Codespaces: "This Codespace is currently running in recovery mode..."

[iaindillingham]

@wjchulme encountered the following when starting a Codespace from opensafely/covid-vaccine-history's main branch:¹

It looks as through the creation.log contains some useful and interesting information.

(Thanks, @tomodwyer for the screenshot and the creation log.)

Footnotes

1. https://bennettoxford.slack.com/archives/C33TWNQ1J/p1729177307365829 ↩

[iaindillingham]

@Mary-Anya, @wjchulme, and I talked about this issue on Thursday 14th October. A patchy Internet connection meant we didn't get far, so, on Slack, I asked Will to gather some extra information about "stunning memory", the troublesome Codespace.¹

• stunning-memory-creation.log
• logs.zip

I also asked Will about his download and upload speeds. More in #77.

Footnotes

1. https://bennettoxford.slack.com/archives/C33TWNQ1J/p1729769704779559?thread_ts=1729177307.365829&cid=C33TWNQ1J ↩

[lucyb]

This issue was resolved by creating a new Codespace.

Documentation for explaining and resolving this issue added here: opensafely/documentation#1660

[wjchulme]

this issue was resolved by creating a new Codespace.

Just to clarify that this wasn't a one-off, it happened a few times in different codespaces. The stunning-memory codespace was my nth attempt to create a new codespace over the course of a few days (in both the covid-vaccine-history and bennett.ox.ac.uk repos). Great that it's now working but it wasn't a single isolated incident.

[StevenMaude]

@wjchulme: If you observe similar behaviour again, please flag it to support again, of course.

(From the logs, this particular problem could have been at GitHub's end, or I think another log I saw pointed to some transient issues accessing Canonical's Ubuntu-related repositories. So it doesn't seem like it's anything that we've directly changed to blame.)

[lucyb]

Steve is correct, the bennett.ox.ac.uk problems could have been affected by an outage at Canonical that was happening around that time. I'm less sure about the research Codespaces though. You should definitely contact support if it happens again. From the documentation, it sounds like GitHub expect these issues to happen very occasionally, so we'd like to know that's not the case.