Mailing lists using usenet newsgroup

[Rick Moen]

On Wed, Jan 7, 2015 at 8:44 AM, jim <jim at well.com> wrote:

>     I'm not trying to defend or excuse myself; rather I
> hope my confession helps someone reflect and possibly
> improve.

I didn't fault you for anything, Jim.  I just wanted to remind
everyone about fundamentals.  Heavens knows I've cut corners myself
and made errors, but I've also tried to never make the same exact
error twice. I even once lost a couple of weeks of updates to my Web
server, received mailing list traffic,received e-mail, sent e-mail,
and updates to my and others' files in home directories, because my
very casual backup scheme hadn't been run in a while.  But at least I
did have backups, however intermittent.

Another one of my rules of thumb is that the half-assed, 80% solution
you actually do beats the perfect, polished solution you never quite
got around to, every time.  For example, my backups rely mostly on
periodic backups of critical file trees to an external, detachable
hard drive.  That's pretty half-assed -- but it's good enough, and I
actually do it.

I put heavy reliance[1], for some years, on all of my server's
filesystems that matter on a RAID1 mirrored pair of hard drives.  When
one of the two drives failed, and I lacked time and energy to fix that
problem right away, I rushed out to do the next best thing:  I bought
a detachable external hard drive, and did much more frequent backups
than before.  With my laptop, I became increasingly anxious about the
risk of its SSD failing, with my having in effect _no_ backups of its
files, so I finally found an external drive and now both do regular
backups to it and periodically test the integrity of same.

> I believe that you several times wrote about backing
> up the MailMan data and meta-data, yet I was possibly
> * too unfamiliar with some aspects of Unix and *nix systems
> * too distracted or stressed or lazy to explore
> * needed more involvement with others sharing the same
> sense of immediacy

Well, regard this as an opportunity.

Here's something to ponder:
https://web.archive.org/web/20130331052337/http://linuxmafia.com/faq/Admin/linuxmafia.com-backup.html

That's a scheme that I've been using, and refining, for years.
Recently, I was bowled over to discover, nonetheless, an error in it.
 The steps I had been taking to back up /var/spool/mail, which
directory houses the MTA's spool of received mail for various users,
had been ineffective.  Why?  Because Debian in its dubious wisdom had
at some point decided that the MTA mail spool should be in /var/mail,
and left /var/spool/mail as a symlink pointing there.  My rsync
command to back up /var/spool/mail was copying null contents, and I
hadn't noticed.  Fortunately, I corrected this error and added
/var/mail/* to my backup set.

The major value of that Web page for me is its documentation of which
subtrees on my server system it makes sense to back up.  This was the
fruit of careful study -- and yet, even with that caution, it
accidentally omitted one significant subtree because Debian moved
those files and, although I knew they did that, I hadn't made the
connection about my backup scheme needing revision.

> Assuming so, seems to me that sf-lug and dv-lug might
> benefit with a shared backup regimen.

Might I suggest process thinking?  Set up Mailman, play with it, get
to know it well.  Get to know how to backup and restore.

You can do that starting right now.

Talking about 'ideas' is just blather.  You want to get something
done, good:  Go get your hands dirty.

> My idea is to find out what MailMan facilities allow full
> backup as well as incremental backup:

My suggestion:  Stop talking about 'ideas' and how to do things with
all the bells and whistles, e.g., full as well as incremental.  (At
your leisure, by the way, look up differential backups and learn why
they make a great deal more sense than incrementals.)  Get your hands
dirty, start simple, go for the low-hanging fruit.  You don't need to
barrage people with questions or bounce 'ideas' off them to get this
process underway.  What you need to do is work with technology and
apply process thinking to it.

> * somebody (hopefully not me, but I'm willing) catches the
>    appropriate MailMan data to their laptop system, verifies
>    the data restore, then makes sure the data is on some
>    redundant storage device (my vote is USB stick or SSD
>    or some such).

USB stick?  Really?  Honestly?

Damn, Jim.

Tell you what:  I'm not going to keep acting like some sort of guru
for you and tell you why that's not a very bright idea.  You need to
stop the armchair theorising, do some work with Linux, and learn from
it.

Stop telling everyone what you think someone-nobody-in-particular
should aim to achieve as an end-goal (classic non-process thinking).
Instead, go do things.  Because this other stuff is not the way to get
anything done.