Roots Discourse

Provisioning production fails : no "Release file"

[edit: I removed some stuff that had nothing to do with the issue at hand. Don’t use <? tags, folks]
Hello, I’ve run into problems re-provisioning my server.

Failed to update apt cache: W:Updating from such a repository can't be done
securely, and is therefore disabled by default., W:See apt-secure(8) manpage
for repository creation and user configuration details., E:The repository
'http://nyc2.mirrors.digitalocean.com/mariadb/repo/10.5/ubuntu bionic
Release' no longer has a Release file.
fatal: [production_host]: FAILED! => {"changed": false}

Seems like Ubuntu Bionic is out of date…
Now, I’ve looked around to see if there’s any sense in upgrading Ubuntu with ‘do-release-upgrade’ like it suggests, but I have a feeling that doing so would break everything without any hope of repair.

I’m running trellis-cli 1.0.0 and the recommended ansible 2.9.10

2 Likes

The culprit seems to be /trellis/roles/mariadb/defaults/main.yml. I removed the [arch=amd64] tag like in commit 32b1b6fcec8cb9fe14c6ec43cc1fccd3851fac2f just to see if that did anything but it didn’t.

Actually, so it turns out that the provisioning problem and the deploy playbook problem were unrelated. Still having issues with that mariadb mirror seemingly standing in the way of provisioning but the other stuff was just… a <? tag that should have been a <?php tag instead. :man_facepalming:t2:

Same issue here. I think I provisioned successfully a couple weeks ago, so this seems to be a recent issue.

Running into the same issue since a few hours ago, I had run a successful provision as recent as yesterday.

Thinking it could be something to do with the DigitalOcean mirror for MariaDB?

Chatter on MariaDB Zulipchat seems to indicate it may be an upstream issue on MariaDB’s mirrors. https://mariadb.zulipchat.com/

The way I managed to solve this was replacing http://nyc2.mirrors.digitalocean.com/mariadb/repo/10.5/ubuntu with https://mirror.mva-n.net/mariadb/repo/10.5/ubuntu in trellis/roles/mariadb/defaults/main.yml.

It shouldn’t be considered a long-term solution, but at least it did the trick for the time being.

Same issue. I used http://ftp.osuosl.org/pub/mariadb/repo/10.5/ubuntu, which was used before: Add MariaDB 10.2 PPA · roots/trellis@9d1bdad · GitHub

I tried changing the URL but the provision keeps calling the DigitalOcean link.

It seems like the mirror is cached somewhere, change it to any of the proposed mirrors won’t resolve the error for me.

This worked for me:

Change the url:

SSH into the guest machine:

$ trellis ssh development

In the guest machine:

List of all repositories (just to check)

$ grep ^ /etc/apt/sources.list /etc/apt/sources.list.d/*

Go to directory:
$ cd /etc/apt/sources.list.d

Delete the file:
$ sudo rm nyc2_mirrors_digitalocean_com_mariadb_repo_10_5_ubuntu.list

From host machine (trellis folder):
$ vagrant provision

4 Likes

apt-get update should generally clear the sources cache.

Now I’m wondering if we should move from Digitalocean as the default mirror though. I think this is the second time they’ve had issues.

We’ve released a new version of Trellis to address this by changing the mirror of the MariaDB repo: Release v1.9.1 · roots/trellis · GitHub

3 Likes

Just encountered this issue.

The PR fixes the apt source, however, running the playbook will still terminate with an ansible error as apt is apparently updated first (task Update apt packages somewhere before task Add MariaDB PPA that fixes the apt source).

New version: Thats perfect! :+1:

Does reprovisioning fix the issue directly on an existing Trellis system?
I had to manually remove the sources and then reprovision, as an apt upate caused ansible to fail before changing the source.