Fork me on GitHub
Slack Archive

This page is not created by, affiliated with, or supported by Slack Technologies, Inc.

2020-02-10

Channels

#clojars
<
2020-02-10
>
vemv04:02:41

As a (non-urgent) FYI, clojars deployments have been deploying for me with: 

Could not transfer metadata xxxxx:xxxxxxxx/maven-metadata.xml from/to clojars (): Read timed out
Failed to deploy metadata: Could not transfer metadata xxxxx:xxxxxxxx/maven-metadata.xml from/to clojars (): Read timed out
This is the second day in a row. Can be worked around by retrying

tcrawley04:02:16

Usually when that happens the deploy actually succeeds. I did some work earlier today to do less during a deploy to prevent this from happening. It should be even less tomorrow evening which I hope will help out with this issue.

vemv04:02:48

You're right - all my deploys had actually succeed! (but the responsible CI build had failed) Hadn't realised this. Thanks for all the info!

tcrawley04:02:25

My pleasure! And just to give more detail - after your client uploads the maven-metadata.xml file, but before the server responds, the server does the following: • it verifies the deploy is valid • it writes the deployed artifacts to Cloudfiles • it writes the deployed artifacts to S3 • until earlier today, it added the deploy to the search index Once that is all done, it responds to your client. So if that takes too long, the client times out. I made indexing async today, and hope to drop writing to Cloudfiles tomorrow (we're in the middle of a transition from Cloudfiles to S3, and have to dual-write for the time being - see https://groups.google.com/d/msg/clojars-maintainers/LvfZMiGTYWg/QxBo7PG3AgAJ for details).

✅ 4
👏 4
tcrawley04:02:25

My pleasure! And just to give more detail - after your client uploads the maven-metadata.xml file, but before the server responds, the server does the following: • it verifies the deploy is valid • it writes the deployed artifacts to Cloudfiles • it writes the deployed artifacts to S3 • until earlier today, it added the deploy to the search index Once that is all done, it responds to your client. So if that takes too long, the client times out. I made indexing async today, and hope to drop writing to Cloudfiles tomorrow (we're in the middle of a transition from Cloudfiles to S3, and have to dual-write for the time being - see https://groups.google.com/d/msg/clojars-maintainers/LvfZMiGTYWg/QxBo7PG3AgAJ for details).

✅ 4
👏 4