[mirror-admin] rsync and updates-testing -> updates-released

[Axel Thimm]

On Wed, Jul 22, 2009 at 07:44:58PM -0300, Carlos Carvalho wrote:
> Jesse Keating (jkeating at redhat.com) wrote on 21 July 2009 20:38:
>  >On Tue, 2009-07-21 at 23:03 -0300, Carlos Carvalho wrote:
>  >> Right now they cannot be absolute for the same reason. I meant
>  >> "absolute" starting from the root of the mirror tree, of course.
>  >> 
>  >
>  >I got push back even on that.  Doing it in repodata is a tad bit
>  >dangerous, so I'd rather do it with file hardlinks.
>  >
>  >Why is there an aversion to hardlinks?
> 
> It's not an aversion but it's an inferior solution for the following
> reasons:

Inferior to what?

> - more error prone, as has happened here more than once

The errors was sometimes not having the hardlinks between different
distributions, e.g. a fresh cut of Fedora N out of rawhide w/o proper
hardlinks. It wasn't a worse situation than if hardlinks had never
been used.

> - less clear repository structure (admitedly subjective)
> - rsync needs more memory to handle hardlinks
> - increases the number of file names, thus needing more disk accesses;
>   if the requests are close enough in time the (same) inode will
>   already be in ram and the cost will be negligible, but if they're
>   already out of cache another disk read is necessary. Without the
>   large amount of hardlinks the number of names would be non-neglibly
>   smaller

It's just a stat.
-- 
Axel.Thimm at ATrpms.net
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
Url : http://mail.ale.org/pipermail/mirror-admin/attachments/20090723/36e065f1/attachment.bin 
-------------- next part --------------
--