIGITAL Alpha. It had a single 9GB SCSI drive and 80MB of RAM. It
was a nice enough machine, though a tad slow by modern standards.
Unfortunately, the 64 bit Alpha processor sometimes made it hard to find
the software we/I wanted to run, causing a lot of frustration.

The new machine was intended to remedy the major problems of Hektor. I
talked about it (lauded it with many ambitious ramblings) with the other
list administrators at the time, and built it up, and even sort of showed
it off, but never actually brought it live.

The hardware is starting to sound a bit dated, but it was very good for
January :)

  * Athlon XP1700+ (1.5Ghz)
  * Soyo DRAGON+ motherboard. 
  * 512MB DDR RAM
  * Two 60GB UDMA-66 drives in a RAID 1 mirror set. 
  * Plextor 12/10/32 CD-RW
  * Advansys SCSI controller

(The video card is the only "junk" grade hardware in the machine...some
old PCI card I had lying around. It is a server, after all...)

Up until yesterday it had been running Debian Linux (same as the old
server) with Ext2 filesystems. It has almost everything configured,
including a mailing list server and most of the web site. However,
yesterday I put an end to that dream by accidentally destroying the OS
installation.

I have switched the server to Gentoo Linux (http://www.gentoo.org/).
Gentoo is a more performance-oriented distribution (or meta-distribution
as they call it) different from most in that it is source-only... all
software is compiled on the target system. I am using it mainly because I
have it on my laptop (where the performance gain was significant and
important) and have been rather impressed with it. Not for the faint of
heart, and not for those with slow network connections or slow machines
(e.g. it took me about 7 hours to install/compile the basics on my laptop,
or an hour and 15 minutes to do the same on the new server -- it took 4
days to do the same over a dial-up Internet connection to another machine
I tried), but in the long haul the performance difference seems to be
worthwhile.

Other than that...

The new server is using ReiserFS, a fast journaling filesystem, for all of
its files. The journaling means that the filesystem doesn't become corrupt
if the system crashes or loses power, so we won't have to wait several
hours for the drive to re-check itself in those cases. ReiserFS is also
supposed to be noticeably faster than the older Ext2 filesystem. That,
coupled with the RAID mirror, means that we are a lot better protected
from drive failures and the like. Not only does the RAID improve
reliability, but it also improves disk read access speed (almost doubles
it, to be specific), and that can't hurt. And, because the system is based
on a common CPU, if another hardware failure occurs we can just change
whatever part failed, rather than rebuilding the whole machine.

The system had been running for something like 70 days continuous before
hektor went down... this isn't very long, but it is usually long enough to
show any flakey hardware. Seems rock solid. 

It is obvious that the mailing list software is now Mailman. If you are
subscribed to very many mailing lists, then you have used Mailman
elsewhere. It isn't the fanciest list manager out, but it is much prettier
than majordomo.

As always: Please direct any questions/comments to me directly and I'll
get them sorted out as quickly as I can.

Thanks,

-- 
Jon


_______________________________________________
Diy_efi mailing list
Diy_efi at www.diy-efi.org
http://diy-efi.org/mailman/listinfo/diy_efi