[opensuse] Where to report bad mirrors?

older
[opensuse] Trying to create a new...

Carlos E. R.

10 Jun 2019 10 Jun '19

17:22

This mirror is too slow: --2019-06-10 19:17:10-- http://download.opensuse.org/update/leap/15.1/oss/x86_64/kernel-default-deve... Connecting to download.opensuse.org (download.opensuse.org)|195.135.221.134|:80... connected. HTTP request sent, awaiting response... 302 Found Location: http://ftp.rnl.tecnico.ulisboa.pt/pub/opensuse/update/leap/15.1/oss/x86_64/k... [following] --2019-06-10 19:17:11-- http://ftp.rnl.tecnico.ulisboa.pt/pub/opensuse/update/leap/15.1/oss/x86_64/k... Connecting to ftp.rnl.tecnico.ulisboa.pt (ftp.rnl.tecnico.ulisboa.pt)|193.136.164.6|:80... connected. HTTP request sent, awaiting response... 200 OK Length: 8314408 (7,9M) [application/x-redhat-package-manager] Saving to: ‘kernel-default-devel-4.12.14-lp151.28.4.1.x86_64.rpm’ kernel-default-devel-4.12.14-lp151.28.4.1.x86_64.r 98%[==========================================================================================================> ] 7,78M 325KB/s eta 1s Whom do I report it to, admin@o.o? -- Cheers / Saludos, Carlos E. R. (from 15.0 x86_64 at Telcontar)

Attachments:

signature.asc (application/pgp-signature — 195 bytes)

Show replies by date

Per Jessen

10 Jun 10 Jun

17:53

Carlos E. R. wrote:

...

This mirror is too slow:

--2019-06-10 19:17:10-- [snip]

Whom do I report it to, admin@o.o?

It's worth a try, it'll go to the mirror admin - that's me :-) There isn't much we can do about a mirror being "slow". Anton Aylward has brought up the topic too. Most likely it is not the mirror that is slow, but the end-to-end connection. Your problem in Spain is also that there are no Spanish mirrors. -- Per Jessen, Zürich (15.9°C) member, openSUSE Heroes. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos E. R.

21:25

On 10/06/2019 19.53, Per Jessen wrote:

...

Carlos E. R. wrote:

...
This mirror is too slow:

--2019-06-10 19:17:10-- [snip]

Whom do I report it to, admin@o.o?

It's worth a try, it'll go to the mirror admin - that's me :-)

There isn't much we can do about a mirror being "slow". Anton Aylward has brought up the topic too. Most likely it is not the mirror that is slow, but the end-to-end connection.

300 Kb/s when my connection is 100 Mb/s is very slow... I remember that thread, but couldn't find it. Yes, we would need a local blacklist of mirrors. Otherwise I'll have to bypass mirrorbrain and list manually some mirrors.

...

Your problem in Spain is also that there are no Spanish mirrors.

Right, but we use other central European servers, like gwdg, which are very fast. -- Cheers / Saludos, Carlos E. R. (from 15.0 x86_64 at Telcontar)

Patrick Shanahan

23:18

* Carlos E. R. [06-10-19 17:28]:

...

On 10/06/2019 19.53, Per Jessen wrote:

...
Carlos E. R. wrote:

...
This mirror is too slow:

--2019-06-10 19:17:10-- [snip]

Whom do I report it to, admin@o.o?

It's worth a try, it'll go to the mirror admin - that's me :-)

There isn't much we can do about a mirror being "slow". Anton Aylward has brought up the topic too. Most likely it is not the mirror that is slow, but the end-to-end connection.

300 Kb/s when my connection is 100 Mb/s is very slow...

I remember that thread, but couldn't find it. Yes, we would need a local blacklist of mirrors. Otherwise I'll have to bypass mirrorbrain and list manually some mirrors.

...
Your problem in Spain is also that there are no Spanish mirrors.

Right, but we use other central European servers, like gwdg, which are very fast.

then add the *specific* servers as I described earlier, and do not let mirrorbrained decide which server you use. -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.org openSUSE Community Member facebook/ptilopteri Registered Linux User #207535 @ http://linuxcounter.net Photos: http://wahoo.no-ip.org/piwigo paka @ IRCnet freenode -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos E. R.

11 Jun 11 Jun

06:30

On 11/06/2019 01.18, Patrick Shanahan wrote:

...

* Carlos E. R. [06-10-19 17:28]:

...
On 10/06/2019 19.53, Per Jessen wrote:

...
Carlos E. R. wrote:

...
This mirror is too slow:

--2019-06-10 19:17:10-- [snip]

Whom do I report it to, admin@o.o?

It's worth a try, it'll go to the mirror admin - that's me :-)

There isn't much we can do about a mirror being "slow". Anton Aylward has brought up the topic too. Most likely it is not the mirror that is slow, but the end-to-end connection.

300 Kb/s when my connection is 100 Mb/s is very slow...

I remember that thread, but couldn't find it. Yes, we would need a local blacklist of mirrors. Otherwise I'll have to bypass mirrorbrain and list manually some mirrors.

...
Your problem in Spain is also that there are no Spanish mirrors.

Right, but we use other central European servers, like gwdg, which are very fast.

then add the *specific* servers as I described earlier, and do not let mirrorbrained decide which server you use.

Which is what I said two paragraphs above (re bypass mirrorbrain). -- Cheers / Saludos, Carlos E. R. (from 15.0 x86_64 at Telcontar)

Patrick Shanahan

11:37

* Carlos E. R. [06-11-19 02:32]:

...

On 11/06/2019 01.18, Patrick Shanahan wrote:

...
* Carlos E. R. [06-10-19 17:28]:

...
On 10/06/2019 19.53, Per Jessen wrote:

...
Carlos E. R. wrote:

...
This mirror is too slow:

--2019-06-10 19:17:10-- [snip]

Whom do I report it to, admin@o.o?

It's worth a try, it'll go to the mirror admin - that's me :-)

There isn't much we can do about a mirror being "slow". Anton Aylward has brought up the topic too. Most likely it is not the mirror that is slow, but the end-to-end connection.

300 Kb/s when my connection is 100 Mb/s is very slow...

I remember that thread, but couldn't find it. Yes, we would need a local blacklist of mirrors. Otherwise I'll have to bypass mirrorbrain and list manually some mirrors.

...
Your problem in Spain is also that there are no Spanish mirrors.

Right, but we use other central European servers, like gwdg, which are very fast.

then add the *specific* servers as I described earlier, and do not let mirrorbrained decide which server you use.

Which is what I said two paragraphs above (re bypass mirrorbrain).

but *are* you doing that? -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.org openSUSE Community Member facebook/ptilopteri Registered Linux User #207535 @ http://linuxcounter.net Photos: http://wahoo.no-ip.org/piwigo paka @ IRCnet freenode -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos E. R.

11:56

On 11/06/2019 13.37, Patrick Shanahan wrote:

...

* Carlos E. R. <> [06-11-19 02:32]:

...
On 11/06/2019 01.18, Patrick Shanahan wrote:

...
* Carlos E. R. <> [06-10-19 17:28]:

...
On 10/06/2019 19.53, Per Jessen wrote:

...
Carlos E. R. wrote:

...

...
...
...
...
Your problem in Spain is also that there are no Spanish mirrors.

Right, but we use other central European servers, like gwdg, which are very fast.

then add the *specific* servers as I described earlier, and do not let mirrorbrained decide which server you use.

Which is what I said two paragraphs above (re bypass mirrorbrain).

but *are* you doing that?

Not when I reported. And doesn't work with wget. The problem is not with zypper, but wget. -- Cheers / Saludos, Carlos E. R. (from openSUSE 15.0 (Legolas))

Patrick Shanahan

13:05

* Carlos E. R. [06-11-19 07:58]:

...

On 11/06/2019 13.37, Patrick Shanahan wrote:

...
* Carlos E. R. <> [06-11-19 02:32]:

...
On 11/06/2019 01.18, Patrick Shanahan wrote:

...
* Carlos E. R. <> [06-10-19 17:28]:

...
On 10/06/2019 19.53, Per Jessen wrote:

...
Carlos E. R. wrote:

...
...
...
...
...
Your problem in Spain is also that there are no Spanish mirrors.

Right, but we use other central European servers, like gwdg, which are very fast.

then add the *specific* servers as I described earlier, and do not let mirrorbrained decide which server you use.

Which is what I said two paragraphs above (re bypass mirrorbrain).

but *are* you doing that?

Not when I reported. And doesn't work with wget. The problem is not with zypper, but wget.

you aren't directing wget to a *specific* site? ??? -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.org openSUSE Community Member facebook/ptilopteri Registered Linux User #207535 @ http://linuxcounter.net Photos: http://wahoo.no-ip.org/piwigo paka @ IRCnet freenode -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos E. R.

13:12

On 11/06/2019 15.05, Patrick Shanahan wrote:

...

* Carlos E. R. <> [06-11-19 07:58]:

...
On 11/06/2019 13.37, Patrick Shanahan wrote:

...
* Carlos E. R. <> [06-11-19 02:32]:

...
On 11/06/2019 01.18, Patrick Shanahan wrote:

...
* Carlos E. R. <> [06-10-19 17:28]:

...
On 10/06/2019 19.53, Per Jessen wrote: > Carlos E. R. wrote:

...
...
...
...
> Your problem in Spain is also that there are no Spanish mirrors.

Right, but we use other central European servers, like gwdg, which are very fast.

then add the *specific* servers as I described earlier, and do not let mirrorbrained decide which server you use.

Which is what I said two paragraphs above (re bypass mirrorbrain).

but *are* you doing that?

Not when I reported. And doesn't work with wget. The problem is not with zypper, but wget.

you aren't directing wget to a *specific* site? ???

Nope. It worked while 15.1 was beta. It has been ages since I had to do that. -- Cheers / Saludos, Carlos E. R. (from openSUSE 15.0 (Legolas))

Patrick Shanahan

13:39

* Carlos E. R. [06-11-19 09:14]:

...

On 11/06/2019 15.05, Patrick Shanahan wrote:

...
* Carlos E. R. <> [06-11-19 07:58]:

...
On 11/06/2019 13.37, Patrick Shanahan wrote:

...
* Carlos E. R. <> [06-11-19 02:32]:

...
On 11/06/2019 01.18, Patrick Shanahan wrote:

...
* Carlos E. R. <> [06-10-19 17:28]: > On 10/06/2019 19.53, Per Jessen wrote: >> Carlos E. R. wrote:

...
...
...
>> Your problem in Spain is also that there are no Spanish mirrors. > > Right, but we use other central European servers, like gwdg, which are > very fast.

then add the *specific* servers as I described earlier, and do not let mirrorbrained decide which server you use.

Which is what I said two paragraphs above (re bypass mirrorbrain).

but *are* you doing that?

Not when I reported. And doesn't work with wget. The problem is not with zypper, but wget.

you aren't directing wget to a *specific* site? ???

Nope. It worked while 15.1 was beta. It has been ages since I had to do that.

and now you have problems that the ages brought about or have not solved. perhaps time to try and alternate solution. -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.org openSUSE Community Member facebook/ptilopteri Registered Linux User #207535 @ http://linuxcounter.net Photos: http://wahoo.no-ip.org/piwigo paka @ IRCnet freenode -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Anton Aylward

13:07

On 10/06/2019 19:18, Patrick Shanahan wrote:

...

then add the *specific* servers as I described earlier, and do not let mirrorbrained decide which server you use.

This makes sense when you are using one of a very few repositories.. As I've said before, doing the step-and-repeat to determine the most efficacious sites for a number of repository becomes a combinational explosion as you multiply it out. I've been referred to my 'local" (aka here in Canada or northern USA) sites, but the reality is that they only cover the 'core', not the kernel_Stable, the KDE5 or the additional photographic stuff. And a simple script based around ping/traceroute tells me nothing about how the server at the far end can actually deliver. One of those electrically distant sites in Russia actually gives a better throughput than ones here in Canada! "Well if you used a rolling distribution, Anton ..." "Well if you didn't use so many repositories, Anton ..." yes, yea, yea, "A minimis incipe". -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Per Jessen

13:41

Anton Aylward wrote:

...

On 10/06/2019 19:18, Patrick Shanahan wrote:

...
then add the *specific* servers as I described earlier, and do not let mirrorbrained decide which server you use.

This makes sense when you are using one of a very few repositories.. As I've said before, doing the step-and-repeat to determine the most efficacious sites for a number of repository becomes a combinational explosion as you multiply it out.

I've been referred to my 'local" (aka here in Canada or northern USA) sites, but the reality is that they only cover the 'core', not the kernel_Stable, the KDE5 or the additional photographic stuff.

If your "local" mirrors are okay for the core, and you get the extra stuff from elsewhere, problem solved? I thought the problem was that your "local" mirrors were also pretty poor, bandwidth wise.

...

And a simple script based around ping/traceroute tells me nothing about how the server at the far end can actually deliver.

Yes, run a wget on an ISO to get a useful impression. -- Per Jessen, Zürich (14.4°C) http://www.dns24.ch/ - your free DNS host, made in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Anton Aylward

13:56

On 11/06/2019 09:41, Per Jessen wrote:

...

If your "local" mirrors are okay for the core, and you get the extra stuff from elsewhere, problem solved? I thought the problem was that your "local" mirrors were also pretty poor, bandwidth wise.

No. I don't see many core updates to 42.3 these days. I do see updates to kernel_stable and the various other photographic stuff that I actively use that isn't in the core. Oh, and the KDF stuff! The more remote sites, the ones mirrorbrain sends me to for resolution of those matters, are the ones that deliver single and double digit download speeds.

...

...
And a simple script based around ping/traceroute tells me nothing about how the server at the far end can actually deliver.

Yes, run a wget on an ISO to get a useful impression.

So we are back to sheer combinatorial volume of doing that for every site that mirrorbrain _might_ direct me to for each repository I use. -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Patrick Shanahan

14:05

* Anton Aylward [06-11-19 09:58]:

...

On 11/06/2019 09:41, Per Jessen wrote:

...
If your "local" mirrors are okay for the core, and you get the extra stuff from elsewhere, problem solved? I thought the problem was that your "local" mirrors were also pretty poor, bandwidth wise.

No. I don't see many core updates to 42.3 these days. I do see updates to kernel_stable and the various other photographic stuff that I actively use that isn't in the core. Oh, and the KDF stuff! The more remote sites, the ones mirrorbrain sends me to for resolution of those matters, are the ones that deliver single and double digit download speeds.

...
...
And a simple script based around ping/traceroute tells me nothing about how the server at the far end can actually deliver.

Yes, run a wget on an ISO to get a useful impression.

So we are back to sheer combinatorial volume of doing that for every site that mirrorbrain _might_ direct me to for each repository I use.

or *just* add known *good* dl sites to your repos and not the opensuse url(s), *just* to repos where you continually experience problems. -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.org openSUSE Community Member facebook/ptilopteri Registered Linux User #207535 @ http://linuxcounter.net Photos: http://wahoo.no-ip.org/piwigo paka @ IRCnet freenode -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Per Jessen

14:52

Anton Aylward wrote:

...

On 11/06/2019 09:41, Per Jessen wrote:

...
If your "local" mirrors are okay for the core, and you get the extra stuff from elsewhere, problem solved? I thought the problem was that your "local" mirrors were also pretty poor, bandwidth wise.

No. I don't see many core updates to 42.3 these days. I do see updates to kernel_stable and the various other photographic stuff that I actively use that isn't in the core. Oh, and the KDF stuff! The more remote sites, the ones mirrorbrain sends me to for resolution of those matters, are the ones that deliver single and double digit download speeds.

Okay - well, then I don't think there is much you can do. Mirrorbrain sends you to those places because that is where those packages/repos are available. The only way to improve on that would be to convince a local mirror to also mirror repositories:/ -- Per Jessen, Zürich (15.1°C) -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Anton Aylward

17:47

I've edited zypp.conf and increased the number of connections from the default of 5 to 12. It does nothing to help the download of the metadata! That seems to remain in single or low double digit rates :-( I've also increased download.min_download_speed from the default of 0 to 385. Maybe that should be in the thousands? What is, bit/sec or bytes/sec? Multiple connections? Well having multiple connections to a single site, ftp.opensuse.org to whatever, is, perhaps not exactly brilliant. Well, it seems to have improved things: now into three digit byes/sec speeds consistently. Next big batch I'll increase the download.min_download_speed further. -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Anton Aylward

18:23

On 11/06/2019 13:47, Anton Aylward wrote:

...

I've edited zypp.conf and increased the number of connections from the default of 5 to 12. It does nothing to help the download of the metadata! That seems to remain in single or low double digit rates :-(

I've also increased download.min_download_speed from the default of 0 to 385. Maybe that should be in the thousands? What is, bit/sec or bytes/sec?

Multiple connections? Well having multiple connections to a single site, ftp.opensuse.org to whatever, is, perhaps not exactly brilliant.

Well, it seems to have improved things: now into three digit byes/sec speeds consistently. Next big batch I'll increase the download.min_download_speed further.

That was fine for a while, but now regressed before I could make any other changes. There's a lot of inconsistency out there! -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Per Jessen

18:37

Anton Aylward wrote:

...

I've edited zypp.conf and increased the number of connections from the default of 5 to 12. It does nothing to help the download of the metadata! That seems to remain in single or low double digit rates :-(

I've also increased download.min_download_speed from the default of 0 to 385. Maybe that should be in the thousands? What is, bit/sec or bytes/sec?

It says in the comments: "bytes per second". Nice find, I'll have to take note of that one.

...

Well, it seems to have improved things: now into three digit byes/sec speeds consistently.

...

From where, Anton?

-- Per Jessen, Zürich (15.1°C) http://www.dns24.ch/ - free dynamic DNS, made in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Anton Aylward

19:38

On 11/06/2019 14:37, Per Jessen wrote:

...

Anton Aylward wrote:

...
I've edited zypp.conf and increased the number of connections from the default of 5 to 12. It does nothing to help the download of the metadata! That seems to remain in single or low double digit rates :-(

I've also increased download.min_download_speed from the default of 0 to 385. Maybe that should be in the thousands? What is, bit/sec or bytes/sec?

It says in the comments: "bytes per second". Nice find, I'll have to take note of that one.

...
Well, it seems to have improved things: now into three digit byes/sec speeds consistently.

...
From where, Anton?

This is after kernel_default which came down at an astounding speed of 364.9 KiB/s) # netstat -t --notrim -p| grep zypper tcp 1 0 MainBox:36482 195.135.221.157:www-http CLOSE_WAIT 13716/zypper tcp 0 0 MainBox:36504 195.135.221.157:www-http ESTABLISHED 13716/zypper tcp 1 0 MainBox:52058 ftp.opensuse.org:www-http CLOSE_WAIT 13716/zypper tcp 1 0 MainBox:52082 ftp.opensuse.org:www-http CLOSE_WAIT 13716/zypper tcp 1 0 MainBox:36470 195.135.221.157:www-http CLOSE_WAIT 13716/zypper tcp 1 0 MainBox:52068 ftp.opensuse.org:www-http CLOSE_WAIT 13716/zypper tcp 1 0 MainBox:52074 ftp.opensuse.org:www-http CLOSE_WAIT 13716/zypper tcp 1 0 MainBox:52070 ftp.opensuse.org:www-http CLOSE_WAIT 13716/zypper tcp 1 0 MainBox:36492 195.135.221.157:www-http CLOSE_WAIT 13716/zypper tcp 1 0 MainBox:36496 195.135.221.157:www-http CLOSE_WAIT 13716/zypper -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Per Jessen

19:50

Anton Aylward wrote:

...

On 11/06/2019 14:37, Per Jessen wrote:

...
Anton Aylward wrote:

...
I've edited zypp.conf and increased the number of connections from the default of 5 to 12. It does nothing to help the download of the metadata! That seems to remain in single or low double digit rates :-(

I've also increased download.min_download_speed from the default of 0 to 385. Maybe that should be in the thousands? What is, bit/sec or bytes/sec?

It says in the comments: "bytes per second". Nice find, I'll have to take note of that one.

...
Well, it seems to have improved things: now into three digit byes/sec speeds consistently.

...
From where, Anton?

This is after kernel_default which came down at an astounding speed of 364.9 KiB/s)

# netstat -t --notrim -p| grep zypper tcp 1 0 MainBox:36482 195.135.221.157:www-http CLOSE_WAIT 13716/zypper

Which repository would have been better, but ftp.opensuse.org is in Nuernberg, 195.135.221.157 is in Provo, Utah. I would report to your provider that these are very slow. -- Per Jessen, Zürich (14.8°C) http://www.hostsuisse.com/ - virtual servers, made in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos E. R.

20:13

On 11/06/2019 19.47, Anton Aylward wrote:

...

I've edited zypp.conf and increased the number of connections from the default of 5 to 12. It does nothing to help the download of the metadata!

I understand the metadata is downloaded from the master copy, not the mirrors. -- Cheers / Saludos, Carlos E. R. (from 15.0 x86_64 at Telcontar)

Anton Aylward

12 Jun 12 Jun

14:30

On 11/06/2019 16:13, Carlos E. R. wrote:

...

On 11/06/2019 19.47, Anton Aylward wrote:

...
I've edited zypp.conf and increased the number of connections from the default of 5 to 12. It does nothing to help the download of the metadata!

I understand the metadata is downloaded from the master copy, not the mirrors.

Yea ${DEITY}s! ## Maximum time in seconds that you allow a transfer operation to take. ## ## This is useful for preventing your batch jobs from hanging for hours due ## to slow networks or links going down. Limiting operations to less than a ## few minutes risk aborting perfectly normal operations. ## ## Valid values: [0,3600] ## Default value: 180 ## # download.transfer_timeout = 180 Realistically, it's taking more than that! Next time I should run this using 'time' and mark with 'date' in the log THis too abut 45 minutes according to the logs Retrieving repository 'openSUSE-Leap-42.3-Update' metadata ---------------------------------------------[/] Timeout exceeded when accessing 'http://download.opensuse.org/update/leap/42.3/oss/repodata/beb927a2f16f0a199...'. Abort, retry, ignore? [a/r/i] (r): Trying again... Retrieving repository 'openSUSE-Leap-42.3-Update' metadata ---------------------------------------------[\] Timeout exceeded when accessing 'http://download.opensuse.org/update/leap/42.3/oss/repodata/beb927a2f16f0a199...'. Abort, retry, ignore? [a/r/i] (r): Trying again... Retrieving repository 'openSUSE-Leap-42.3-Update' metadata ---------------------------------------------[|] Timeout exceeded when accessing 'http://download.opensuse.org/update/leap/42.3/oss/repodata/beb927a2f16f0a199...'. Abort, retry, ignore? [a/r/i] (r): Trying again... Retrieving repository 'openSUSE-Leap-42.3-Update' metadata ---------------------------------------------[/] Timeout exceeded when accessing 'http://download.opensuse.org/update/leap/42.3/oss/repodata/beb927a2f16f0a199...'. Abort, retry, ignore? [a/r/i] (a): Trying again... Retrieving repository 'openSUSE-Leap-42.3-Update' metadata .........................................[error] Repository 'openSUSE-Leap-42.3-Update' is invalid. [repo-update|http://download.opensuse.org/update/leap/42.3/oss/] Valid metadata not found at specified URL Please check if the URIs defined for this repository are pointing to a valid repository. Warning: Skipping repository 'openSUSE-Leap-42.3-Update' because of the above error. Some of the repositories have not been refreshed because of an error. netstat -t --notrim -p| grep zypper tcp 1 0 MainBox:38414 ftp.opensuse.org:www-http CLOSE_WAIT 3888/zypper tcp 1 0 MainBox:37490 195.135.221.157:www-http CLOSE_WAIT 3888/zypper tcp 0 0 MainBox:39156 195.135.221.157:www-http ESTABLISHED 3888/zypper tcp 1 0 MainBox:37584 195.135.221.157:www-http CLOSE_WAIT 3888/zypper main:~ # date Wed Jun 12 09:22:16 EDT 2019 main:~ # time zypper up Retrieving repository 'openSUSE-Leap-42.3-Update' metadata ---------------------------------------------[|] Timeout exceeded when accessing 'http://download.opensuse.org/update/leap/42.3/oss/repodata/beb927a2f16f0a199...'. Abort, retry, ignore? [a/r/i] (r): Trying again... Retrieving repository 'openSUSE-Leap-42.3-Update' metadata ---------------------------------------------[/] Timeout exceeded when accessing 'http://download.opensuse.org/update/leap/42.3/oss/repodata/beb927a2f16f0a199...'. Abort, retry, ignore? [a/r/i] (r): Trying again... Retrieving repository 'openSUSE-Leap-42.3-Update' metadata ---------------------------------------------[\] Timeout exceeded when accessing 'http://download.opensuse.org/update/leap/42.3/oss/repodata/beb927a2f16f0a199...'. Abort, retry, ignore? [a/r/i] (r): Trying again... Retrieving repository 'openSUSE-Leap-42.3-Update' metadata ---------------------------------------------[-] Timeout exceeded when accessing 'http://download.opensuse.org/update/leap/42.3/oss/repodata/beb927a2f16f0a199...'. Abort, retry, ignore? [a/r/i] (a): Trying again... Retrieving repository 'openSUSE-Leap-42.3-Update' metadata .........................................[error] Repository 'openSUSE-Leap-42.3-Update' is invalid. [repo-update|http://download.opensuse.org/update/leap/42.3/oss/] Valid metadata not found at specified URL Please check if the URIs defined for this repository are pointing to a valid repository. Warning: Skipping repository 'openSUSE-Leap-42.3-Update' because of the above error. Some of the repositories have not been refreshed because of an error. Loading repository data... Reading installed packages... The following NEW package is going to be installed: kernel-default-5.1.9-5.1.ge68f829 1 new package to install. Overall download size: 65.2 MiB. Already cached: 0 B. After the operation, additional 351.3 MiB will be used. Continue? [y/n/...? shows all options] (y): n real 65m14.892s <================================================== user 0m8.372s sys 0m1.291s main:~ # -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Per Jessen

14:58

Anton Aylward wrote:

...

THis too abut 45 minutes according to the logs

Retrieving repository 'openSUSE-Leap-42.3-Update' metadata ---------------------------------------------[/] Timeout exceeded when accessing 'http://download.opensuse.org/update/leap/42.3/oss/repodata/beb927a2f16f0a199...'. Abort, retry, ignore? [a/r/i] (r):

That's not about a slow connection, that is _no_ connection. Would you mind trying the same url with wget? -- Per Jessen, Zürich (20.3°C) member, openSUSE Heroes -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos E. R.

17:46

On 12/06/2019 16.30, Anton Aylward wrote:

...

On 11/06/2019 16:13, Carlos E. R. wrote:

...
On 11/06/2019 19.47, Anton Aylward wrote:

...
I've edited zypp.conf and increased the number of connections from the default of 5 to 12. It does nothing to help the download of the metadata!

I understand the metadata is downloaded from the master copy, not the mirrors.

Yea ${DEITY}s!

If you define a mirror instead of the master, then all comes from the mirror.

...

## Maximum time in seconds that you allow a transfer operation to take. ## ## This is useful for preventing your batch jobs from hanging for hours due ## to slow networks or links going down. Limiting operations to less than a ## few minutes risk aborting perfectly normal operations. ## ## Valid values: [0,3600] ## Default value: 180 ## # download.transfer_timeout = 180

Realistically, it's taking more than that! Next time I should run this using 'time' and mark with 'date' in the log

THis too abut 45 minutes according to the logs

Retrieving repository 'openSUSE-Leap-42.3-Update' metadata ---------------------------------------------[/] Timeout exceeded when accessing 'http://download.opensuse.org/update/leap/42.3/oss/repodata/beb927a2f16f0a199...'. Abort, retry, ignore? [a/r/i] (r): Trying again...

But you are manually telling it to start again, that resets the timer. -- Cheers / Saludos, Carlos E. R. (from 15.0 x86_64 at Telcontar)

Per Jessen

11 Jun 11 Jun

07:20

Carlos E. R. wrote:

...

On 10/06/2019 19.53, Per Jessen wrote:

...
Carlos E. R. wrote:

...
This mirror is too slow:

--2019-06-10 19:17:10-- [snip]

Whom do I report it to, admin@o.o?

It's worth a try, it'll go to the mirror admin - that's me :-)

There isn't much we can do about a mirror being "slow". Anton Aylward has brought up the topic too. Most likely it is not the mirror that is slow, but the end-to-end connection.

300 Kb/s when my connection is 100 Mb/s is very slow...

Like I said, you have to consider the end-to-end connection, looking at either end is of little use. I am certain the University of Lissabon has a very decent uplink, and yours is also fine. Try doing a traceroute, that might give you an indication of the quality of the interconnection between your provider and Uni Lissabon. From my location, I see about 10 hops (ipv4), but the first goes via Milan. The IPv6 route is virtually identical. Right now I am downloading an ISO from ulisboa.pt - the speed goes up and down quite a bit, but it seems to mostly run between 1 and 2Mb/s. A concurrent download from Nuernberg is running at 25-26Mb/s. A 3rd concurrent download from a Hetzner datacentre is racing along at 40Mb/s. 4Gb ISO download from Uni of Lissabon - Hetzner @ Falkenstein - 89seconds, 43.3Mb/s. openSUSE, Nuernberg - 3m35, 18Mb/s. Hostsuisse, Switzerland - avg speed 2.34Mb/s. I think we can conclude (with some guessing) - Hetzner has a very direct interconnect to Portugal (it does not go via Milan, for instance). openSUSE in Nuernberg isn't so bad, but does go via Frankfurt. For Portugal, Hostsuisse is pretty poor. The mirror at Uni of Lisboa is not slow. It can easily feed 50-60Mb/sec. I suspect much more locally. It might be worth mentioning that measuring with wget (as I also did) only gives an indication of one mirror - whereas zypper will use chunked downloading from multiple mirrors, which should give you an overall much better performance. Feel free to report slow mirrors, but there is very little we (openSUSE) can do. -- Per Jessen, Zürich (13.8°C) member, openSUSE Heroes -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos E. R.

08:01

On 11/06/2019 09.20, Per Jessen wrote:

...

Carlos E. R. wrote:

...
On 10/06/2019 19.53, Per Jessen wrote:

...
Carlos E. R. wrote:

...
This mirror is too slow:

--2019-06-10 19:17:10-- [snip]

Whom do I report it to, admin@o.o?

It's worth a try, it'll go to the mirror admin - that's me :-)

There isn't much we can do about a mirror being "slow". Anton Aylward has brought up the topic too. Most likely it is not the mirror that is slow, but the end-to-end connection.

300 Kb/s when my connection is 100 Mb/s is very slow...

Like I said, you have to consider the end-to-end connection, looking at either end is of little use. I am certain the University of Lissabon has a very decent uplink, and yours is also fine.

Try doing a traceroute, that might give you an indication of the quality of the interconnection between your provider and Uni Lissabon. From my location, I see about 10 hops (ipv4), but the first goes via Milan. The IPv6 route is virtually identical.

20 hops. cer@Telcontar:~> traceroute ftp.rnl.tecnico.ulisboa.pt traceroute to ftp.rnl.tecnico.ulisboa.pt (193.136.164.6), 30 hops max, 60 byte packets 1 router.valinor (192.168.1.1) 0.602 ms 0.811 ms 0.794 ms 2 * * * 3 * * * 4 134.red-81-46-33.customer.static.ccgg.telefonica.net (81.46.33.134) 14.459 ms * 142.red-81-46-33.customer.static.ccgg.telefonica.net (81.46.33.142) 14.428 ms 5 * * 97.red-80-58-106.staticip.rima-tde.net (80.58.106.97) 15.918 ms 6 216.184.113.52 (216.184.113.52) 15.200 ms 15.251 ms ae1-400-gramadte2.net.telefonicaglobalsolutions.com (216.184.113.186) 16.058 ms 7 94.142.97.139 (94.142.97.139) 16.287 ms 94.142.107.95 (94.142.107.95) 12.818 ms 11.365 ms 8 prs-bb3-link.telia.net (213.155.131.152) 51.013 ms prs-bb4-link.telia.net (80.91.248.130) 49.946 ms 52.344 ms 9 ffm-bb3-link.telia.net (62.115.123.12) 53.266 ms prs-bb4-link.telia.net (80.91.248.130) 51.260 ms ffm-bb3-link.telia.net (62.115.123.12) 52.687 ms 10 ffm-bb3-link.telia.net (62.115.123.12) 54.366 ms ffm-b12-link.telia.net (62.115.142.5) 49.979 ms ffm-b12-link.telia.net (62.115.142.47) 54.245 ms 11 geant-ic-326862-ffm-b12.c.telia.net (80.239.135.137) 49.255 ms 49.223 ms 52.678 ms 12 * * * 13 * * * 14 ae0.mx2.lis.pt.geant.net (62.40.98.106) 96.756 ms * 97.454 ms 15 fccn-ias-fccn-gw.mx2.lis.pt.geant.net (83.97.88.210) 92.432 ms ae0.mx2.lis.pt.geant.net (62.40.98.106) 96.191 ms 96.528 ms 16 ROUTER60.10GE.CR2.Lisboa.fccn.pt (193.137.0.28) 95.772 ms ROUTER60.10GE.CR1.Lisboa.fccn.pt (193.137.0.14) 92.029 ms fccn-ias-fccn-gw.mx2.lis.pt.geant.net (83.97.88.210) 95.037 ms 17 ROUTER60.10GE.CR2.Lisboa.fccn.pt (193.137.0.28) 96.082 ms ROUTER60.10GE.CR1.Lisboa.fccn.pt (193.137.0.14) 96.070 ms gtulisboa-ist.fccn.pt (193.136.1.94) 97.281 ms 18 gtulisboa-ist.fccn.pt (193.136.1.94) 98.114 ms 194.117.12.134 (194.117.12.134) 92.686 ms 97.227 ms 19 b0-2.gwcore2.tecnico.ulisboa.pt (193.136.134.143) 98.790 ms 95.169 ms 98.341 ms 20 b0-2.gwcore2.tecnico.ulisboa.pt (193.136.134.143) 93.150 ms * *

...

Right now I am downloading an ISO from ulisboa.pt - the speed goes up and down quite a bit, but it seems to mostly run between 1 and 2Mb/s. A concurrent download from Nuernberg is running at 25-26Mb/s. A 3rd concurrent download from a Hetzner datacentre is racing along at 40Mb/s.

4Gb ISO download from Uni of Lissabon -

Hetzner @ Falkenstein - 89seconds, 43.3Mb/s. openSUSE, Nuernberg - 3m35, 18Mb/s. Hostsuisse, Switzerland - avg speed 2.34Mb/s.

I think we can conclude (with some guessing) -

Hetzner has a very direct interconnect to Portugal (it does not go via Milan, for instance). openSUSE in Nuernberg isn't so bad, but does go via Frankfurt. For Portugal, Hostsuisse is pretty poor.

The mirror at Uni of Lisboa is not slow. It can easily feed 50-60Mb/sec. I suspect much more locally.

But not to me...

...

It might be worth mentioning that measuring with wget (as I also did) only gives an indication of one mirror - whereas zypper will use chunked downloading from multiple mirrors, which should give you an overall much better performance.

Yes, but I had to use wget to populate a partial local mirror. Maybe if I could use aria2c instead that would make a difference, but I don't know how.

...

Feel free to report slow mirrors, but there is very little we (openSUSE) can do.

Not offer them from the Brain. As we can not configure a local file to disable them... -- Cheers / Saludos, Carlos E. R. (from 15.0 x86_64 at Telcontar)

Patrick Shanahan

11:41

* Carlos E. R. [06-11-19 04:04]:

...

On 11/06/2019 09.20, Per Jessen wrote:

...
Carlos E. R. wrote:

...
On 10/06/2019 19.53, Per Jessen wrote:

...
Carlos E. R. wrote:

...
This mirror is too slow:

--2019-06-10 19:17:10-- [snip]

Whom do I report it to, admin@o.o?

It's worth a try, it'll go to the mirror admin - that's me :-)

There isn't much we can do about a mirror being "slow". Anton Aylward has brought up the topic too. Most likely it is not the mirror that is slow, but the end-to-end connection.

300 Kb/s when my connection is 100 Mb/s is very slow...

Like I said, you have to consider the end-to-end connection, looking at either end is of little use. I am certain the University of Lissabon has a very decent uplink, and yours is also fine.

Try doing a traceroute, that might give you an indication of the quality of the interconnection between your provider and Uni Lissabon. From my location, I see about 10 hops (ipv4), but the first goes via Milan. The IPv6 route is virtually identical.

20 hops.

cer@Telcontar:~> traceroute ftp.rnl.tecnico.ulisboa.pt traceroute to ftp.rnl.tecnico.ulisboa.pt (193.136.164.6), 30 hops max, 60 byte packets 1 router.valinor (192.168.1.1) 0.602 ms 0.811 ms 0.794 ms 2 * * * 3 * * * 4 134.red-81-46-33.customer.static.ccgg.telefonica.net (81.46.33.134) 14.459 ms * 142.red-81-46-33.customer.static.ccgg.telefonica.net (81.46.33.142) 14.428 ms 5 * * 97.red-80-58-106.staticip.rima-tde.net (80.58.106.97) 15.918 ms 6 216.184.113.52 (216.184.113.52) 15.200 ms 15.251 ms ae1-400-gramadte2.net.telefonicaglobalsolutions.com (216.184.113.186) 16.058 ms 7 94.142.97.139 (94.142.97.139) 16.287 ms 94.142.107.95 (94.142.107.95) 12.818 ms 11.365 ms 8 prs-bb3-link.telia.net (213.155.131.152) 51.013 ms prs-bb4-link.telia.net (80.91.248.130) 49.946 ms 52.344 ms 9 ffm-bb3-link.telia.net (62.115.123.12) 53.266 ms prs-bb4-link.telia.net (80.91.248.130) 51.260 ms ffm-bb3-link.telia.net (62.115.123.12) 52.687 ms 10 ffm-bb3-link.telia.net (62.115.123.12) 54.366 ms ffm-b12-link.telia.net (62.115.142.5) 49.979 ms ffm-b12-link.telia.net (62.115.142.47) 54.245 ms 11 geant-ic-326862-ffm-b12.c.telia.net (80.239.135.137) 49.255 ms 49.223 ms 52.678 ms 12 * * * 13 * * * 14 ae0.mx2.lis.pt.geant.net (62.40.98.106) 96.756 ms * 97.454 ms 15 fccn-ias-fccn-gw.mx2.lis.pt.geant.net (83.97.88.210) 92.432 ms ae0.mx2.lis.pt.geant.net (62.40.98.106) 96.191 ms 96.528 ms 16 ROUTER60.10GE.CR2.Lisboa.fccn.pt (193.137.0.28) 95.772 ms ROUTER60.10GE.CR1.Lisboa.fccn.pt (193.137.0.14) 92.029 ms fccn-ias-fccn-gw.mx2.lis.pt.geant.net (83.97.88.210) 95.037 ms 17 ROUTER60.10GE.CR2.Lisboa.fccn.pt (193.137.0.28) 96.082 ms ROUTER60.10GE.CR1.Lisboa.fccn.pt (193.137.0.14) 96.070 ms gtulisboa-ist.fccn.pt (193.136.1.94) 97.281 ms 18 gtulisboa-ist.fccn.pt (193.136.1.94) 98.114 ms 194.117.12.134 (194.117.12.134) 92.686 ms 97.227 ms 19 b0-2.gwcore2.tecnico.ulisboa.pt (193.136.134.143) 98.790 ms 95.169 ms 98.341 ms 20 b0-2.gwcore2.tecnico.ulisboa.pt (193.136.134.143) 93.150 ms * *

...
Right now I am downloading an ISO from ulisboa.pt - the speed goes up and down quite a bit, but it seems to mostly run between 1 and 2Mb/s. A concurrent download from Nuernberg is running at 25-26Mb/s. A 3rd concurrent download from a Hetzner datacentre is racing along at 40Mb/s.

4Gb ISO download from Uni of Lissabon -

Hetzner @ Falkenstein - 89seconds, 43.3Mb/s. openSUSE, Nuernberg - 3m35, 18Mb/s. Hostsuisse, Switzerland - avg speed 2.34Mb/s.

I think we can conclude (with some guessing) -

Hetzner has a very direct interconnect to Portugal (it does not go via Milan, for instance). openSUSE in Nuernberg isn't so bad, but does go via Frankfurt. For Portugal, Hostsuisse is pretty poor.

The mirror at Uni of Lisboa is not slow. It can easily feed 50-60Mb/sec. I suspect much more locally.

But not to me...

...
It might be worth mentioning that measuring with wget (as I also did) only gives an indication of one mirror - whereas zypper will use chunked downloading from multiple mirrors, which should give you an overall much better performance.

Yes, but I had to use wget to populate a partial local mirror. Maybe if I could use aria2c instead that would make a difference, but I don't know how.

...
Feel free to report slow mirrors, but there is very little we (openSUSE) can do.

Not offer them from the Brain.

As we can not configure a local file to disable them...

*again* sure you can, but not directly. SPECIFY specific sites rather than allowing mirrorbrain to pick them. -- (paka)Patrick Shanahan Plainfield, Indiana, USA @ptilopteri http://en.opensuse.org openSUSE Community Member facebook/ptilopteri Registered Linux User #207535 @ http://linuxcounter.net Photos: http://wahoo.no-ip.org/piwigo paka @ IRCnet freenode -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Per Jessen

09:27

Per Jessen wrote:

...

The mirror at Uni of Lisboa is not slow. It can easily feed 50-60Mb/sec. I suspect much more locally.

[snip]

...

Feel free to report slow mirrors, but there is very little we (openSUSE) can do.

A poor-end-to-end connection is often due to a poor interconnection between one provider and another, in Carlos' case between Telefonica and the University of Lisboa. (geant.org, fccn.pt). I have over the years occasionally had such issues myself (in my business) and have reported them to my uplink provider. AFAIR, they were always solved by establishing different peerings. This is however up to the providers and their uplink providers. So a better answer to $SUBJ would be: - if a mirror is slow, report to your access provider. - if a mirror is down/inactive/outdated, report to admin@o.o. -- Per Jessen, Zürich (13.8°C) member, openSUSE Heroes. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos E. R.

12:04

On 11/06/2019 11.27, Per Jessen wrote:

...

Per Jessen wrote:

...
The mirror at Uni of Lisboa is not slow. It can easily feed 50-60Mb/sec. I suspect much more locally.

[snip]

...
Feel free to report slow mirrors, but there is very little we (openSUSE) can do.

A poor-end-to-end connection is often due to a poor interconnection between one provider and another, in Carlos' case between Telefonica and the University of Lisboa. (geant.org, fccn.pt). I have over the years occasionally had such issues myself (in my business) and have reported them to my uplink provider. AFAIR, they were always solved by establishing different peerings. This is however up to the providers and their uplink providers.

So a better answer to $SUBJ would be:

- if a mirror is slow, report to your access provider. - if a mirror is down/inactive/outdated, report to admin@o.o.

I'll try that. Please, I will need sample test outputs from other sites like you did to add to the report, so that I can tell them that the speed is fast from other countries. I just tried tethering my laptop to my mobile phone (via wifi), and speed went up to 1.3..2.0 Mb -- Cheers / Saludos, Carlos E. R. (from openSUSE 15.0 (Legolas))

Per Jessen

12:24

Carlos E. R. wrote:

...

On 11/06/2019 11.27, Per Jessen wrote:

...
Per Jessen wrote:

...
The mirror at Uni of Lisboa is not slow. It can easily feed 50-60Mb/sec. I suspect much more locally.

[snip]

...
Feel free to report slow mirrors, but there is very little we (openSUSE) can do.

A poor-end-to-end connection is often due to a poor interconnection between one provider and another, in Carlos' case between Telefonica and the University of Lisboa. (geant.org, fccn.pt). I have over the years occasionally had such issues myself (in my business) and have reported them to my uplink provider. AFAIR, they were always solved by establishing different peerings. This is however up to the providers and their uplink providers.

So a better answer to $SUBJ would be:

- if a mirror is slow, report to your access provider. - if a mirror is down/inactive/outdated, report to admin@o.o.

I'll try that. Please, I will need sample test outputs from other sites like you did to add to the report, so that I can tell them that the speed is fast from other countries.

I'll send them off-line. -- Per Jessen, Zürich (14.5°C) http://www.dns24.ch/ - free dynamic DNS, made in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos E. R.

12:42

On 11/06/2019 14.24, Per Jessen wrote:

...

Carlos E. R. wrote:

...

...
...
So a better answer to $SUBJ would be:

- if a mirror is slow, report to your access provider. - if a mirror is down/inactive/outdated, report to admin@o.o.

I'll try that. Please, I will need sample test outputs from other sites like you did to add to the report, so that I can tell them that the speed is fast from other countries.

I'll send them off-line.

Thanks :-) I'll open the ticket when I get back home. -- Cheers / Saludos, Carlos E. R. (from openSUSE 15.0 (Legolas))

Anton Aylward

13:46

On 11/06/2019 03:20, Per Jessen wrote:

...

I think we can conclude (with some guessing) -

Granted. But ... It used to be that time of day mattered. I could CRON a download for the early hours local to wherever and expect that to be low load, but now the internet has grown and runs 36/9 you can't predict time in advance as opposed to statistical fluctuations. And to get ruond those you need more sampling and better predictive algorithms, not simply a larger, faster pipe to your own site. I recall back in the hinterland of the 'net, the early 1990s, drawing an analogy with flushing toilets. They generally have 4" out-pipes to the common drain and that to the main sewer, which probably also deals with rain run-off from the street gutters. But the common drain running past the houses is only 6". that's fine, statistically speaking, but if all 46 houses on the street decide to flush at the same time ... and perhaps while there is a storm, that is moving in from the high ground pushing a flood of water ahead into the drainage system ... Re call the old saw about drowning in a river of average depth 4 inches? Yes, statistically speaking, traceroute might tell you stuff about getting to that server in Hong Kong from Toronto is a fast hop to a router in New York then a fantastic pipe that goes past the Horn of Africa: just 4 hops, TO to HK! bB comparison the route to the server in Kosmograd is 9 hops away and 'ping' says it is four times the delay. But what's the load on the server in HK and the server in Kosmograd? What's the packet loss rate, not on the route, but discarded at the end due to lack of buffer space? There's a lot going on out there that isn't under your control and which you only have indirect evidence for. of course if you've had to debug network/server problems or had to deal with overloaded systems or inherited a badly configured system and had to figure out how to 'optimise' it .... that all helps with the speculation. But closer (aka quicker ping, fewer hops) might not mean better performance, better throughput. I can set up my old Newton 2100 touchpad as a HTTPS server and make it internet accessible (and I did once just to prove I could), but it doesn't have much capacity, either in terms of storage or computational power. But to people who are local to me, such as James Knott, it would "look" like a great repository since it is just a couple of hops away! That it was always empty had little to do with matters, eh? And so to, as far as I'm concerned when I'm updating kernel, photo-stuff and the like, the local repositories here in Canada are useless to me. Mirrorbrain sends me to Asia. A more populous part of the world. -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Per Jessen

15:00

Anton Aylward wrote:

...

On 11/06/2019 03:20, Per Jessen wrote:

...
I think we can conclude (with some guessing) -

Granted. But ...

[snip]

...

Yes, statistically speaking, traceroute might tell you stuff about getting to that server in Hong Kong from Toronto is a fast hop to a router in New York then a fantastic pipe that goes past the Horn of Africa: just 4 hops, TO to HK! bB comparison the route to the server in Kosmograd is 9 hops away and 'ping' says it is four times the delay. But what's the load on the server in HK and the server in Kosmograd? What's the packet loss rate, not on the route, but discarded at the end due to lack of buffer space?

For this sort of thing, I use traceroute to actually show me the route. I'm certainly no expert on internet peering setups, but when I see an odd route, like my own connection to Uni Lisboa going via some location in Milan, I suspect poor peering. No doubt much smarter people than me will point to better tools to investigate (looking glass, https://www.peeringdb.com/ maybe) -- Per Jessen, Zürich (15.2°C) http://www.cloudsuisse.com/ - your owncloud, hosted in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Anton Aylward

13:13

On 10/06/2019 17:25, Carlos E. R. wrote:

...

300 Kb/s when my connection is 100 Mb/s is very slow...

Oh, WOW, all off 300kb/s! Three digits worth! Try loading a kernel update at single digit speeds.

...

I remember that thread, but couldn't find it. Yes, we would need a local blacklist of mirrors. Otherwise I'll have to bypass mirrorbrain and list manually some mirrors.

I prefer the blacklist idea. Manually determining the groups of possible servers for each repository and then timing them and editing the dot-repo files will get time consuming. -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos E. R.

13:26

On 11/06/2019 15.13, Anton Aylward wrote:

...

On 10/06/2019 17:25, Carlos E. R. wrote:

...
300 Kb/s when my connection is 100 Mb/s is very slow...

Oh, WOW, all off 300kb/s! Three digits worth! Try loading a kernel update at single digit speeds.

I did, years ago. Wait, 3 digits? It is less than one: 0.3 mbits/s.

...

...
I remember that thread, but couldn't find it. Yes, we would need a local blacklist of mirrors. Otherwise I'll have to bypass mirrorbrain and list manually some mirrors.

I prefer the blacklist idea. Manually determining the groups of possible servers for each repository and then timing them and editing the dot-repo files will get time consuming.

At least here zypper downloads concurrently from several repos, and the rest come from Germany or thereabouts and run normally. Luckily only one "local" mirror. Sigh. But for directly downloading, I don't see a solution. One has first to find the official URL, then the different URL at a mirror, and mirrors do not always work fine, they change. -- Cheers / Saludos, Carlos E. R. (from openSUSE 15.0 (Legolas))

Carlos E. R.

20:15

On 11/06/2019 15.26, Carlos E. R. wrote:

...

On 11/06/2019 15.13, Anton Aylward wrote:

...
On 10/06/2019 17:25, Carlos E. R. wrote:

...
300 Kb/s when my connection is 100 Mb/s is very slow...

Oh, WOW, all off 300kb/s! Three digits worth! Try loading a kernel update at single digit speeds.

I did, years ago. Wait, 3 digits? It is less than one: 0.3 mbits/s.

The 15.1 iso took 2h 54m to download from Lisboa to my house. 3.78G 508KB/s in 2h 54m -- Cheers / Saludos, Carlos E. R. (from 15.0 x86_64 at Telcontar)

Per Jessen

14:05

Anton Aylward wrote:

...

On 10/06/2019 17:25, Carlos E. R. wrote:

...
300 Kb/s when my connection is 100 Mb/s is very slow...

Oh, WOW, all off 300kb/s! Three digits worth! Try loading a kernel update at single digit speeds.

...
I remember that thread, but couldn't find it. Yes, we would need a local blacklist of mirrors. Otherwise I'll have to bypass mirrorbrain and list manually some mirrors.

I prefer the blacklist idea. Manually determining the groups of possible servers for each repository and then timing them and editing the dot-repo files will get time consuming.

FWIW, I have just timed the 15.1 DVD iso from Hetzner (Germany) to the mirror at Dalhousie (20Mb/s), freemirror.org (25Mb/s), Manitoba (very slow) and Uwaterloo (too slow, hovering at 1Mb/s). If I get a better service in Germany than you do in Canada, something is wrong with your interconnect. As I suggested to Carlos, try asking your provider. All four Canadian mirrors showed very poor service from my office location. Entirely expected. -- Per Jessen, Zürich (14.2°C) http://www.dns24.ch/ - free dynamic DNS, made in Switzerland. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Knurpht-openSUSE

22:00

Op dinsdag 11 juni 2019 15:13:16 CEST schreef Anton Aylward:

...

On 10/06/2019 17:25, Carlos E. R. wrote:

...
300 Kb/s when my connection is 100 Mb/s is very slow...

Oh, WOW, all off 300kb/s! Three digits worth! Try loading a kernel update at single digit speeds.

...
I remember that thread, but couldn't find it. Yes, we would need a local blacklist of mirrors. Otherwise I'll have to bypass mirrorbrain and list manually some mirrors.

I prefer the blacklist idea. Manually determining the groups of possible servers for each repository and then timing them and editing the dot-repo files will get time consuming. > Q: Are you sure? > >> A: Because it reverses the logical flow of conversation. >> >>> Q: Why is top posting frowned upon? sudo dracut -f

-- Gertjan Lettink a.k.a. Knurpht openSUSE Board Member openSUSE Forums Team -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org

Carlos E. R.

12 Jun 12 Jun

01:00

On 12/06/2019 00.00, Knurpht-openSUSE wrote:

...

Op dinsdag 11 juni 2019 15:13:16 CEST schreef Anton Aylward:

...
On 10/06/2019 17:25, Carlos E. R. wrote:

...
300 Kb/s when my connection is 100 Mb/s is very slow...

Oh, WOW, all off 300kb/s! Three digits worth! Try loading a kernel update at single digit speeds.

...
I remember that thread, but couldn't find it. Yes, we would need a local blacklist of mirrors. Otherwise I'll have to bypass mirrorbrain and list manually some mirrors.

I prefer the blacklist idea. Manually determining the groups of possible servers for each repository and then timing them and editing the dot-repo files will get time consuming. > Q: Are you sure? > >> A: Because it reverses the logical flow of conversation. >> >>> Q: Why is top posting frowned upon? sudo dracut -f

I suspect you wanted to answer on another thread ;-) -- Cheers / Saludos, Carlos E. R. (from 15.0 x86_64 at Telcontar)

1787

Age (days ago)

1789

Last active (days ago)

List overview

Download

38 comments

5 participants

participants (5)

Anton Aylward
Carlos E. R.
Knurpht-openSUSE
Patrick Shanahan
Per Jessen

[opensuse] Where to report bad mirrors?

tags

participants (5)