[OpenAFS] [Fwd: OpenAFS-info digest, Vol 1 #2127 - 12 msgs]

[Ray Pasetes]

Hi,

I understand the case of replicated volumes, but why would the update be
< the create time for non-replicated volumes?


bash-2.03# /usr/afs/bin/vos exam XXXXX
XXXXX                    1685424002 RW     845553 K  On-line
     FFFFFFF /vicepd
     RWrite 1685424002 ROnly          0 Backup 1685424004
     MaxQuota    3000000 K
     Creation    Sun Dec 19 01:50:29 1999
     Copy        Tue Jun 17 14:39:36 2003
     Backup      Tue Dec 14 11:10:29 2004
     Last Update Fri Jan  2 13:22:15 1998
     3019 accesses in the past day (i.e., vnode references)

     RWrite: 1685424002    Backup: 1685424004
     number of sites -> 1
        server FFFFFF partition /vicepd RW Site


Would this be the case if we moved a volume too since technically
it creates a new one on a different server and deletes the old one?


Thanks.

-------- Original Message --------

--__--__--

Message: 8
Date: Tue, 21 Dec 2004 17:36:20 -0500
From: Jeffrey Hutzelman <jhutz@cmu.edu>
To: "Kristen J. Webb" <kwebb@teradactyl.com>,
	openafs-info@openafs.org
Subject: Re: [OpenAFS] Differences in vos e output



On Tuesday, December 21, 2004 10:33:03 -0700 "Kristen J. Webb"
<kwebb@teradactyl.com> wrote:

> It appears that there may be more issues related
> to the display of the Last Update field
>
> 1.3 output:
>
>     Creation    Sun Dec 19 06:50:29 1999
>     Copy        Fri Oct 31 12:09:16 2003
>     Backup      Wed Dec 15 15:11:18 2004
>     Last Update Wed Jul 14 06:13:18 1999
>
> 1.2 output:
>
>     Creation    Sun Dec 19 06:50:29 1999
>     Last Update Sun Dec 19 06:50:29 1999
>
> In the 1.3 output the last update is
> older than the creation date in this example.
>
> It seems that it should be the same as
> the Creation date or Never in this instance.

If you were parsing the output of "vos examine" to see when a volume was
modified in 1.2, then you are going to be unhappy some of the time.

You see, volumes have something like 5 or 6 different timestamps.  If
you're doing an incremental dump, then the one you want is the one printed
by 1.3 as the "last update" stamp.  In certain circumstances this can be
older than the creation stamp, because of what constitutes "creation" of a
volume.  For example, on a cloned volume this is the date the clone was
created, which is always more recent than the last time the source volume
was modified.

Unfortunately, in OpenAFS 1.2.x and earlier, vos tries to be "friendly".
The number it prints next to "Last Update" is actually the more recent of
the creationDate or the updateDate.  Starting in 1.3, it will print the
actual updateDate, even if it is older than the creationDate.


-- Jeffrey T. Hutzelman (N3NHS) <jhutz+@cmu.edu>
    Sr. Research Systems Programmer
    School of Computer Science - Research Computing Facility
    Carnegie Mellon University - Pittsburgh, PA


-- 
==============================================
Ray Pasetes             Email: rayp@fnal.gov
CD/CSS/CSI              Phone: 630-840-5250
Fermilab, Batavia, IL   Fax  : 630-840-6345
==============================================