Boston Linux Meeting Wednesday, July 16, 2008 with abstract: Beowulf clusters

[Jerry Feldman]

When: July 16, 2008 7:00PM (6:30 for Q&A)
Topic: Beowulf clusters
Moderator: Don Becker, CTO of Penguin Computing
Location:  MIT Building E51 Room 315 

Beowulf clusters are scalable performance clusters based on commodity
computers connected with a private system network.  They were named
after the NASA Beowulf Project, an effort to develop software for and
demonstrate the effectiveness of commodity cluster computing.

The challenge of commodity clusters has moved from basic machine
communication and communication library support to effective
administration and monitoring of large and changing numbers of machines.

Based on their Beowulf Project experience, Scyld has developed a
innovative cluster system that dramatically simplifies creating, using
monitoring and maintaining Beowulf clusters.  This talk will describe
the evolution of Beowulf systems, using the Scyld system as an
illustration of how cluster software has advanced from a collection of
individual ad hoc installations to elegant and efficient single system
image clusters that incrementally scale and tolerate failures.

For much more information, and Parking please
refer to http://www.blu.org/cgi-bin/calendar/2008-jul
There is a parking lot adjacent to building E-51 at 2 Amherst St. 

Note: The after-meeting meeting will be at The Cambridge Brewery. 

Also remember that the BLU BarBQue is on Saturday July 19
http://www.blu.org/cgi-bin/calendar/2008-bbq14 I'll post a separate
message next week.


--
Jerry Feldman <[EMAIL PROTECTED]>
Boston Linux and Unix
PGP key id: 537C5846
PGP Key fingerprint: 3D1B 8377 A3C0 A5F2 ECBB  CA3B 4607 4319 537C 5846


signature.asc
Description: PGP signature
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Example of Linux initramfs

[Thomas Charron]

On Fri, Jul 11, 2008 at 1:52 PM, Michael Nolin <[EMAIL PROTECTED]> wrote:
> --- On Fri, 7/11/08, Thomas Charron <[EMAIL PROTECTED]> wrote:
>>   I was talking specifically about an initramfs, which
>> specifically
>> uses cpio.  initramfs.c checks for a magic flag in the
> I would recommend "Understanding the Linux Kernel" and "Linux Kernel 
> Development Second Edition"
> kernel/init
>do_mounts.h do_mounts_md.cKconfig  Makefile   version.o
> do_mounts.cdo_mounts_initrd.c   initramfs.c main.c
> do_mounts_devfs.cdo_mounts_rd.c  version.c
> These files and kernel initialization is explained in these references. Very 
> little if any ARM specific dependencies here.

  The point is

  His issue has nothing to do with ARM really.  His issue is, he's got
the kernel booting to the point of loading the initramfs, at which
point it dies horribly.  My point was that the CPIO archive which
contains the root filesystem that is loaded from initramfs.c requires
a magic header which marks the filesystem contained as being a valid
root initramfs.

-- 
-- Thomas
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Example of ARM based linux board using initramfs and serial console

[Michael Nolin]

--- On Fri, 7/11/08, Thomas Charron <[EMAIL PROTECTED]> wrote:

> From: Thomas Charron <[EMAIL PROTECTED]>
> Subject: Re: Example of ARM based linux board using initramfs and serial 
> console

> > > says it can be a root FS by the kernel.
> >  We're still talking Linux, right?
>  (The kernel will panic if it then cannot
> find an initial
> > program to start, but it will still mount the
> filesystem first.)
> >  You may be thinking of the "rdev" field in
> the kernel boot image
> > file.  It gives major and minor numbers for the
> default root device,
> > which are used if you don't specify a
> "root=" argument on the kernel
> > command line.
> 
>   I was talking specifically about an initramfs, which
> specifically
> uses cpio.  initramfs.c checks for a magic flag in the

I would recommend "Understanding the Linux Kernel" and "Linux Kernel 
Development Second Edition"

kernel/init 

do_mounts.h do_mounts_md.cKconfig  Makefile   version.o
do_mounts.cdo_mounts_initrd.c   initramfs.c main.c   
do_mounts_devfs.cdo_mounts_rd.c  version.c

These files and kernel initialization is explained in these references. Very 
little if any ARM specific dependencies here.

grep -i ARM *
Binary file built-in.o matches
Binary file do_mounts_initrd.o matches
Binary file do_mounts.o matches
Binary file do_mounts_rd.o matches
Binary file initramfs.o matches
Kconfig:default y if ARM || H8300
Kconfig:config OBSOLETE_MODPARM
main.c: printk("early_parm_test: %s\n", rest ?: "(null)");
Binary file main.o matches

My own project had a generous block of DDR memory for the creation of this 
ramfs. 

Michael Nolin
Embedded Solutions Unlimited, LLC





  
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Example of ARM based linux board using initramfs and serial console

[Thomas Charron]

On 7/11/08, Ben Scott <[EMAIL PROTECTED]> wrote:
> On Fri, Jul 11, 2008 at 11:53 AM, Thomas Charron <[EMAIL PROTECTED]> wrote:
> >  There IS something that I can't recall the exact specifics of, but
> > it's along the lines of a 'magic identifier' in the filesystem that
> > says it can be a root FS by the kernel.
>  We're still talking Linux, right?
>  To the best of my knowledge, there is no such thing.  Any filesystem
> can be mounted as a root filesystem.  There's no magic in the
> filesystem.  (The kernel will panic if it then cannot find an initial
> program to start, but it will still mount the filesystem first.)
>  You may be thinking of the "rdev" field in the kernel boot image
> file.  It gives major and minor numbers for the default root device,
> which are used if you don't specify a "root=" argument on the kernel
> command line.

  I was talking specifically about an initramfs, which specifically
uses cpio.  initramfs.c checks for a magic flag in the header which
tells the kernel this is indeed a cpio compressed filesystem,
specifically for mounting as the initial root filesystem by the
kernel.  Sorry I wasn't more specific when I said 'filesystem',
perhaps I should have clarified with 'initramfs filesystem'.

-- 
-- Thomas
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Example of ARM based linux board using initramfs and serial console

[Ben Scott]

On Fri, Jul 11, 2008 at 11:53 AM, Thomas Charron <[EMAIL PROTECTED]> wrote:
>  There IS something that I can't recall the exact specifics of, but
> it's along the lines of a 'magic identifier' in the filesystem that
> says it can be a root FS by the kernel.

  We're still talking Linux, right?

  To the best of my knowledge, there is no such thing.  Any filesystem
can be mounted as a root filesystem.  There's no magic in the
filesystem.  (The kernel will panic if it then cannot find an initial
program to start, but it will still mount the filesystem first.)

  You may be thinking of the "rdev" field in the kernel boot image
file.  It gives major and minor numbers for the default root device,
which are used if you don't specify a "root=" argument on the kernel
command line.

-- Ben
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Example of ARM based linux board using initramfs and serial console

[Thomas Charron]

On 7/11/08, Alexander Wolfson <[EMAIL PROTECTED]> wrote:
> This is what I am doing as well.
> Usually there is a statically linked program in the /bin called busybox which 
> provides big chunk of expected command line functionality.
> All the programms - bash, ash, ls, ping, ... are slinks to busybox (init as 
> well) and busybox checks ARGV[0] for the name that was used to call it and 
> acts accordingly. It still has to be part of some filesystem and this is 
> where the problem is. I am doing something wrong creating the initramfs and 
> linking it with the kernel or something is wrong with my busybox or this is 
> something else that I am not even think of.

  There IS something that I can't recall the exact specifics of, but
it's along the lines of a 'magic identifier' in the filesystem that
says it can be a root FS by the kernel.  Forgive me for not recalling
exactly what it is, and I have no time ATM to STFW for it.  :-D

-- 
-- Thomas
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: How to troubleshoot wide area network performance problem?

[Alex Hewitt]

On Fri, 2008-07-11 at 11:30 -0400, Hewitt_Tech wrote:
> Mark Greene wrote:
> > 
> > 
> > On Thu, Jul 10, 2008 at 7:36 PM, Alex Hewitt <[EMAIL PROTECTED] 
> > > wrote:
> > 
> > I have clients with an interesting network problem. One location in
> > Bedford New Hampshire using a fractionated T1 has routinely been
> > transmitting studies to an office in Nashua New Hampshire. There have
> > been no problems with this for at least 18 months. However recently
> > (about a week ago), the transmissions suddenly became slow, really slow.
> > A transmission that was taking around 10 minutes suddenly jumped to 2-3
> > hours. The customer in Bedford New Hampshire is using One
> > Communications. 
> > 
> > 
> > I'd bet money that One Communications is the culprit, and that they are 
> > doing different routing on their network to you vs. to your Nashua 
> > client's office.  They *may* be doing selective throttling based on 
> > content ala Comcast, but this may also be a non-malicious mistaken 
> > config problem too.
> > 
> > mark
> > 
> > 
> > 
> > 
> > 
> > 
> > ___
> > gnhlug-discuss mailing list
> > gnhlug-discuss@mail.gnhlug.org
> > http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
> 
> I've had suggestions from at least two colleagues that we may be the 
> victims of peer to peer throttling. I'm going back to the Nashua site 
> later today and I'm going to replace a small internal router that used 
> to replace a failed router Monday. I don't believe the internal router 
> has any bearing on the problem because the customer noticed the problem 
> when there was no internal router in place (we bypassed it as a 
> workaround). I'm not sure if there is any kind of tool that can be used 
> to check for throttling. One of my colleagues ran into a Comcast 
> throttling problem while doing an rsync at a different location. He said 
> the rsync ran at full speed for about 30 seconds and then basically 
> dropped to about ten percent performance after that. I need to see if 
> something similar is going on at the Bedford site.
> 
> -Alex
> 
> P.S. I'll probably put in a call to One Communications today to have 
> them check the connection/routing.

Actually the site I'm going to replace the router at is the Bedford
site. I want to make sure I've done everything humanly possible to be
100% sure the problem isn't in equipment that I can control.

-Alex

> 
> ___
> gnhlug-discuss mailing list
> gnhlug-discuss@mail.gnhlug.org
> http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: How to troubleshoot wide area network performance problem?

[Hewitt_Tech]

Mark Greene wrote:
> 
> 
> On Thu, Jul 10, 2008 at 7:36 PM, Alex Hewitt <[EMAIL PROTECTED] 
> > wrote:
> 
> I have clients with an interesting network problem. One location in
> Bedford New Hampshire using a fractionated T1 has routinely been
> transmitting studies to an office in Nashua New Hampshire. There have
> been no problems with this for at least 18 months. However recently
> (about a week ago), the transmissions suddenly became slow, really slow.
> A transmission that was taking around 10 minutes suddenly jumped to 2-3
> hours. The customer in Bedford New Hampshire is using One
> Communications. 
> 
> 
> I'd bet money that One Communications is the culprit, and that they are 
> doing different routing on their network to you vs. to your Nashua 
> client's office.  They *may* be doing selective throttling based on 
> content ala Comcast, but this may also be a non-malicious mistaken 
> config problem too.
> 
> mark
> 
> 
> 
> 
> 
> 
> ___
> gnhlug-discuss mailing list
> gnhlug-discuss@mail.gnhlug.org
> http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

I've had suggestions from at least two colleagues that we may be the 
victims of peer to peer throttling. I'm going back to the Nashua site 
later today and I'm going to replace a small internal router that used 
to replace a failed router Monday. I don't believe the internal router 
has any bearing on the problem because the customer noticed the problem 
when there was no internal router in place (we bypassed it as a 
workaround). I'm not sure if there is any kind of tool that can be used 
to check for throttling. One of my colleagues ran into a Comcast 
throttling problem while doing an rsync at a different location. He said 
the rsync ran at full speed for about 30 seconds and then basically 
dropped to about ten percent performance after that. I need to see if 
something similar is going on at the Bedford site.

-Alex

P.S. I'll probably put in a call to One Communications today to have 
them check the connection/routing.

___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

RE: Example of ARM based linux board using initramfs and serial console

[John Abreau]

Just as a simple sanity check, I'd test it without trying to
overcomplicate things. Forget the symlinks to busybox; just
rename the binary to /bin/sh, or whatever name busybox uses
to decide to be a shell; then modify the test kernel to run
/bin/sh instead of /sbin/init. That will at least prove that
the kernel can successfully boot.

And from that, you can try running by hand each command that
your proposed initramfs is supposed to run automatically via init.
By doing that, you should be able to identify where the init
process was failing.



On Fri, July 11, 2008 7:59 am, Alexander Wolfson said:
> This is what I am doing as well.
> Usually there is a statically linked program in the /bin called busybox
> which provides big chunk of expected command line functionality.
>
> All the programms - bash, ash, ls, ping, ... are slinks to busybox (init
> as well) and busybox checks ARGV[0] for the name that was used to call it
> and acts accordingly. It still has to be part of some filesystem and this
> is where the problem is. I am doing something wrong creating the initramfs
> and linking it with the kernel or something is wrong with my busybox or
> this is something else that I am not even think of.
>
> Alex
>
> -Original Message-
> From: [EMAIL PROTECTED] on behalf of John Abreau
> Sent: Thu 7/10/2008 5:59 PM
> To: gnhlug-discuss@mail.gnhlug.org
> Subject: RE: Example of ARM based linux board using initramfs and serial
> console
>
>
> On Thu, July 10, 2008 4:08 pm, Alexander Wolfson said:
>> This is our own board based on our own chip which among other things has
>> ARM926EJ-S (ARMv5TEJ) core.
>> There is no BSP yet, no Flash or USB drivers - only limited access to
>> the board over JTAG. We can boot the board up to the point when kernel
>> dies because there is no init. This why I need initramfs now.
>>
>
>
> While i don't know the details, I understand that it's common practice
> when porting to a new platform to use bash, or perhaps a simpler shell,
> in place of init, as a first step to achieve an initial boot.
>
>
> --
> John Abreau / Executive Director, Boston Linux & Unix
> IM: [EMAIL PROTECTED] / [EMAIL PROTECTED] / [EMAIL PROTECTED] / [EMAIL 
> PROTECTED]
> Email [EMAIL PROTECTED] / WWW http://www.abreau.net / PGP-Key-ID 0xD5C7B5D9
> PGP-Key-Fingerprint 72 FB 39 4F 3C 3B D6 5B E0 C8 5A 6E F1 2C BE 99
>
>
> --
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.
>
> ___
> gnhlug-discuss mailing list
> gnhlug-discuss@mail.gnhlug.org
> http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
>
>
> --
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.
>


-- 
John Abreau / Executive Director, Boston Linux & Unix
IM: [EMAIL PROTECTED] / [EMAIL PROTECTED] / [EMAIL PROTECTED] / [EMAIL 
PROTECTED]
Email [EMAIL PROTECTED] / WWW http://www.abreau.net / PGP-Key-ID 0xD5C7B5D9
PGP-Key-Fingerprint 72 FB 39 4F 3C 3B D6 5B E0 C8 5A 6E F1 2C BE 99


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: How to troubleshoot wide area network performance problem?

[Mark Greene]

On Thu, Jul 10, 2008 at 7:36 PM, Alex Hewitt <[EMAIL PROTECTED]>
wrote:

> I have clients with an interesting network problem. One location in
> Bedford New Hampshire using a fractionated T1 has routinely been
> transmitting studies to an office in Nashua New Hampshire. There have
> been no problems with this for at least 18 months. However recently
> (about a week ago), the transmissions suddenly became slow, really slow.
> A transmission that was taking around 10 minutes suddenly jumped to 2-3
> hours. The customer in Bedford New Hampshire is using One
> Communications.


I'd bet money that One Communications is the culprit, and that they are
doing different routing on their network to you vs. to your Nashua client's
office.  They *may* be doing selective throttling based on content ala
Comcast, but this may also be a non-malicious mistaken config problem too.

mark
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Quick DNS perfromance measurement trick

[Chip Marshall]

On July 10, 2008, Jeff Kinz sent me the following:
> It appears that "good" resolvers have lots of ports. 
> 
> Anyone who wants to take a whack at explaining what this means is very
> welcome! 

http://www.kb.cert.org/vuls/id/800113

Basically, if you have a single port or small range of ports that you
generate DNS queries from, it becomes easier to poison your cache with
invalid answers. Since that security announcement, there's been a big
push to deploy updated versions of BIND that use a wider souce port
range.

-- 
Chip Marshall <[EMAIL PROTECTED]>
http://weblog.2bithacker.net/PGP key ID 43C4819E
v4sw5PUhw4/5ln5pr5FOPck4ma4u6FLOw5Xm5l5Ui2e4t4/5ARWb7HKOen6a2Xs5IMr2g6CM


signature.asc
Description: Digital signature
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

On the evils of Comcast

[Coleman Kane]

I saw this in the news today, FCC ruled *against* COMCAST. Since this is
like the "secondary topic" of the mailing list, here's the link:

  * http://biz.yahoo.com/ap/080711/internet_regulation.html?.v=5

-- 
Coleman Kane


signature.asc
Description: This is a digitally signed message part
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

RE: Example of ARM based linux board using initramfs and serial console

[Alexander Wolfson]

This is what I am doing as well.
Usually there is a statically linked program in the /bin called busybox which 
provides big chunk of expected command line functionality.

All the programms - bash, ash, ls, ping, ... are slinks to busybox (init as 
well) and busybox checks ARGV[0] for the name that was used to call it and acts 
accordingly. It still has to be part of some filesystem and this is where the 
problem is. I am doing something wrong creating the initramfs and linking it 
with the kernel or something is wrong with my busybox or this is something else 
that I am not even think of.

Alex

-Original Message-
From: [EMAIL PROTECTED] on behalf of John Abreau
Sent: Thu 7/10/2008 5:59 PM
To: gnhlug-discuss@mail.gnhlug.org
Subject: RE: Example of ARM based linux board using initramfs and serial console
 

On Thu, July 10, 2008 4:08 pm, Alexander Wolfson said:
> This is our own board based on our own chip which among other things has
> ARM926EJ-S (ARMv5TEJ) core.
> There is no BSP yet, no Flash or USB drivers - only limited access to
> the board over JTAG. We can boot the board up to the point when kernel
> dies because there is no init. This why I need initramfs now.
>


While i don't know the details, I understand that it's common practice
when porting to a new platform to use bash, or perhaps a simpler shell,
in place of init, as a first step to achieve an initial boot.


-- 
John Abreau / Executive Director, Boston Linux & Unix
IM: [EMAIL PROTECTED] / [EMAIL PROTECTED] / [EMAIL PROTECTED] / [EMAIL 
PROTECTED]
Email [EMAIL PROTECTED] / WWW http://www.abreau.net / PGP-Key-ID 0xD5C7B5D9
PGP-Key-Fingerprint 72 FB 39 4F 3C 3B D6 5B E0 C8 5A 6E F1 2C BE 99


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/


___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Boston Linux Meeting Wednesday, July 16, 2008 Don Becker CTO Penguin Computing

[Jerry Feldman]

When: June 18, 2008 7:00PM (6:30 for Q&A)
Topic: To be announced
Moderator: Don Becker, CTO of Penguin Computing
Location:  MIT Building E51 Room 315 

Don Becker is currently CTO of Penguin Computing, and many of us know
Don for his work both in Linux drivers as well as in Beowulf clusters.
Please check our website in the next few days for more details. I'll be
able to post more information shortly. 

For much more information, and Parking please
refer to http://www.blu.org/cgi-bin/calendar/2008-jul
There is a parking lot adjacent to building E-51 at 2 Amherst St. 

Note: The after-meeting meeting will be at The Cambridge Brewery. 

Also remember that the BLU BarBQue is on Saturday July 19
http://www.blu.org/cgi-bin/calendar/2008-bbq14 I'll post a separate
message next week.



--
Jerry Feldman <[EMAIL PROTECTED]>
Boston Linux and Unix
PGP key id: 537C5846
PGP Key fingerprint: 3D1B 8377 A3C0 A5F2 ECBB  CA3B 4607 4319 537C 5846


signature.asc
Description: PGP signature
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/

Re: Quick DNS perfromance measurement trick

[Brian Chabot]

Michael ODonnell wrote:

>"aaa.bbb.ccc.ddd is POOR: 26 queries in 3.1 seconds from 1 ports with std 
> dev 0.00"
> 
> That aaa.bbb.ccc.ddd address seems to be the (possibly NAT'd) IP
> addr that the target site sees mentioned in the inbound packets;
> I have no idea about the rest of it...

It looks like a responding DNS server to me... whether the authoritative
or (more likely) a cached one.

[EMAIL PROTECTED] wrote:
> Here are my results: 
> z.y.x.w.v.u.t.s.r.q.p.o.n.m.l.k.j.i.h.g.f.e.d.c.b.a.pt.dns-oarc.net.
> "209.244.7.43 is POOR: 38 queries in 1.9 seconds from 2 ports with std dev 
> 0.94"

$ host 209.244.7.43
43.7.244.209.in-addr.arpa domain name pointer keynote2.Phoenix1.Level3.net.
$

My results:
$ dig +short porttest.dns-oarc.net TXT
z.y.x.w.v.u.t.s.r.q.p.o.n.m.l.k.j.i.h.g.f.e.d.c.b.a.pt.dns-oarc.net.
"216.231.41.2 is GOOD: 26 queries in 0.6 seconds from 26 ports with std
dev 18409.11"
$ host 216.231.41.2
2.41.231.216.in-addr.arpa domain name pointer ns-legacy.speakeasy.net.
$ cat /etc/resolv.conf
# Dynamic resolv.conf(5) file for glibc resolver(3) generated by
resolvconf(8)
# DO NOT EDIT THIS FILE BY HAND -- YOUR CHANGES WILL BE OVERWRITTEN
nameserver 216.254.95.2
nameserver 216.231.41.2
search datasquire.net
$


___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/