abra incorrectly handles missing recipe folders with commands new and app ls --status
#749
Reference in New Issue
Block a user
No description provided.
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
edit
This issue can be related to your firewall setup. If you're encountering this issue, one possible cause is that you're using both
iptablesandnftables. If that doesn't resolve the issue, verify whether you have firewall rules in place that could prevent abra from pulling the repositories (includind DNS resolution).if you're on archlinux (like me) and have both
iptablesandnftables, you can fix it by installingiptables-nft(see Wiki). This can happen if you installdocker(since it depends onnftables), since a standard archlinux installation depends oniptables.original issue
I stumbled on the following issue (with
abra version 0.12.0-beta-db7c404), possibly because I copied my.abraover and muddled around with it a bit. I can't entirely reconstruct how this state arose, but I also figured out a fix (see below).Initial setup:
Bugreport
I typed
abra app ls -Sto see the status of my running services.Expected behavior: get a list of running services and their status
Actual behavior:
FATA <app/list.go:153> unable to retrieve tags for lauti.dyn.namnatulco.eu: repository does not exist
``´
Aftter git init in the new folder, it works as expected.
I have attached a .txt with some outputs that describe the debugging in detail.
okay, and I can confirm the issue for the
newcommand;This also happens with other recipes I haven't used, so maybe the issue is that I broke something while doing stuff to my
.abrafolder.can reproduce the issue with a fresh install and a new .abra directory.
It could be related to the IPv6 lookup; I tried to force IPv4 by disabling IPv6 on the loopback and wireless interfaces (via
/proc/sys/net/ipv6/conf/INTERFACE/disable_ipv6), but it still prints the v6[::]addresses. Pulling works fine, though. Unfortunately that doesn't seem to be the case for the recipe fetching.The
newissue seems to be a separate problem. I tried to chase it down through the go-code, but it looks like the git clone indeed just fails (and the debug output that the clone was successful seems to be incorrect -- this suggests it was either an authentication error or an empty remote)Replication via commandline git:
skipping the redirect gives the same output without the warning.
Removing the
/info/refs?service=git-upload-packworks:unfortunately it seems tracking down where the DNS issue comes from is beyond my go skills. Does it make sense to split this into a separate issue at this point?
In case anyone reads this with the same issue -- a workaround seems to be manually pulling the repos and then using
--offlineclosed as discussed on matrix, since better errorhandling is kidn of a rabbithole