Well It's difficult to track down a problem I can't reproduce. If I could, it would be easy. I would probably run WireShark in parallel and record the traffic and check what's going wrong.
If you want, Yishay, you could create a recording and somehow send that to me, I could try and have a look ... but I would imagine the file to be "lengthy" .. and ideals don't run any video or audio streamings while recording (Believe me ... I've done several PLC communication recordings while running some 4k festival streams in the background ... that's no fun) Did the changes I submitted change anything? You should be able to profit from them without having to re-release. Chris Am 29.04.20, 17:51 schrieb "Alex Harui" <[email protected]>: Chris, we've had this problem for years. Nobody is stopping you from figuring it out. I don't have the skills to do further analysis and we use the configuration you recommend that works for you and I. I think it is a bug in Maven/Wagon/StagingServer/Networking so I am proposing workarounds instead of putting people down. The problem uploading to staging seem to happen mainly to Windows users. IIRC, Piotr switched to a Mac using the same wires in his home and it got better. But I've used Windows out of my house near Seattle and it worked, and I'm interested to see if the CI server somewhere in the US also works. Also, there is some evidence that the upload will happen much faster from the CI server than from Yishay's home, so it could be a potential timesaver to upload from the CI server. To me and hopefully others, not requiring Macs and not hogging your local computer and network bandwidth are useful improvements. -Alex On 4/29/20, 2:14 AM, "Christofer Dutz" <[email protected]> wrote: Oh gee ... ok ... so instead of trying to find out what the problem is, you're adding a bunch of additional steps to the already extremely lengthy list? I mean it's not that the internet and especially the maven infrastructure hates you guys and therefore discriminates you. I would assume that probably the problem is somehow related to your (maven) setup. Am I correct with the assumption, that the RMs that have had these issues aren't very frequent users of Maven otherwise? I mean ... I ran the steps up to typedefs and all deployment was working fine without any issues, so I could assume that there might be an issue with your general setup. Chris Am 29.04.20, 09:38 schrieb "Alex Harui" <[email protected]>: I would first try #1. Hopefully you can have a good enough connection to upload a bunch of small .asc files. If we have to go with #2, I think there are lots of possibilities. a) have the Ant script zip up all of the .asc files b) create a git repo where you push the .asc files c) figure out how to use dropbox or something like that. HTH, -Alex PS: I'm shutting down for tonight. On 4/29/20, 12:26 AM, "Yishay Weiss" <[email protected]> wrote: >Don't know what the best workflow is until we try a few things. There are at least two workflows: > >1) Upload the artifacts from CI and RM uploads only the asc files as you suggested >2) RM uploads asc files to the CI server and everything gets uploaded to staging from CI. How do you suggest doing (2)? As I see it, we would need some SCP connection configured, or lots of manual steps with RDP. What am I missing?
