Re: New Debian mirror submission (mirrors.tuna.tsinghua.edu.cn)

[Simon Paillard]

Hi,

On Wed, Jul 04, 2012 at 07:57:57AM +0000, TUNA wrote:
> Submission-Type: new

Thanks for mirroring Debian.
Before we can include your mirror in our official list, there are a few items to check.

> Site: mirrors.tuna.tsinghua.edu.cn

Could you please :
* update your ftpsync version
* make sure you are subscribed
to https://lists.debian.org/debian-mirrors-announce/ where are announced new
versions ?
* set MIRRORNAME to mirrors.tuna.tsinghua.edu.cn in ftpsync*.conf

> Type: leaf
> Archive-architecture: amd64 i386
> Archive-http: /debian/
> Backports-http: /debian-backports/

You appear to mirror debian-cd as well, but the mirror is outdated (6.0.5 is current).
Would you consider updating it ?
http://mirrors.tuna.tsinghua.edu.cn/debian-cd/

> IPv6: yes
> Archive-upstream: ftp.fi.debian.org

Why not ftp.cn.debian.org ?

> Backports-upstream: ftp.cn.debian.org
> Updates: once

Given http://www.debian.org/mirror/ftpmirror#when
Could you update the archive 4 times a day ?

> Maintainer: TUNA <thu-opensourc...@googlegroups.com>
> Country: CN China
> Location: Beijing
> Sponsor: Tsinghua University http://www.tsinghua.edu.cn/
> Comment: bandwidth : 1GBps CERNET & CERNET2

Thanks for you answer and best regards.

--
Simon Paillard

[Aron Xu]

我下下个星期到北京，可以试试把 mirror 做成 push 的，另外视硬盘空间情况申请 ftp2.cn.debian.org
域名。建议暂缓这个申请一两个星期。上面提出的要求虽然问得比较客气，但如果不改实际上就无法进 official mirror list。

--
Regards,
Aron Xu

[Qijiang Fan]

Hello,
1. You appear to mirror debian-cd as well, but the mirror is outdated
(6.0.5 is current).
We mirror debian-cd from ftp.cn.debian.org,
but the current version on ftp.cn.debian.org/debian-cd is 6.0.4.
We have change our upstream to ftp.de.d.o.
We use debian
and we only mirror CD-1 and DVD-1 of amd64 and i386 for people to
install debian without internet.
We do NOT mirror other CDs and DVDs for the lack of sapce.That's why
we don't apply to be added to the list of debian-cd.
$ df -h | grep mirror
mirror-nas:/srv/ftp/ 3.7T 3.4T
147G 96% /mirror
The mirror script of debian-cd can be found at
http://git.tuna.tsinghua.edu.cn/?p=mirror/scripts.git;a=blob;f=bin/sync.debcd;h=023acd35bc9f56d060c81cb3deadac2c3eb5aa63;hb=HEAD

2. Our address can not subscribe debian's list because our address
thu-opensourc...@googlegroups.com has the words "admin",
which will fail the procmail's check.

3. Other questions will be answered by Xu Benda ( heroxbd in the CC's List).


2012/7/6 Simon Paillard <spai...@debian.org>:

> You appear to mirror debian-cd as well, but the mirror is outdated (6.0.5 is current).
> Would you consider updating it ?
> http://mirrors.tuna.tsinghua.edu.cn/debian-cd/

--
喵～～

[her...@gmail.com]

Dear Simon,

Simon Paillard <spai...@debian.org> writes:

>> Site: mirrors.tuna.tsinghua.edu.cn
>
> Could you please :
> * update your ftpsync version

Underway. tracked at http://issues.tuna.tsinghua.edu.cn/issue221.

Is another bug report at debbugs (tracking the inclusion of mirror.tuna)
preferable?

> * make sure you are subscribed
> to https://lists.debian.org/debian-mirrors-announce/ where are announced new
> versions ?

Done.

> * set MIRRORNAME to mirrors.tuna.tsinghua.edu.cn in ftpsync*.conf

DONE

>> IPv6: yes
>> Archive-upstream: ftp.fi.debian.org
>
> Why not ftp.cn.debian.org ?

It was changed to fi mirror once ftp.cn.d.o left in an incomplete
status.

as suggest by ftp.cn.d.o maintainer, at
http://issues.tuna.tsinghua.edu.cn/msg1243

Translating from Chinese, it is concluded that the bandwidth between
ftp.cn.d.o and syncproxy is lower than that between ftp.cn.d.o and
ftp.de.d.o. Such problem is the cause of incomplete state.

We tested many eu mirrors, and selected ftp.fi.

>> Backports-upstream: ftp.cn.debian.org
>> Updates: once
>
> Given http://www.debian.org/mirror/ftpmirror#when
> Could you update the archive 4 times a day ?

DONE.

Our final mission is to become another primary mirror in P.R.China
(ftp2.cn.debian.org) and accept push. What other requirements should we
meet in order to realize that?

Yours,
Benda

[her...@gmail.com]

Hi, Aron,

Aron Xu <happya...@gmail.com> writes:

> 我下下个星期到北京，可以试试把 mirror 做成 push 的，

支持！

> 另外视硬盘空间情况申请 ftp2.cn.debian.org域名。

硬盘空间不是问题，我们仍有另外的 2T + 5T = 7TB 的 NAS 未用。

同步所有 arch 没问题。

> 建议暂缓这个申请一两个星期。

为何暂缓，等你归京还是有别的原因？

> 上面提出的要求虽然问得比较客气，但如果不改实际上就无法进 official
> mirror list。

现在的实际工作只剩下 issue221 了。aron 归京后是否愿意来做？

本达

[Aron Xu]

2012/7/6 <her...@gmail.com>:

> Hi, Aron,
>
> Aron Xu <happya...@gmail.com> writes:
>
>> 我下下个星期到北京，可以试试把 mirror 做成 push 的，
>
> 支持！
>
>> 另外视硬盘空间情况申请 ftp2.cn.debian.org域名。
>
> 硬盘空间不是问题，我们仍有另外的 2T + 5T = 7TB 的 NAS 未用。
>
> 同步所有 arch 没问题。
>
>> 建议暂缓这个申请一两个星期。
>
> 为何暂缓，等你归京还是有别的原因？
>

如果规定没变的话提交 bug report 才算正式提出申请，但我们还有一些准备工作要做，比如所有 arch
的初始同步，以及下面说到的同步脚本改写，都最好在申请之前搞定。当前用的这个脚本合并新版难度不大，但是因为改了用法所有做 push
mirror 比较麻烦。如果申请到了 ftp2.cn.debian.org 域名还要承担起向下游服务器推送 push
的工作（虽然可能非常少甚至没有，但得准备好），现在的脚本貌似都要做些改动。

>> 上面提出的要求虽然问得比较客气，但如果不改实际上就无法进 official
>> mirror list。
>
> 现在的实际工作只剩下 issue221 了。aron 归京后是否愿意来做？
>
> 本达

主要问题是现在用一个脚本同步了很多发行版的仓库，脚本第一个参数是发行版名称，但上游服务器发来的 push
不会按照这个格式来做调整。这个问题可能有一点影响，需要有个决定看怎样改比较合适。

--
Regards,
Aron Xu

[Qijiang Fan]

可以给debian单独脚本。就像debian testing cd或者pypi那样。
-----原信息-----
发自： Aron Xu
发送： 2012/07/08, 23:16
收件人 thu-opensourc...@googlegroups.com
主题： Re: [THU Mirror:1500] Re: New Debian mirror submission
(mirrors.tuna.tsinghua.edu.cn)

[Aron Xu]

Hi Simon,

I'm working with the team to make this mirror offical, and eventually get the
ftp2.cn.debian.org alias as I've suggested to you. I'd like to introduce about
some details of our work to address your concerns.

We are working on the initial sync of other architectures that weren't synced
before using an up-to-date version of ftysync, and setting up ftp and rsync
services which are required for the qualification of ftpX.xy.debian.org.

The team is very responsive comparing with most other mirror providers I've
worked with before, and they do have good skills to work out all kinds of
problems.

The mirror is hosted in Tsinghua University's DragonLab, where Tsinghua
University is the national core exchange of CERNET and CERNET2, so I believe
peering is not a problem. We have tested the mirror from several locations and
ISPs, it appears to be good enough to become an primary mirror even though
there might be local mirror being a bit faster for some users.

At last I'd like to ask for your help to set up push mirroring from ftpmaster
or syncproxy for this mirror, essentially to make it a Push-Primary one. It
won't take too long before we finish the initical sync of all architectures and
we'll be able to work on this issue then.


--
Regards,
Aron Xu