[opensuse-factory] mirrors lagging, by 200-400% the lifetime of a TW release - how to get a coherent version? - openSUSE Factory

newer
[opensuse-factory]...

[opensuse-factory] mirrors lagging, by 200-400% the lifetime of a TW release - how to get a coherent version?

older
[opensuse-factory] mirrors not...

L A Walsh

21 Dec 2019 21 Dec '19

02:52

On 2019/12/20 17:32, Knurpht-openSUSE wrote:

...

Linda, to cut a long story short: Mirrors are there for distributing the transfers for people downloading, updating, upgrading etc. If a workaround for this would go public, everyone would try download.o.o with some options first. Result: download.o.o down.

---- Usually the sites I d/l are fine and give me my max d/l speed -- not always the same for d.o.o. I certainly don't mind using a local mirror if one exists, but most importantly, is really a mirror and not 4+ days behind. Problem now, is I keep getting rerouted to what would normally be a very close, performant mirror, but now, I have no way of getting to "correct content" as I seem to have no choice in the mirror chosen, nor a way to work around it. d.o.o is effectively offline for anyone using a mirror less than or equal to the net-distance of kernel.org from anywhere. A four day lag time seems like alot for a distro like TW that updates as often as daily.

...

We may need some faster mirroring though for TW. Sometimes it comepletely rebuilds ( i.e. full package upgrade, f.e. gcc major version upgrade ) , a new release gets distributed before some mirrors have synced the previous released TW repos.

That's a big problem for anyone trying to pull down a coherent release considering the size. I haven't been able to pull down a coherent release since I switched to TW. Sometimes I've gotten remarks from some who when noticing the the noncoherence think it has something to do with changes I've made, but reality is that most of it now has to do with not being able to get a complete copy before another version is released. This supports one of my long time pet-peeves of having specific versions built into everything that they will work with and not a range or 'newer than or equal to Vx.y'. While many people come up with lots of edge cases where that system can fail. The reality I've run into is often ending up with different parts of the system that not only won't load or run, but more often -- that can't even be built from "open source[sic]" because the sources don't stay still long enough to download all of them and their deps. Certainly, its beginning to more and more seem like version requirements and differences are becoming more of a problem noticed by more people. Certainly -- with something like TW, might not cutting back to 1/week help resolve a problem? It's hard to get everything in TW working together if the versions of everything are interlocked. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Show replies by date

Neal Gompa

21 Dec 21 Dec

03:05

On Fri, Dec 20, 2019 at 8:52 PM L A Walsh wrote:

...

On 2019/12/20 17:32, Knurpht-openSUSE wrote:

...
Linda, to cut a long story short: Mirrors are there for distributing the transfers for people downloading, updating, upgrading etc. If a workaround for this would go public, everyone would try download.o.o with some options first. Result: download.o.o down.

---- Usually the sites I d/l are fine and give me my max d/l speed -- not always the same for d.o.o. I certainly don't mind using a local mirror if one exists, but most importantly, is really a mirror and not 4+ days behind. Problem now, is I keep getting rerouted to what would normally be a very close, performant mirror, but now, I have no way of getting to "correct content" as I seem to have no choice in the mirror chosen, nor a way to work around it. d.o.o is effectively offline for anyone using a mirror less than or equal to the net-distance of kernel.org from anywhere.

A four day lag time seems like alot for a distro like TW that updates as often as daily.

...
We may need some faster mirroring though for TW. Sometimes it comepletely rebuilds ( i.e. full package upgrade, f.e. gcc major version upgrade ) , a new release gets distributed before some mirrors have synced the previous released TW repos.

That's a big problem for anyone trying to pull down a coherent release considering the size. I haven't been able to pull down a coherent release since I switched to TW. Sometimes I've gotten remarks from some who when noticing the the noncoherence think it has something to do with changes I've made, but reality is that most of it now has to do with not being able to get a complete copy before another version is released.

This supports one of my long time pet-peeves of having specific versions built into everything that they will work with and not a range or 'newer than or equal to Vx.y'. While many people come up with lots of edge cases where that system can fail. The reality I've run into is often ending up with different parts of the system that not only won't load or run, but more often -- that can't even be built from "open source[sic]" because the sources don't stay still long enough to download all of them and their deps.

Certainly, its beginning to more and more seem like version requirements and differences are becoming more of a problem noticed by more people. Certainly -- with something like TW, might not cutting back to 1/week help resolve a problem? It's hard to get everything in TW working together if the versions of everything are interlocked.

openSUSE Tumbleweed does not move fast enough for mirrors to fail to keep up if they were following the recommended procedure and schedule for maintaining an openSUSE mirror. On average, we see roughly 3 snapshots a week, and in many cases, we see fewer. The problem is that YaST configures repositories on openSUSE to use the download redirector from MirrorBrain, rather than the mirrorlist. The redirector just passes you forward to a "reasonably fresh" mirror and does no further validation. From there, you're at the mercy of the mirror. There are ways to solve this problem, but because it's difficult to get YaST to change its behavior, fixing it will be quite tough. For my personal Tumbleweed systems (where I use DNF instead of Zypper as my package manager), I have an RPM installed that installs me a repo configuration that uses the metalink provided by MirrorBrain so I get a list of mirrors, DNF can do validation and selection of mirrors, and it can download packages in parallel from multiple sources, if need be. This has drastically improved the quality of service I get for software management on openSUSE. And for what it's worth, this is how system repos have been done in Fedora for over a decade. -- 真実はいつも一つ！/ Always, there's only one truth! -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Per Jessen

15:42

Neal Gompa wrote:

...

The problem is that YaST configures repositories on openSUSE to use the download redirector from MirrorBrain, rather than the mirrorlist.

Well, that is the whole point of running Mirrorbrain :-) I don't think that is a problem, Mirrobrain hands out a mirrorlist too, on every request. It is up to the client to use it. For instance, on asking for http://download.opensuse.orgtumbleweed/repo/oss/x86_64/install-initrd-openSU... I am provided with: Link: http://download.opensuse.org/tumbleweed/repo/oss/x86_64/install-initrd-openS...; rel=describedby; type="application/metalink4+xml" Link: http://download.opensuse.org/tumbleweed/repo/oss/x86_64/install-initrd-openS...; rel=describedby; type="application/x-bittorrent" Link: http://mirror.hostsuisse.com/opensuse/tumbleweed/repo/oss/x86_64/install-ini...; rel=duplicate; pri=1; geo=ch Link: http://opensusemirror.lihaso.com/tumbleweed/repo/oss/x86_64/install-initrd-o...; rel=duplicate; pri=2; geo=ch Link: http://mirror1.hs-esslingen.de/pub/Mirrors/ftp.opensuse.org/tumbleweed/repo/...; rel=duplicate; pri=3; geo=de Link: http://fr2.rpmfind.net/linux/opensuse/tumbleweed/repo/oss/x86_64/install-ini...; rel=duplicate; pri=4; geo=fr Link: http://ftp.uni-erlangen.de/pub/mirrors/opensuse/tumbleweed/repo/oss/x86_64/i...; rel=duplicate; pri=5; geo=de

...

The redirector just passes you forward to a "reasonably fresh" mirror and does no further validation. From there, you're at the mercy of the mirror.

We continually scan mirrors and check that they are up-to-date - a mirror found to be out-of-date is disabled. Wrt mirrors.edge.kernel.org, they have at least four sites, but synchronize only from two. I can only presume they have some internal synch to complement that. They use DNS geolocation to hand out the best address. If their internal sync is faulty and the optimal mirror for Linda is not up-to-date, we cannot see it - we only see the mirror in Amsterdam. -- Per Jessen, Zürich (6.1°C) Member, openSUSE Heroes. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Neal Gompa

15:58

On Sat, Dec 21, 2019 at 9:42 AM Per Jessen wrote:

...

Neal Gompa wrote:

...
The problem is that YaST configures repositories on openSUSE to use the download redirector from MirrorBrain, rather than the mirrorlist.

Well, that is the whole point of running Mirrorbrain :-) I don't think that is a problem, Mirrobrain hands out a mirrorlist too, on every request. It is up to the client to use it.

Package manager clients need to be specifically configured to fetch mirrorlists from MirrorBrain.

...

...
The redirector just passes you forward to a "reasonably fresh" mirror and does no further validation. From there, you're at the mercy of the mirror.

We continually scan mirrors and check that they are up-to-date - a mirror found to be out-of-date is disabled.

Wrt mirrors.edge.kernel.org, they have at least four sites, but synchronize only from two. I can only presume they have some internal synch to complement that. They use DNS geolocation to hand out the best address. If their internal sync is faulty and the optimal mirror for Linda is not up-to-date, we cannot see it - we only see the mirror in Amsterdam.

For me, the problem wasn't the freshness, but that I'd be redirected to mirrors that are geographically far away or regularly time out for me. Switching to the mirrorlist and having DNF do the failover selection improved my experience dramatically. -- 真実はいつも一つ！/ Always, there's only one truth! -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Per Jessen

16:14

Neal Gompa wrote:

...

On Sat, Dec 21, 2019 at 9:42 AM Per Jessen wrote:

...
Neal Gompa wrote:

...
The problem is that YaST configures repositories on openSUSE to use the download redirector from MirrorBrain, rather than the mirrorlist.

Well, that is the whole point of running Mirrorbrain :-) I don't think that is a problem, Mirrobrain hands out a mirrorlist too, on every request. It is up to the client to use it.

Package manager clients need to be specifically configured to fetch mirrorlists from MirrorBrain.

The above was a standard reply to wget which doesn't use any of the provided links, but zypper uses the mirrorlist by default.

...

...
Wrt mirrors.edge.kernel.org, they have at least four sites, but synchronize only from two. I can only presume they have some internal synch to complement that. They use DNS geolocation to hand out the best address. If their internal sync is faulty and the optimal mirror for Linda is not up-to-date, we cannot see it - we only see the mirror in Amsterdam.

For me, the problem wasn't the freshness, but that I'd be redirected to mirrors that are geographically far away or regularly time out for me.

Ah, I see. Yeah, a lot depends on how good the mirrors are in your country/network.

...

Switching to the mirrorlist and having DNF do the failover selection improved my experience dramatically.

There will be a reason why Mirrorbrain was unable to dish out better mirrors to you, but unless we/I know about it .... -- Per Jessen, Zürich (5.8°C) Member, openSUSE Heroes. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Neal Gompa

16:57

On Sat, Dec 21, 2019 at 10:18 AM Per Jessen wrote:

...

Neal Gompa wrote:

...
On Sat, Dec 21, 2019 at 9:42 AM Per Jessen wrote:

...
Neal Gompa wrote:

...
The problem is that YaST configures repositories on openSUSE to use the download redirector from MirrorBrain, rather than the mirrorlist.

Well, that is the whole point of running Mirrorbrain :-) I don't think that is a problem, Mirrobrain hands out a mirrorlist too, on every request. It is up to the client to use it.

Package manager clients need to be specifically configured to fetch mirrorlists from MirrorBrain.

The above was a standard reply to wget which doesn't use any of the provided links, but zypper uses the mirrorlist by default.

The code in "zypp/repo" in libzypp seems to indicate this is not true. Unless you use mirrorlist=/metalink= instead of baseurl=, you're not going to have client failover of mirrors based on the mirror selection provided by MirrorBrain. This is (mostly) the same behavior that DNF has. The only difference is that DNF will retry for baseurl downloads a couple of times in hopes of getting a working connection from the server. I'm pretty sure that has to do with handling download redirectors, but it's still not as reliable as using a mirrorlist directly.

...

...
...
Wrt mirrors.edge.kernel.org, they have at least four sites, but synchronize only from two. I can only presume they have some internal synch to complement that. They use DNS geolocation to hand out the best address. If their internal sync is faulty and the optimal mirror for Linda is not up-to-date, we cannot see it - we only see the mirror in Amsterdam.

For me, the problem wasn't the freshness, but that I'd be redirected to mirrors that are geographically far away or regularly time out for me.

Ah, I see. Yeah, a lot depends on how good the mirrors are in your country/network.

Well, despite living in the US east coast, I usually get redirected to a mirror in central or eastern Europe when Provo isn't working. When Provo is working, I always get that. I don't seem to get anything else. :(

...

...
Switching to the mirrorlist and having DNF do the failover selection improved my experience dramatically.

There will be a reason why Mirrorbrain was unable to dish out better mirrors to you, but unless we/I know about it ....

I've had this problem for years, and I've asked about it before. It's pretty hard to debug apparently. :( -- 真実はいつも一つ！/ Always, there's only one truth! -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Per Jessen

17:39

Neal Gompa wrote:

...

On Sat, Dec 21, 2019 at 10:18 AM Per Jessen wrote:

...
Neal Gompa wrote:

...
On Sat, Dec 21, 2019 at 9:42 AM Per Jessen wrote:

...
Neal Gompa wrote:

...
The problem is that YaST configures repositories on openSUSE to use the download redirector from MirrorBrain, rather than the mirrorlist.

Well, that is the whole point of running Mirrorbrain :-) I don't think that is a problem, Mirrobrain hands out a mirrorlist too, on every request. It is up to the client to use it.

Package manager clients need to be specifically configured to fetch mirrorlists from MirrorBrain.

The above was a standard reply to wget which doesn't use any of the provided links, but zypper uses the mirrorlist by default.

The code in "zypp/repo" in libzypp seems to indicate this is not true.

Oh. That doesn't match my experience, but I've been using my own mirror for a few years now. I wrote a caching setup for squid some years back, at that time, an installation would use all provided mirrors and fetch packages in 256K chunks, spread across a number of mirrors. The idea was, AFAIU, to use the maximum available downlink bandwidth whilst still being nice to the mirrors. I hope we still do that.

...

Unless you use mirrorlist=/metalink= instead of baseurl=, you're not going to have client failover of mirrors based on the mirror selection provided by MirrorBrain.

That certainly seems to be a mistake - surely failover is the first thing we would always want.

...

...
...
...
Wrt mirrors.edge.kernel.org, they have at least four sites, but synchronize only from two. I can only presume they have some internal synch to complement that. They use DNS geolocation to hand out the best address. If their internal sync is faulty and the optimal mirror for Linda is not up-to-date, we cannot see it - we only see the mirror in Amsterdam.

For me, the problem wasn't the freshness, but that I'd be redirected to mirrors that are geographically far away or regularly time out for me.

Ah, I see. Yeah, a lot depends on how good the mirrors are in your country/network.

Well, despite living in the US east coast, I usually get redirected to a mirror in central or eastern Europe when Provo isn't working. When Provo is working, I always get that. I don't seem to get anything else. :(

The US mirroring situation is not as good as the European ditto, but depending on what you need, I would think there should still be sufficient mirrors available. One possibility is that we're not able to properly geolocate your IP address. The other is that TW is not very well mirrored in the US.

...

...
...
Switching to the mirrorlist and having DNF do the failover selection improved my experience dramatically.

There will be a reason why Mirrorbrain was unable to dish out better mirrors to you, but unless we/I know about it ....

I've had this problem for years, and I've asked about it before. It's pretty hard to debug apparently. :(

Please report to admin@o.o and I'll be happy to have a look, although after Christmas. I need an example of IP addresses you would typically be accessing from. -- Per Jessen, Zürich (5.3°C) Member, openSUSE Heroes. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Neal Gompa

18:47

On Sat, Dec 21, 2019 at 11:39 AM Per Jessen wrote:

...

Neal Gompa wrote:

...
On Sat, Dec 21, 2019 at 10:18 AM Per Jessen wrote:

...
Neal Gompa wrote:

...
On Sat, Dec 21, 2019 at 9:42 AM Per Jessen wrote:

...
Neal Gompa wrote:

...
The problem is that YaST configures repositories on openSUSE to use the download redirector from MirrorBrain, rather than the mirrorlist.

Well, that is the whole point of running Mirrorbrain :-) I don't think that is a problem, Mirrobrain hands out a mirrorlist too, on every request. It is up to the client to use it.

Package manager clients need to be specifically configured to fetch mirrorlists from MirrorBrain.

The above was a standard reply to wget which doesn't use any of the provided links, but zypper uses the mirrorlist by default.

The code in "zypp/repo" in libzypp seems to indicate this is not true.

Oh. That doesn't match my experience, but I've been using my own mirror for a few years now. I wrote a caching setup for squid some years back, at that time, an installation would use all provided mirrors and fetch packages in 256K chunks, spread across a number of mirrors. The idea was, AFAIU, to use the maximum available downlink bandwidth whilst still being nice to the mirrors. I hope we still do that.

...
Unless you use mirrorlist=/metalink= instead of baseurl=, you're not going to have client failover of mirrors based on the mirror selection provided by MirrorBrain.

That certainly seems to be a mistake - surely failover is the first thing we would always want.

Ideally, repos should be configured to fetch the mirrorlist.

...

...
...
...
...
Wrt mirrors.edge.kernel.org, they have at least four sites, but synchronize only from two. I can only presume they have some internal synch to complement that. They use DNS geolocation to hand out the best address. If their internal sync is faulty and the optimal mirror for Linda is not up-to-date, we cannot see it - we only see the mirror in Amsterdam.

For me, the problem wasn't the freshness, but that I'd be redirected to mirrors that are geographically far away or regularly time out for me.

Ah, I see. Yeah, a lot depends on how good the mirrors are in your country/network.

Well, despite living in the US east coast, I usually get redirected to a mirror in central or eastern Europe when Provo isn't working. When Provo is working, I always get that. I don't seem to get anything else. :(

The US mirroring situation is not as good as the European ditto, but depending on what you need, I would think there should still be sufficient mirrors available. One possibility is that we're not able to properly geolocate your IP address. The other is that TW is not very well mirrored in the US.

Equal possibilities. I suspect the latter more than the former, though.

...

...
...
...
Switching to the mirrorlist and having DNF do the failover selection improved my experience dramatically.

There will be a reason why Mirrorbrain was unable to dish out better mirrors to you, but unless we/I know about it ....

I've had this problem for years, and I've asked about it before. It's pretty hard to debug apparently. :(

Please report to admin@o.o and I'll be happy to have a look, although after Christmas. I need an example of IP addresses you would typically be accessing from.

When I'm back home after Christmas, I'll try to send a report. :) -- 真実はいつも一つ！/ Always, there's only one truth! -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Felix Miata

19:33

Per Jessen composed on 2019-12-21 18:39 (UTC+0100):

...

Neal Gompa wrote:

...

...
Well, despite living in the US east coast, I usually get redirected to a mirror in central or eastern Europe when Provo isn't working. When Provo is working, I always get that. I don't seem to get anything else. :(

...

The US mirroring situation is not as good as the European ditto, but depending on what you need, I would think there should still be sufficient mirrors available. One possibility is that we're not able to properly geolocate your IP address. The other is that TW is not very well mirrored in the US. There *is* a serious dearth of North American mirrors of TW repos, only in the western quarter AFAICT (San Jose & Provo), though there is a bunch that provide TW isos. Here in the southeast peninsula I'm usually directed by d.o.o. to western Europe. -- Evolution as taught in public schools is religion, not science.

Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Jan Engelhardt

20:28

On Saturday 2019-12-21 20:33, Felix Miata wrote:

...

There *is* a serious dearth of North American mirrors of TW repos,[...] in the southeast peninsula I'm usually directed by d.o.o. to western Europe.

Ah, the sound of globalization! Just as local farms shut down, because importing everything from overseas is cheaper. Now it happens for bits and bytes too :-D -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Felix Miata

22:23

Jan Engelhardt composed on 2019-12-21 21:28 (UTC+0100):

...

On Saturday 2019-12-21 20:33, Felix Miata wrote:

...

...
There *is* a serious dearth of North American mirrors of TW repos,[...] in the southeast peninsula I'm usually directed by d.o.o. to western Europe.

...

Ah, the sound of globalization! Just as local farms shut down, because importing everything from overseas is cheaper. Now it happens for bits and bytes too :-D

I have no idea what you might be implying. I was only pointing out that distance to Europe's TW mirrors is rather long. From where I am, some distances: Miami 377 km none found Atlanta 608 km none found St Louis 1,332 km none found Dallas 1,454 km none found Chicago 1,556 km none found Toronto 1,700 km none found Boston 1,842 km none found Provo 2,965 km Manaus 4,204 km ufam.edu.br mirror San Jose 3,783 km Seattle 4,024 km none found Dublin 6,654 km heanet.ie mirror Oxford 7,017 km ox.ac.uk mirror Paris 7,343 km free.fr mirror Amsterdam 7,408 km ?nluug.nl mirror Göttingen 7,763 km gwdg.de mirror -- Evolution as taught in public schools is religion, not science. Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/ -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Andrei Borzenkov

22 Dec 22 Dec

06:47

21.12.2019 19:57, Neal Gompa пишет:

...

On Sat, Dec 21, 2019 at 10:18 AM Per Jessen wrote:

...
Neal Gompa wrote:

...
On Sat, Dec 21, 2019 at 9:42 AM Per Jessen wrote:

...
Neal Gompa wrote:

...
The problem is that YaST configures repositories on openSUSE to use the download redirector from MirrorBrain, rather than the mirrorlist.

Well, that is the whole point of running Mirrorbrain :-) I don't think that is a problem, Mirrobrain hands out a mirrorlist too, on every request. It is up to the client to use it.

Package manager clients need to be specifically configured to fetch mirrorlists from MirrorBrain.

The above was a standard reply to wget which doesn't use any of the provided links, but zypper uses the mirrorlist by default.

The code in "zypp/repo" in libzypp seems to indicate this is not true. Unless you use mirrorlist=/metalink= instead of baseurl=, you're not going to have client failover of mirrors based on the mirror selection provided by MirrorBrain.

The problem is that as of now there seems to be no user facing frontend to configure mirror lists instead of base URL. So while libzypp supports failover, users must manually edit repo definitions to enable it. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Carlos E. R.

21 Dec 21 Dec

12:54

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 21/12/2019 03.52, L A Walsh wrote: ...

...

That's a big problem for anyone trying to pull down a coherent release considering the size. I haven't been able to pull down a coherent release since I switched to TW. Sometimes I've gotten remarks from some who when noticing the the noncoherence think it has something to do with changes I've made, but reality is that most of it now has to do with not being able to get a complete copy before another version is released.

There is another posibility: there is an archive of photos taken of Tumbleweed, that stays put for weeks, perhaps. Let me see if I can find the URL. [...] http://download.opensuse.org/history/ Goes back a month. But sources are not included, only oss and non-oss. - -- Cheers / Saludos, Carlos E. R. (from 15.1 x86_64 at Telcontar) -----BEGIN PGP SIGNATURE----- iF0EARECAB0WIQQZEb51mJKK1KpcU/W1MxgcbY1H1QUCXf4V7gAKCRC1MxgcbY1H 1YLUAJ90m+GELsM19kMMom1M640HOKOzUgCfb6H0IqoqLafhYTs68q4qLuPmOFA= =j/62 -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

Per Jessen

14:14

L A Walsh wrote:

...

On 2019/12/20 17:32, Knurpht-openSUSE wrote:

...
Linda, to cut a long story short: Mirrors are there for distributing the transfers for people downloading, updating, upgrading etc. If a workaround for this would go public, everyone would try download.o.o with some options first. Result: download.o.o down.

---- Usually the sites I d/l are fine and give me my max d/l speed -- not always the same for d.o.o. I certainly don't mind using a local mirror if one exists, but most importantly, is really a mirror and not 4+ days behind.

Any mirror that is that far behind should have been automagically disabled. For now, mirror.edge.kernel.org seems up-to-date though. (at least the copy in Amsterdam),

...

Problem now is I keep getting rerouted to what would normally be a very close, performant mirror, but now, I have no way of getting to "correct content" as I seem to have no choice in the mirror chosen, nor a way to work around it.

Your only option is to manually select a specific mirror.

...

A four day lag time seems like alot for a distro like TW that updates as often as daily.

According to the logs, for the last 21 days, kernel.org have been running an rsync twice a day, from two mirrors. If one of the sites (hosted by "Packet") were out of date, it sounds like it might be an internal issue. -- Per Jessen, Zürich (9.3°C) Member, openSUSE Heroes. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org

1593

Age (days ago)

1594

Last active (days ago)

List overview

Download

13 comments

7 participants

participants (7)

Andrei Borzenkov
Carlos E. R.
Felix Miata
Jan Engelhardt
L A Walsh
Neal Gompa
Per Jessen

[opensuse-factory] mirrors lagging, by 200-400% the lifetime of a TW release - how to get a coherent version?

tags

participants (7)