Hi Maxim, On Wed, 28 Sep 2022 at 20:32, Maxim Cournoyer wrote: > Can this issue be closed? This “meta” bug raises 2 levels of issues for long-term: 1. save the source code, 2. save the current binary substitutes. For #1, we have now a roadmap to tackle this via Disarchive or sources.json and SWH [1,2]. Therefore, this point of the “meta” bug can be closed. However, about #2, we do not have a roadmap, AFAIK. For instance «Substitute retention» [3] is still an issue. The recent outage of Berlin exemplifies the potential problems. (Note that because of energy troubles in Europe, I do not exclude some potential short blackout or power outage for short period of time of the Berlin server.) Why kept binary substitutes? For instance, it is not possible to rebuild some packages on modern CPU; see OpenBLAS [4]. Therefore, waiting a better solution, it appears to me a pragmatic solution to keep as much substitutes as we are able to. We do not have a clear policy between Berlin and Bordeaux. 1: 2: 3: 4: > Otherwise, what remains to be acted upon? The next action for this #2 is to address what I sent on guix-sysadmin about duplicating the storage of binary substitutes in some machine offered by INRAe / Univ. of Montpellier. And we need to draw a roadmap to tackle this #2. Then, this “meta” bug could be closed, IMHO. Cheers, simon