Hi garden team! I've been helping out raising issues in github earlier but today I'd like to use discord because I feel the problem at hand is super weird and might need some back and forth. In our setup we have around 5 remote sources. When issuing

garden update-remote source --parallel

(via workflow), the relevant log looks like in attached file

garden.log

. So far, so good. However, sometimes -- maybe every 30th run -- the module resolution mysteriously fails. See attached

garden-error.log

. Please note that in the successful run there are 44 modules; in the failed run garden only finds 39: 5 are missing. When retrying the failed pipeline, it normally works. I'd be really happy to eliminate this flakiness in our CI 🙂 Do you have any idea what is going on?

Hi @brief-dawn-88958! I've asked our internal support team to take a look at this as I'm not too familiar with our remote dependency resoluton

Hey @brief-dawn-88958! A few questions to (hopefully) help us narrow this down: 1. When this fails, does it consistently happen for the same build dependency,

bex-os-db-utils

? Or does it seem random? 2. Is there something different about the repo/repos that sometimes fail? E.g. something requiring an auth token to be refreshed? Just thought I'd rule that out. 3. Could you share more details on the workflow config in question (e.g. the step commands involved) and any relevant parts of the module config of the

analytics

module (to the extent you can without sharing sensitive details)? In the meantime, I'll ask around some more and see if we come up with some suggestions. Thanks!