[Bug 238785] sysutils/py-google-compute-engine: Some GCE packages fill the logs with an error message

bugzilla-noreply at freebsd.org bugzilla-noreply at freebsd.org
Mon Jul 8 11:32:24 UTC 2019


https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=238785

--- Comment #6 from Mage <mage at mage.gold> ---
It got interesting. 

First, I could not reproduce the issue in a VM on my home computer. I thought
the reason was something like the daemon couldn’t connect and refused to run.

Then I created a test machine on Google cloud, using an image I created on
28.1.2019.

The issue also didn’t appear on this machine. That was normal as I started
having it only this year, likely after January.

However, when I upgraded the test machine’s packages to the current version
with pkg, the issue also did not appear.

Yet every machine that has been running in the last few months and that I kept
upgrading on a regular basis has the issue. In addition that I know I didn’t
edit any related config file (I don’t even know if py27-google-compute-engine
has one in the first place), the machines are on two separated accounts. On one
of the accounts, I hardly ever edit any config.

I thought that somewhere between 28 January and today, a specific version of
py27-google-compute-engine must have created some file that causes the issue.
This would explain why I don’t see it on either a fresh install (home VM) or on
the server I upgraded from 28 January to today’s state in a single step.

By the way, the upgrade was:

py27-google-compute-engine: 2.8.6 -> 2.8.13_5

I uninstalled every package that had "google" in their name. I executed pkg
autoremove as well.

There aren’t many files left in the system that have "google" in their path.

There is a folder called "/var/lib/google". That reminds me of the bad days of
Linux. Shouldn’t it be /var/db/google?

/var/lib/google # ls -l
total 14
drwxr-xr-x  7 root  wheel   7 Nov 15  2016 per-instance
-rw-r--r--  1 root  wheel  20 Nov 15  2016 vm-instance-id
-rw-rw-rw-  1 root  wheel   0 Sep  9  2018 vm-instance-id.lock

The 'per-instance' directory contains empty subdirectories that are named as a
19 digits decimal number. There are five of them from November 2016. 

That’s probably when I created my first instance. All my servers in Google
Compute Engine are a descendant or let’s say, a clone of my first instance that
I created from the FreeBSD image on GCE. I have been upgrading them for years. 

The vm-instance-id is neither the same nor different on each of my servers. On
some of the servers, it’s the same. On a few other servers, it’s unique. 

The file is from 2016 Movember from each of my servers. The lock is from 2018
September 9 everywhere. I’m not sure, but the latter might be when I migrated
from the deprecated google-cloud-something-cant-remember to the
py27-google-compute-engine.

I feel like deleting the /var/lib/google everywhere, or more like rm -rf
/var/lib. Can I do that?

In addition to making things right, I’m trying to figure out the cause of the
issue that’s the original subject of this ticket.

-- 
You are receiving this mail because:
You are on the CC list for the bug.


More information about the freebsd-python mailing list