cryptofreak.org cryptofreak home projects
contact about
Contact:


projects
News Agenda
Antera Antera
News Commentator
News fcreate
Linux Porting Linux Porting
mod-chal mod-chal
Quake III Quake III
News Zope
Contact: webmaster

From: Perry and Lorae Merritt (plmerritt, hypermall dot net)
Date: 2001.10.08 - 21.57 MDT


X-Mailer: Microsoft Outlook Express 5.50.4522.1200

So your HSM scenarios center around the more typical usage, where the system
is getting full and we need to clean it up. I was thinking of a third, more
lame, scenario, where a file would be removed immediately upon close. I
think, and am probably wrong, that we shouldn't unload the data until the
HSM bit is set. That way if we go down before the operation is complete, we
can recover by checking the bit. If the bit is set and the length is not
zero, we can zero the length. If its the other way around, we won't know if
its a legitimate zero length file or one that sorta got HSM'd.

On the backup sequence diagram:

Perhaps its a moot point, since you are changing your mind based on your
Amanda reading, but...

Query the LVM??? From my understanding of an LVM, it knows about blocks, not
files. Do you mean looking at the snapped volume? If so, it still gets
mounted as a file-system, just under a different mount point. After it's
mounted, you can reference the files in the snapped volume through the file
system. If thats the case, I guess you could read the EA from the snapped
image, but you'd want to set them in the unsnapped image. Which brings up an
issue of marking a file as backed up when in fact it may have been changed
while the static image was being backed up. Hmmm.

More than one batch? So what you're saying is the software making a list
might deem it beneficial to start the backup before the entire list is
created?

I'd like to know more about what you learned from your Amanda reading. I
take it you think that Amanda will more than suffice for wnat we need done
and may infact take care of some of the things we thought we might have to
write. Is that accurate?

Man, I wish we could do this sort of stuff all day long. I spent the morning
changing IP addresses and then managed to upgrade three NT clients to
Win2000. I also got to meet our new COO. What a thrill. He's going to change
the world. Yawn.

I talked to Dan this evening for a few minutes. He's making progress on the
presentation to give to the two guys we met with Friday. They're bugging him
daily for it. I just call him to make sure they've been bugging him. ;-)

P



----- Original Message -----
From: "Don jessup" <djessup72, yahoo dot com>
To: <antera, cryptofreak dot org>
Sent: Monday, October 08, 2001 9:00 PM
Subject: Re: AMANDA


>
>   I'm thinking something like following for the HSM:
>
>      1. HSM check to see if the file is candidate for
>         HSMd
>      2. If it is,
>      3.        open file remove data.
>      4.        close file.
>      5.        Set EA to say it's HSMd.
>
>                    or
>      1. HSM check to see if the file is a candidate
> for
>          HSMd
>      2. If it is,
>      3.             get file Atrributes.
>      4.             remove file.
>      5.             touch file and set EA to say it's
>                     HSMd.
>
>
>    Below are comments about questions you asked. I
>   think we should stay away from writting scenarios
> for now at least ones as detailed as I wrote.  For
> example, After reading the AMANDA chapter I would
> change the entire scenario which means I wasted a
>  lot of time.  We should stick to the other documents
>   in the list.
>
>
>
>
> Reading through sequence diagram, I have some comments
> and questions.
>
> I'm impressed, who knew ASCII could look so good. I've
> got to believe it was a pain though.
>     No more of a pain then using power point.
>
> Item 4, you say the ABM creates a list of files. Is it
> getting them from the cron or are is it going to
> querry the portal/ADML?
>     I'm thinking it will query the LVM.
>
> Would it be safer to move steps 5, 6 and 7 to the
> bottom? That way if something bad happened during the
> backup, we could more acurately reflect it in the
> saved metadata.
>     Sounds great.
>
> In step 8 are you referring to the metadata file being
> backed up?
>    Yes
>
> It seems that sending a batch of files in step 11 to
> be backed up would preclude the looping of steps 9
> though 15
>
>         I was thinking that there might be more than
> one batch.
>
> Is step 14 through the portal or through the VFS call?
>          fs
>
> __________________________________________________
> Do You Yahoo!?
> NEW from Yahoo! GeoCities - quick and easy web site hosting, just
$8.95/month.
> http://geocities.yahoo.com/ps/info1
> --
> This is the antera mailing list.  To unsubscribe, email
> majordomo, cryptofreak dot org with message body `unsubscribe antera'.
> Or, for more information, visit http://www.cryptofreak.org/.
>

--
This is the antera mailing list.  To unsubscribe, email
majordomo, cryptofreak dot org with message body `unsubscribe antera'.
Or, for more information, visit http://www.cryptofreak.org/.



This archive was generated by hypermail 2b30 : 2001.10.09 - 03.00 MDT