Greetings.·
As you may know we have been working on a new version of mirrormanager, and we are finally ready to roll it out in production (barring any· show stoppers).·
mirrormanager 2 is re-written in flask and has a number of improvements over mirrormanager 1.·
We have a set of staging instances setup:·
https://admin.stg.fedoraproject.org/mirrormanager2/
Which is the new flask frontend. The data for mirror admins should· be pretty much the same. You should be able to login and check your· mirrors settings (which have been copied from the production instance).· (Note that any changes made here will not be reflected in production data)
and
https://mirrors.stg.fedoraproject.org/
Which is a mirrorlist server using the data from the staging database. You should be able to use this in place of 'mirrors.fedoraproject.org' in· yum config and the like.·
and
There is also an internal crawler instance, checking mirrors and· removing out of date ones or readding up to date ones.·
Source is of course available at:· https://github.com/fedora-infra/mirrormanager2
and bugs or issues can be reported at:· https://fedorahosted.org/mirrormanager/
Please do let us know if you see any problems or issues.· We will likely be scheduling a short outage next week to roll the new· version out to production.·
Thanks,·
kevin
Should this go to the mirror-list rather than just infrastructure?
What units is the "Bandwidth" parameter in?
How is "Max connections" used?
My mirror isn't crawled because it is a private mirror, but I do run the check-in script after every sync. Is there a way to test checking-in to the staging instance? It says last crawled in 2013, and last checked in 2014-12-05.
Thanks.
On Fri, Apr 24, 2015 at 11:32:21AM -0600, Kevin Fenzi wrote:
Greetings.·
As you may know we have been working on a new version of mirrormanager, and we are finally ready to roll it out in production (barring any· show stoppers).·
mirrormanager 2 is re-written in flask and has a number of improvements over mirrormanager 1.·
We have a set of staging instances setup:·
https://admin.stg.fedoraproject.org/mirrormanager2/
Which is the new flask frontend. The data for mirror admins should· be pretty much the same. You should be able to login and check your· mirrors settings (which have been copied from the production instance).· (Note that any changes made here will not be reflected in production data)
and
https://mirrors.stg.fedoraproject.org/
Which is a mirrorlist server using the data from the staging database. You should be able to use this in place of 'mirrors.fedoraproject.org' in· yum config and the like.·
and
There is also an internal crawler instance, checking mirrors and· removing out of date ones or readding up to date ones.·
Source is of course available at:· https://github.com/fedora-infra/mirrormanager2
and bugs or issues can be reported at:· https://fedorahosted.org/mirrormanager/
Please do let us know if you see any problems or issues.· We will likely be scheduling a short outage next week to roll the new· version out to production.·
Thanks,·
kevin
On Fri, Apr 24, 2015 at 07:50:02PM -0400, Chuck Anderson wrote:
Should this go to the mirror-list rather than just infrastructure?
It was actually send to one of the three mirror-lists. But you are right, I will forward it to the other two mirror-lists.
What units is the "Bandwidth" parameter in?
This has already been part of the old MirrroManager and is in MBit/s it used to sort the mirrorlist. The higher your bandwidth the higher the likelihood your mirror will be on the top of the list.
How is "Max connections" used?
In the old MirrorManager those fields had a description. These seems to have been lost. This is from MM1:
Maximum parallel download connections per client, suggested via metalinks.
I will try to get those descriptions also in the new MM2, so that it is clear what all the fields mean.
My mirror isn't crawled because it is a private mirror, but I do run the check-in script after every sync. Is there a way to test checking-in to the staging instance? It says last crawled in 2013, and last checked in 2014-12-05.
I was successfully using: https://admin.stg.fedoraproject.org/mirrormanager2/xmlrpc
Adrian
On Fri, Apr 24, 2015 at 11:32:21AM -0600, Kevin Fenzi wrote:
Greetings.·
As you may know we have been working on a new version of mirrormanager, and we are finally ready to roll it out in production (barring any· show stoppers).·
mirrormanager 2 is re-written in flask and has a number of improvements over mirrormanager 1.·
We have a set of staging instances setup:·
https://admin.stg.fedoraproject.org/mirrormanager2/
Which is the new flask frontend. The data for mirror admins should· be pretty much the same. You should be able to login and check your· mirrors settings (which have been copied from the production instance).· (Note that any changes made here will not be reflected in production data)
and
https://mirrors.stg.fedoraproject.org/
Which is a mirrorlist server using the data from the staging database. You should be able to use this in place of 'mirrors.fedoraproject.org' in· yum config and the like.·
and
There is also an internal crawler instance, checking mirrors and· removing out of date ones or readding up to date ones.·
Source is of course available at:· https://github.com/fedora-infra/mirrormanager2
and bugs or issues can be reported at:· https://fedorahosted.org/mirrormanager/
Please do let us know if you see any problems or issues.· We will likely be scheduling a short outage next week to roll the new· version out to production.·
Thanks,·
kevin
On Sat, 25 Apr 2015 09:35:16 +0200 Adrian Reber adrian@lisas.de wrote:
On Fri, Apr 24, 2015 at 07:50:02PM -0400, Chuck Anderson wrote:
Should this go to the mirror-list rather than just infrastructure?
It was actually send to one of the three mirror-lists. But you are right, I will forward it to the other two mirror-lists.
I actually sent it to all the mirror lists I know about. If I missed one, please let me know what it is. ;)
kevin
On Sat, Apr 25, 2015 at 09:54:40AM -0600, Kevin Fenzi wrote:
On Sat, 25 Apr 2015 09:35:16 +0200 Adrian Reber adrian@lisas.de wrote:
On Fri, Apr 24, 2015 at 07:50:02PM -0400, Chuck Anderson wrote:
Should this go to the mirror-list rather than just infrastructure?
It was actually send to one of the three mirror-lists. But you are right, I will forward it to the other two mirror-lists.
I actually sent it to all the mirror lists I know about. If I missed one, please let me know what it is. ;)
Yes I have seen it shortly after my mail. The old mirror mailing lists run at redhat.com are just a bit slower to deliver the mail. That's why I missed it.
Adrian
In MM1, Max Connections isn't used. I would be surprised if it's used in MM2. I never quite figured out how to use that value in a metalink and have it be meaningful. And yum wasn't doing parallel downloads anyhow, so it didn't ever matter. Maybe DNF will do parallel downloads and it should then matter? I don't know enough about DNF to know.
-- Matt Domsch Senior Distinguished Engineer & Executive Director Dell | Software Group, Office of the CTO
________________________________________ From: infrastructure-bounces@lists.fedoraproject.org [infrastructure-bounces@lists.fedoraproject.org] on behalf of Chuck Anderson [cra@WPI.EDU] Sent: Friday, April 24, 2015 6:50 PM To: Fedora Infrastructure Subject: Re: Upcoming Mirrormanager 2 production deployment
Should this go to the mirror-list rather than just infrastructure?
What units is the "Bandwidth" parameter in?
How is "Max connections" used?
My mirror isn't crawled because it is a private mirror, but I do run the check-in script after every sync. Is there a way to test checking-in to the staging instance? It says last crawled in 2013, and last checked in 2014-12-05.
Thanks.
On Fri, Apr 24, 2015 at 11:32:21AM -0600, Kevin Fenzi wrote:
Greetings.·
As you may know we have been working on a new version of mirrormanager, and we are finally ready to roll it out in production (barring any· show stoppers).·
mirrormanager 2 is re-written in flask and has a number of improvements over mirrormanager 1.·
We have a set of staging instances setup:·
https://admin.stg.fedoraproject.org/mirrormanager2/
Which is the new flask frontend. The data for mirror admins should· be pretty much the same. You should be able to login and check your· mirrors settings (which have been copied from the production instance).· (Note that any changes made here will not be reflected in production data)
and
https://mirrors.stg.fedoraproject.org/
Which is a mirrorlist server using the data from the staging database. You should be able to use this in place of 'mirrors.fedoraproject.org' in· yum config and the like.·
and
There is also an internal crawler instance, checking mirrors and· removing out of date ones or readding up to date ones.·
Source is of course available at:· https://github.com/fedora-infra/mirrormanager2
and bugs or issues can be reported at:· https://fedorahosted.org/mirrormanager/
Please do let us know if you see any problems or issues.· We will likely be scheduling a short outage next week to roll the new· version out to production.·
Thanks,·
kevin
_______________________________________________ infrastructure mailing list infrastructure@lists.fedoraproject.org https://admin.fedoraproject.org/mailman/listinfo/infrastructure
Dne 24.4.2015 v 19:32 Kevin Fenzi napsal(a):
Should there be any difference between these two links? Opening them in browser, I endup at:
https://admin.stg.fedoraproject.org/mirrormanager2/
and
https://admin.stg.fedoraproject.org/mirrormanager2///
respectively.
And what is Fedora 64 and RHEL 7? Is this just due to some testing data?
Vít
On Mon, Apr 27, 2015 at 09:37:36AM +0200, Vít Ondruch wrote:
Dne 24.4.2015 v 19:32 Kevin Fenzi napsal(a):
Should there be any difference between these two links? Opening them in browser, I endup at:
https://admin.stg.fedoraproject.org/mirrormanager2/
and
mirrors.stg is now redirecting to admin.stg../mirrormanager2 that is because in MirrorManager1 the matrix of products/version/arch was outside of the main application while in MirrorManager2 we integrated it.
And what is Fedora 64 and RHEL 7? Is this just due to some testing data?
AFAICS, RHEL7 might in fact be RHL7 and Fedora 64 is some kind of artifact coming from some invalid parsing of atomic data iiuc. We should try to nuke them and see if the UMDL script brings them back.
Note: MM1 [1] is even better as it has Fedora 386 \ó/ [1] https://mirrors.fedoraproject.org/publiclist/
Pierre
On Mon, 27 Apr 2015 09:59:09 +0200 Pierre-Yves Chibon pingou@pingoured.fr wrote:
On Mon, Apr 27, 2015 at 09:37:36AM +0200, Vít Ondruch wrote:
Dne 24.4.2015 v 19:32 Kevin Fenzi napsal(a):
Should there be any difference between these two links? Opening them in browser, I endup at:
https://admin.stg.fedoraproject.org/mirrormanager2/
and
mirrors.stg is now redirecting to admin.stg../mirrormanager2 that is because in MirrorManager1 the matrix of products/version/arch was outside of the main application while in MirrorManager2 we integrated it.
Right, the top level of those two links are in fact the same.
But they aren't the same otherwise... for example, in yum or dnf config you would use something like:
https://mirrors.stg.fedoraproject.org/metalink?repo=fedora-21&arch=x86_6...
to get all base fedora-21 x86_64 mirrors.
And what is Fedora 64 and RHEL 7? Is this just due to some testing data?
AFAICS, RHEL7 might in fact be RHL7 and Fedora 64 is some kind of artifact coming from some invalid parsing of atomic data iiuc. We should try to nuke them and see if the UMDL script brings them back.
RHEL7 is rhel7 beta and rc. We carried them to help with the beta and rc, but they didn't end up using mirrormanager in the urls they listed in their repo files, so it was kind of a waste. ;(
Fedora 64 might be ia64 secondary? Not sure on that one.
kevin
infrastructure@lists.fedoraproject.org
-
Adrian Reber -
Chuck Anderson -
Kevin Fenzi -
Matt Domsch -
Pierre-Yves Chibon -
Vít Ondruch