-
Notifications
You must be signed in to change notification settings - Fork 47
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
torquebox.org is DOWN #124
Comments
As a workaround, is the old IP still available and known? |
Current tests should report the issue https://github.com/torquebox/jruby-maven-plugins/blob/master/gem-maven-plugin/src/it/include-rubygems-in-test-resources/pom.xml. In the past, we used this #88 (comment) but it's also failing now. Any suggestions or alternative? |
@abelsromero what exactly is the problem with using the maven gem plugin? JRuby currently uses that plugin as part of our regular build. |
thanks!
|
Thank you for the information. I was able to reproduce by simply adding this gem to the JRuby build, alongside the few dozen other gems that resolve correctly. The question now is why this gem fails! |
I'm not sure what it is but both asciidoctor and asciidoctor-multipage are not working while other gems I try do seem to work (and of course all those dozens we fetch during JRuby's build). The gemspecs for those gems do contain more fields. I'm not seeing anything I would expect to break in this way, though. The v1 and v2 API calls for rubygems.org also produce json for the gem that doesn't look unusual. It might help if we knew how the gems were being built and pushed; perhaps it is something specific to the rubygems version or Ruby version being used? cc @mojavelinux |
I seriously doubt it. The asciidoctor gem is built using As far as I know, it's a typical gem. Nothing special about it. And there's never been a problem retrieving the asciidoctor gem. |
@mojavelinux Thanks... yeah it was a long shot. I'm going to have to hook up a debugger to figure out why some gems are failing to download via mavengem. |
Ok so here's a new wrinkle: everything just worked on my most recent build. I'm wondering if there's something happening on the RG.org side to cause these issues. |
So after my environment started working, I thought I'd try your suggestion. And it appeared to build ok. I'm now trying to do a build with the rubygems maven artifacts wiped out (to force re-fetching) just to make sure. I don't want to theorize yet, but all I changed was Java version and wiping out the rubygems maven artifacts. |
I already did, since Maven leaves a metadata file to avoid extra fetchings, it's a must usually.
Me too, I just got it working again on MacOS and Linux, both Java 17 (Linux is where I did the previous testing in different days).
@headius So my understanding now then is that when using gem-maven-plugin to download a gem, no other connection is done right? The plugin only interacts with https://rubygems.org/ and nothing else? |
@abelsromero Yes, I believe that is correct (only rubygems.org). The intermittent quality of this makes me wonder if we are seeing some sort of rate-limiting on RG.org, or possibly hitting some bad instance(s) in a cluster that reject some gem requests. I will do a bit more research today, but then I will have to move on to something else until we can get a reliable way to reproduce this. |
I may have found a rather stupid and frustrating cause for these intermittent failures. This In fact, while poking around the mavengem implementation, I also tried some URLs like the v1 dependencies search. Hopefully this isn't confirmation bias, but it appear to fail for a while and then suddenly start working. More worryingly, this appears to be the trailing edge of a phase-out of the v1 API altogether, which will happen on May 10. If this is the case, then we have about 8 days to rework the mavengem backend to use the v2 API, or else beg the RG.org maintainers for an extension. |
I've entered a comment on an issue another v1 dependencies API user filed here: https://bundler.slack.com/archives/C02A04XKP/p1683047044420869 I have also reached out to @hsbt and @indirect to ask for more time to make this transition. |
Well there's good news after all. I conferred with the RubyGems.org team and we have a path forward:
Embedded users of JRuby or this plugin may still run into trouble and need updates, but I'm not sure how far-reaching that impact will be. @abelsromero Thank you for bringing this to our attention! We just recently took ownership of this plugin, and until yesterday I had no idea we were even dependent on this API. Thank goodness we found it when we did! |
Whatever you need 🙇 As Asciidoctor we are impacted in that some users may find this, but that's the Java+Maven ecosystem, which is just a subset of the whole family. And now that we have a WIP story, that's some way we can help them.
Glad it can work out in the end. I only feel bad for not reporting sooner, I've seen some indications someone fishy was going on and hadn't prioritized until now. Please do not stress over this. PS: we hijacked the issue, maybe re-title it or create a new one? |
Yeah it is unfortunate timing but I'm hopeful they'll have the whitelist in place very soon. Then we can set about fixing mavengem.
This issue was originally about the maven/gem proxy going down, and since it will never come back I'll just resolve this. The recommendation for users is to migrate to mavengem which remains, despite current problems, the path forward. |
Users still depending on these servers (which would be weird, since they've been dead for years) should migrate to https://github.com/jruby/mavengem, which performs all the rubygem/maven fakery on the client side. Recent issues with mavengem discussed here are summarized in jruby/mavengem#8 and will be fixed soon. |
jruby/mavengem#9 is largely complete and updates the backend of mavengem to use newer APIs. I will put out a release this week. |
Can't pull any maven dependencies. WHOIS showed that the domain expired on the 1st of May. Is the maintainer alive??
The text was updated successfully, but these errors were encountered: