[9fans] Odd dirwstat behavior

[9fans] Odd dirwstat behavior

[Steven Stallion]

[-- Attachment #1: Type: text/plain, Size: 782 bytes --]

9fans,

Somewhat recently I've been doing some work to preserve older fs
hierarchies. When running replica/pull I'm seeing something quite strange.
Both the log and database are reporting the original (and correct) mtime
for directories, however for some reason replica/applylog (or the
filesystem) is setting the mtime to time(0). After digging through the
applylog source, I found a couple of conditions that could cause this,
however my modifications have been for naught. This is annoying as hell
given the fact I'm going through this rigamarole to preserve permissions
and dates.

FWIW, replica/applylog is running against a fossil filesystem served out of
devmnt with the following configuration:

fsys main config
fsys main open -APVW

Any ideas?

Steve

[-- Attachment #2: Type: text/html, Size: 925 bytes --]

Re: [9fans] Odd dirwstat behavior

[Steven Stallion]

[-- Attachment #1: Type: text/plain, Size: 1038 bytes --]

On Sun, Feb 22, 2015 at 2:53 PM, Steven Stallion <sstallion@gmail.com>
wrote:

> 9fans,
>
> Somewhat recently I've been doing some work to preserve older fs
> hierarchies. When running replica/pull I'm seeing something quite strange.
> Both the log and database are reporting the original (and correct) mtime
> for directories, however for some reason replica/applylog (or the
> filesystem) is setting the mtime to time(0). After digging through the
> applylog source, I found a couple of conditions that could cause this,
> however my modifications have been for naught. This is annoying as hell
> given the fact I'm going through this rigamarole to preserve permissions
> and dates.
>
> FWIW, replica/applylog is running against a fossil filesystem served out
> of devmnt with the following configuration:
>
> fsys main config
> fsys main open -APVW
>
> Any ideas?
>

Ugh. I see the problem - a second pass is not made by applylog to update
the mtime after files have been applied. Patch forthcoming.

Steve

[-- Attachment #2: Type: text/html, Size: 1450 bytes --]

Re: [9fans] Odd dirwstat behavior

[erik quanstrom]

> Ugh. I see the problem - a second pass is not made by applylog to update
> the mtime after files have been applied. Patch forthcoming.
>

steve, there are a bunch of changes in the replica utilities to help porting history
from one file server to another in 9atom.  cphist is intended for that purpose
(see replica(8) on 9atom).  the paper on the project is here: http://www.quanstro.net/plan9/history.pdf

- erik