[fink-core] Re: Not being happy with the Master Distfiles server.

[David H. <dm...@us...>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: RIPEMD160

Peter O'Gorman wrote:
| D. H=C3=B6hn wrote:
|
| Hi David,

Hello.
|
| I have added Rob Braun to the CC line. In case you don't recall, it was
| Rob who did the coding to get fink selfupdate working with rsync and
| mirrors, and also set up the distfiles mirroring for both fink and
| opendarwin. He also has some responsibility for maintaining the
| opendarwin machines which host the master mirrors.
|
That is fine, thank you.

<snip>
|
| You give anecdotal evidence of users/mirror maintainers complaining,
| but, since you are the only contact point for the mirror maintainers, w=
e
| have seen no such complaints.
Since all of the mirroring stuff has been handled by me in the past, the
mi...@fi... email is onkly read by me. I can try to CC the
core list, however I would prefer it if the maintainers have one person
to speak to. This has worked well in the past and I would preferr it to
remain that way.

| Perhaps the mirror maintainers should be
| asked to contact fink-core with any issues?

See above. I dislike that idea a great deal. Not only will it introduce
an organisational chaos, it will also confuse them.
CC'ing the core list with updates and complaints seems like a good idea
though.

| Is a mirror maintainers
| mailing list required?
No. If so it'd run under the finkmirros domain anyways. But last time I
asked all of the maintainers they did not deem it necessary.

| Also, please post links to the users complaints,
| I could not find them.
Just skim the archive for "could not update". There are various
complaints thats ome files are not found on the Master distfiles servers
ad those files are not "restrictive". Also DRM is aware that the
distfiles had/have a lag now and then of several days, which is
something I do not deem tolerable.

|
|> There are two possibilities.
|>
|> a) We replace the current Master Distfiles server with a better
|> maintained system
|> b) We start hosting authoritative distfiles Masters in Europe, japan a=
nd
|> so on.
|
|
| I don't see the issue as being that major, the distfiles tarballs are
| always available at the original source location, if it does not update
| for a day or two what is the issue?
|
It affects the way the Userr perceive the reliability of our mirroring
system. Should they be convinced by such issues that our system is not
reliable, because it failed them 50 our of 60 times, they will resort to
using the main (original mirrors) in the first place before even trying
our mirrors and that is something I do not wish to see developing. To me
it is a major issue, because I get to deal with the rather annoyed
mirror maintainers. Especially very big mirrors like mirrors.co.uk and
belnet do not exactly like it when their scripts croak at them that some
mirror was unable to update.

| As for the reliability issues, yes, www.opendarwin.org (a.k.a. sancho)
| has some problems on occasion, I believe that the major issue with
| bringing it back to life when it dies is that it is remotely hosted and
| I'm not sure any opendarwin core members nearby have an actual key to
| the ISC, your email appears to imply laziness on their part, which is
| certainly not the case.
My email states facts, not more. It was never meant to imply anything,
nor do I say anything about their willingness to get the system up and
running or maintain it properly. I know this i a voluntary effort.
However sancho is a single point of failure when it comes to distfiles
distribution and I was planning for a long time to change that. Even if
Sancho blows up, we still have a Master in Australia and japan to update
the info files from, because they suck the data directly from CVS. I
also want to achieve that for the distfiles. THis is merely giving me
more reason to do so.

|
- -d
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.3.6 (Darwin)

iD4DBQFBl2KvPMoaMn4kKR4RA+IlAJiFarabq2AOot6znpleb7hVjcXMAKCdqmJn
XYKolJKWb//A0eKx9aFhJQ=3D=3D
=3DxjJR
-----END PGP SIGNATURE-----