ERROR while fetching sources
hannesm opened this issue · 4 comments
Dear Madam or Sir,
I can see quite some issues since a few days that look like your worker cluster is unhappy. Rebuilding such jobs does not result in proper solutions. If you have some time, it would be great if you could have a look into the permission issue below.
#=== ERROR while fetching sources for randomconv.0.2.0 ========================#
OpamSolution.Fetch_fail("Cannot copy /home/opam/.opam/4.14/.opam-switch/sources/randomconv.0.2.0/randomconv-0.2.0.tbz to /home/opam/.opam/download-cache/sha512/37/376c36da6b67ed1d817ea13fbbc3490f356f0890f9e009a55d12946c6811611ea9bd31aec733dd961de7209effa23649505f61d1ed3fc18af225ca4ec9131c38 (/usr/bin/opam: \"open\" failed on /home/opam/.opam/download-cache/sha512/37/376c36da6b67ed1d817ea13fbbc3490f356f0890f9e009a55d12946c6811611ea9bd31aec733dd961de7209effa23649505f61d1ed3fc18af225ca4ec9131c38: Permission denied).")
#=== ERROR while fetching sources for ppx_bench.v0.16.0 =======================#
OpamSolution.Fetch_fail("Cannot copy /home/opam/.opam/4.14/.opam-switch/sources/ppx_bench.v0.16.0/ppx_bench-v0.16.0.tar.gz to /home/opam/.opam/download-cache/sha256/e3/e307fc25b4cb38125685fa01888255d00aaf6c1b82f52c4f02ebd48a4471761d (/usr/bin/opam: \"open\" failed on /home/opam/.opam/download-cache/sha256/e3/e307fc25b4cb38125685fa01888255d00aaf6c1b82f52c4f02ebd48a4471761d: Permission denied).")
#=== ERROR while fetching sources for core_kernel.v0.16.0 =====================#
OpamSolution.Fetch_fail("Cannot copy /home/opam/.opam/4.14/.opam-switch/sources/core_kernel.v0.16.0/core_kernel-v0.16.0.tar.gz to /home/opam/.opam/download-cache/sha256/e3/e37370bad978cfb71fdaf2b1a25ab1506b98ef0b91e0dbd189ffd9d853245ce2 (/usr/bin/opam: \"open\" failed on /home/opam/.opam/download-cache/sha256/e3/e37370bad978cfb71fdaf2b1a25ab1506b98ef0b91e0dbd189ffd9d853245ce2: Permission denied).")
FWIW, I can as well spot this issue on the opam repository CI.
Thanks for the report, @hannesm.
I've tried investigating the issue locally using the Dockerfile, but haven't been able to reproduce it. We'll look into it, further.
All of the failed jobs are on the machine asteria. Investigating the cache on this machine showed that the permissions on a few of the cache files were root:root rather than opam:opam. These permissions have been updated. Rebuilding the jobs now works. There is nothing obvious in the log to indicate why these permissions were wrong.
Sounds like this has been resolved for the time being. If it recurs, we have have enough context here to help support a root cause investigation, and we can reopen.