[arch-mirrors] miscellaneous
Dear all, I am an administrator of ftp.halifax.rwth-aachen.de, which is already registered as an Archlinux mirror. I just noticed two strange things about being an Archlinux mirror. 1) The wiki at https://wiki.archlinux.org/index.php/Mirrors does not show my mirror, although it is registered. I cannot edit this page. I think it is outdated and should be replaced by some automated list. 2) On https://wiki.archlinux.org/index.php/DeveloperWiki:NewMirrors it is stated "Do not rsync more rapidly than every hour". However, on http://www.archlinux.org/mirrors/status/ I see several mirrors with a low (good) score and an average delay of less than an hour. Most of these are not listed as tier 1 mirrors. As an admin of ftp.halifax.rwth-aachen.de I'd like to get as much traffic as possible. However, if I play be the rules, my score is not as good as it should be. Please clarify. Since then, we're synchronizing every 30 minutes. PS: What about push triggers? Thanks, -- Carsten Otto otto@informatik.rwth-aachen.de LuFG Informatik 2 http://verify.rwth-aachen.de/otto/ RWTH Aachen phone: +49 241 80-21211
Am 29.11.2011 15:06, schrieb Carsten Otto:
Dear all,
I am an administrator of ftp.halifax.rwth-aachen.de, which is already registered as an Archlinux mirror. I just noticed two strange things about being an Archlinux mirror.
Hi. I am not sure if anybody answered your mail yet, so I just go ahead.
1) The wiki at https://wiki.archlinux.org/index.php/Mirrors does not show my mirror, although it is registered. I cannot edit this page. I think it is outdated and should be replaced by some automated list.
This wiki entry just contains a few "unofficial" mirrors. The official source for such information is https://www.archlinux.org/mirrors/ These lists are automatically generated.
2) On https://wiki.archlinux.org/index.php/DeveloperWiki:NewMirrors it is stated "Do not rsync more rapidly than every hour". However, on http://www.archlinux.org/mirrors/status/ I see several mirrors with a low (good) score and an average delay of less than an hour. Most of these are not listed as tier 1 mirrors.
Tier 1 mirrors sync hourly. So it does not make much sense to sync more often than your upstream mirror. However, you might reduce the delay if you have try to sync eight after your tier 1 mirror has finished. E.g. if you upstream still is https://www.archlinux.org/mirrors/uk2.net/ you might sync a few minutes after **:30 every hour.
As an admin of ftp.halifax.rwth-aachen.de I'd like to get as much traffic as possible. However, if I play be the rules, my score is not as good as it should be. Please clarify. Since then, we're synchronizing every 30 minutes.
As I said, this wont help that much if your upstream mirror only syncs hourly. Syncing on the right minute will ensure that most of the time the overall delay will be below one hour. Also note that we setup this multi tier scheme to reduce traffic on our main server which has a bandwidth limit of 12 MBit and also has to serve the website, email and bug tracker. So we have to be careful about how many mirrors sync how often.
PS: What about push triggers?
I didn't look into details of this. This might give us more control about when which mirrors sync from us. However, this might not work well with our multi tier setup. E.g. we could notify our tier 1 mirror about changes, but it would be up to them to notify their tier 2 mirrors. (none of the tier 1 mirrors that serves as tier 2 upstream is controlled by us) I hope this clarifies a few things. If not don't hesitate to ask me. Greetings, Pierre -- Pierre Schmitz, http://pierre-schmitz.com
Hi, On Sun, Dec 11, 2011 at 12:09:05AM +0100, Pierre Schmitz wrote:
Hi. I am not sure if anybody answered your mail yet, so I just go ahead.
Your mail is the first reply - thanks for that!
This wiki entry just contains a few "unofficial" mirrors. The official source for such information is https://www.archlinux.org/mirrors/ These lists are automatically generated.
I see. Confusing, nevertheless :)
Tier 1 mirrors sync hourly. So it does not make much sense to sync more often than your upstream mirror. However, you might reduce the delay if you have try to sync eight after your tier 1 mirror has finished. E.g. if you upstream still is https://www.archlinux.org/mirrors/uk2.net/ you might sync a few minutes after **:30 every hour.
OK.
Also note that we setup this multi tier scheme to reduce traffic on our main server which has a bandwidth limit of 12 MBit and also has to serve the website, email and bug tracker. So we have to be careful about how many mirrors sync how often.
I see the problem, but do not really understand how that relates to my question.
I didn't look into details of this. This might give us more control about when which mirrors sync from us. However, this might not work well with our multi tier setup. E.g. we could notify our tier 1 mirror about changes, but it would be up to them to notify their tier 2 mirrors. (none of the tier 1 mirrors that serves as tier 2 upstream is controlled by us)
You can learn a lot for Debian here, there is no need to reinvent the wheel.
I hope this clarifies a few things. If not don't hesitate to ask me.
Yes, it does. Thanks! Bye, -- Carsten Otto otto@informatik.rwth-aachen.de LuFG Informatik 2 http://verify.rwth-aachen.de/otto/ RWTH Aachen phone: +49 241 80-21211
Hi, rsync rsync://archlinux.mirrors.uk2.net/ rsync: failed to connect to archlinux.mirrors.uk2.net (77.92.68.20): Connection refused (111) rsync error: error in socket IO (code 10) at clientserver.c(122) [Receiver=3.0.9] Mitry.
On Thu, 15 Dec 2011 13:00:18 +0300 Mitry Matyushkov <ftp@mgts.by> wrote:
rsync rsync://archlinux.mirrors.uk2.net/ rsync: failed to connect to archlinux.mirrors.uk2.net (77.92.68.20): Connection refused (111) rsync error: error in socket IO (code 10) at clientserver.c(122) [Receiver=3.0.9]
This is fixed already, apologies for the inconvenience. I'm still investigating what caused the rsync daemon to exit or be killed, and putting additional monitoring in place to ensure I'm alerted immediately if it happens again. -- David Precious ("bigpresh") <davidp@preshweb.co.uk> http://www.preshweb.co.uk/ www.preshweb.co.uk/twitter www.preshweb.co.uk/linkedin www.preshweb.co.uk/facebook www.preshweb.co.uk/cpan www.preshweb.co.uk/github
participants (4)
-
Carsten Otto
-
David Precious
-
Mitry Matyushkov
-
Pierre Schmitz