Hello. Does it look like valid traceroute
for http://clojars.org?
The problem is that all my lein deps
, lein uprgade
or even e.g. lein new figwheel ...
commands basically hang
@andrewboltachev: that trace looks very similar to what I get (stars past http://core7-aggr403b-9.iad3.rackspace.net), but lein deps
works fine
I don't think lein upgrade
hits clojars, so that might be a clue that something else is going wrong on your end
I mean, when you download new stuff, ofc
The thing is it's on both my local machine (Kirov, Russia) and server (Amsterdam, The Netherlands)
right, I removed deps from my ~/.m2 and ran lein deps
hmm, can you try the new CDN-fronted <http://repo.clojars.org|repo.clojars.org>
and see if that behaves the same?
you'll need to add the following to ~/.lein/profiles.clj
:
:repositories [["clojars" {:url "<https://repo.clojars.org>"}]]
:plugin-repositories [["clojars" {:url "<https://repo.clojars.org>"}]]
in your :user
profile
you'll get a warning from lein about having repos in the :user
profile, but you can ignore that for now
I haven't had this file btw:
{:user {
:repositories [["clojars" {:url "<https://repo.clojars.org>"}]]
:plugin-repositories [["clojars" {:url "<https://repo.clojars.org>"}]]
}}
should this be valid?
Well, with this, I have:
yes, that should be correct
18 sec for 14.6M sounds quite serious
that doesn't use clojars though: https://github.com/technomancy/leiningen/blob/master/bin/lein#L73
so adding the CDN repo should have made no difference
yep, I've had GitHub also working with delays for me
Okay, either I have VPN or not, my physical connection stays the same
So that might be it's issues
thanks @tcrawley and I would check everything more carefully. This CDN should be helpful btw
glad to help
is it feasible to run private clojars install now?
wondering if I could swap archiva with clojars-web
@mpenet: possible? yes. feasible? probably not. It would be a lot of work to set up
we've been working on an ansible setup for it that would ease that, but there are still lots of clojars.org-specific bits in there
alright, anything in particular (I might be able to help with some of it maybe)?
@mpenet: we haven't given specific thought to where we need to make changes to support private instances, we've been focused first on getting a reproducible setup, so don't have a list of what needs to happen. You are welcome to take a look at what you have to see what would need to change if you like: https://github.com/clojars/clojars-server-config
one of the big missing piece might be auth on the artifacts. if I am not mistaken now they are served via nginx directly
then maybe it makes sense to map auth on these to user/groups credentials (thinking out loud)
also another question would be do we want to allow auth per artifact or is it an on/off switch that applies to the whole instance
serving the artifacts via nginx is a good thing, that's where all the pressure is and it's the perfect solution doing the heavy duty work, I don't know if it can be reconfigured "dynamically" for the auth part
maybe with lua + redis (for instance). But yeah, I know, it's another dependency ...
cool stuff: http://nginx.org/en/docs/http/ngx_http_auth_request_module.html
anyway, I ll stop flooding for now and look into this a bit on my free time 😆