Dominique Leuenberger schrieb:
> I think one problem we all experiance here (and I'm not even sure Smart
> would handle this different) is the fact, the a zmd refresh has to get
> the current catalog infos from a server (I think it's downloading
> primary.xml.gz and filelists.xml.gz and maybe even other.xml.gz, can a
> ZMD expert comment on this?)

Almost true. ;-)

zmd never downloads other.xml.gz. If it does, it's a bug. But I've never
seen that.

zmd and smart are downloading the same files for YUM repos: repomd.xml,
primary.xml.gz and filelists.xml.gz. The difference is that zmd
downloads it even if not explicitly asked to do so. But on the other
hand, the files don't need to be downloaded later because they are
already available.

> Especially in the Factory tree which is very unstable at the moment,
> these catalog files are changing regularly (several times a week) and
> thus have to be downloaded over and over again.

Let's say that the tree is "in flux" and not "unstable".

> Once the tree is considered stable, these catalogs get downloaded a
> single time and only the update tree is modified. AFAIK, zmd keeps the
> timestamps of the last update of these catalogs, and if not changed,
> decides not to download them.

Correct, neither zmd nor smart are downloading metadata from unchanged
sources again if they are already available.

> A short look on the files show, that it is downloading around 60MB at
> the moment. This might take a while. Here, probably the FTP/HTTP[S] Get
> method is not as cpu friendly as it could be, or maybe it's something
> completely different.

There is a slight difference between zmd/YaST/zypp and smart for YaST
sources: For this source type, smart does indeed download fewer files
than libzypp does.

But you get less functionality back: No translations, no disk usage
information, no patterns.

What Monkey 9 is experiencing here is probably just a bad mirror. I
blame download.suse.com. Solution: Remove this source and add a good
mirror directly.

Andreas
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to