Re: [opensuse] Update server (download.opensuse.org) slow or crowded like mad - again
I asked the list, why do you reply off list? On Wed, Apr 26, 2017 at 5:33 PM, Richard Brown <RBrownCCB@opensuse.org> wrote:
On 26 April 2017 at 17:21, cagsm <cumandgets0mem00f@gmail.com> wrote:
Once every few days or weeks, we see nonpropagating files especially in the updates directories that somehow only reside on the main primary source for downloads.
<http://download.opensuse.org/update/leap/42.2/oss/repodata/?C=M;O=D>
only lives on the one and only primary source, no mirrors fetched this and other files, although some others in this directory from todays date and current timestamp propagated just fine.
download.opensuse.org or mirrorbrain or the whole deal seems to be often crowded or crashed or not very robust.
Is this gonna improve somewhen in the foreseeable future? In the near future? In the past, I have seen lot of bugreports or bickering on lists or forums on these kind of issues with the download server.
Thank you.
The delta metadata for an update published an hour ago hasn't propagated to other mirrors yet..and you're complaining?
Seriously, how do you expect us to mirror our data? Get a time machine and give the packages to our mirror hosts a day before we make the update?
[Please stop messaging me in private, the list is supposed to know this.TY.] it is not about some file not propagating, it is about the whole update mechanism coming to halt because the metadata blocks the whole zypper ref bytes/sec or even bits/sec trickling in and seemingly blocking and never completing the task. also the binaries of much larger stuff did propagate to mirrors. this update process is really bad if there cant be atomicity of some simple index file or softlink or similar which switches over the outdated metadata to the current metadata once all the mirrors have fetched it and meanwhile not hogging down and stalling everones update experience i am really annoyed by these eternal waits on zypper ref if it could still work and use the stuff already mirrored and already propagated in the mean time. also often there were reports on the infrastructure when the mirrorbrain and responsible scripts hosts or services were crashed or faulty that they needed manual intervention or versight and people only here reported them to the list or even the bug tracker before anyone took care of it. you never know what or if these things need attention or public display on this list or elsewhere before people might need to do something about them or if its a normal slow state until the mirrors fetch even these metadata so if you know of a better check or if anyone has some state display of these opensuse infrastructure if everything is considered to be okay or not, until then i might come forward and ask about them here on the list if i come across such situations. thank you. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
participants (1)
-
cagsm