[Coco] OS-9 Incremental Backups and how they might be done.
Gene Heskett
gheskett at wdtv.com
Sun Apr 5 16:01:02 EDT 2015
On Sunday 05 April 2015 14:48:29 Stephen H. Fischer wrote:
> This whole discussion about incremental backups IMHO is at the wrong
> level.
>
> It appears that Gene wants to set a bit in a file when the file has
> been backed up and need not be backed up again.
>
> Yes that is the way that MSDOS, Windows and perhaps Apple and Linux do
> it but it is unclear if we can free up the single bit.
>
> -----------------------------------------------
> There have been several backup programs offered for OS-9 over the
> years.
>
> They need to be looked at, perhaps the task has already been done or
> we can build on reducing the effort.
>
> Spending our time looking for a solution that has already been done
> has more worth than looking for a bit that may never be found.
>
> Start looking here ftp://www.rtsi.com/OS9/OS9_6X09/
And the only backup which used that bit was Back_AR, also from my commit
I believe. Theres a possibility its the original, and I fixed it but
never committed to RTSI.
But I didn't write the original, just debugged it and made it work, until
the big blowup when Boisy got bit. I had already been bit at that point,
and made some noise before Boisy got bit that EVERYONE should up the
SAS= in their rbf descriptors to avoid that.
But as usual, my voice was ignored. By Boisy it turned out. Sigh.
The only other "cross media format" backup on that site is BRU_1_2, which
does not honor any means of an incremental mode. I should know, that
build came from my machine, a long long time ago. 2 decades maybe. No,
almost 23 years. Unforch, its data handling in a byte by byte format
makes it set stakes and call a surveyor slow when backing up, and a
recovery is about 8x slower yet. I did one, to 720k floppies, took 3
days , and the test recovery was nearly 3 weeks.
I always intended to make it do it a sector at a time which should have
been at least 20x faster, but like you, lost my round tuit. The source
is there, so anyone that wants to try, make like a frog and jump right
in. And while you are doing that, find us an unused bit in the
fd.sector that we can use without any fireworks like the Back_AR bit did
to at least 2 of us. FWIW, the exact same speedup comment could be
applied to rzsz, but its a utility that can overrun the coco's ability
to process data when its receiving is about done, we have far better
ways to move data now with dw.
From the directory where I keep that, an image of the Maxtor 7120s I ran
for 15+ years until stiction made me start beating on a corner of it to
get it started:
0 1992/03/20 02:35 ----r-wr A6DC 46B2 bru1.0_bin.ar
0 1992/03/20 02:35 ----r-wr A6E0 7BCC bru1.0_src.ar
0 1992/10/04 01:45 ----r-wr A6E4 503A bru1.1_bin.ar
0 1992/10/04 01:34 ----r-wr A6E8 85CF bru1.1_src.ar
0 1992/12/13 23:38 ----r-wr A6EC 6E4B bru1.2_bin.ar
0 1992/12/13 23:38 ----r-wr A6F0 9E80 bru1.2_src.ar
Back_AR.ar is almost as old:
0 1993/09/22 20:26 ----r-wr 7800 4F3F BackAr_3.ar
RTSI's occassional recreation of that repo from backups does us zero
favors in determining whats new, and whats old enough to have voted
several times. I should NOT have to resort to my own storage media to
date this stuff. But thats life...
[...]
Cheers, Gene Heskett
--
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>
More information about the Coco
mailing list