Dear all,

Thank you for reaching out to us and reporting this.

The current installation scripts for Debian have a provision that is intended to perform a migration

of the configuration files from the pre-FHS directories to the new post-FHS directories, similar to how

it is done for RPM (many thanks to Robert Scheck for highlighting the code snippet). For Debian, the

relevant snippet can be found in `ripe-atlas-common.preinst` as follows:

migrate_file()
{
	if ( [ -f "$1" ] && ! cmp -s "$1" "$2" 1>/dev/null 2>&1 ); then
		install -D -p -m "$3" -o "$4" -g "$5" "$1" "$2" 1>/dev/null 2>&1
	fi

}

...

case "$1" in
	install)
...
		if ! [ -d "/etc/ripe-atlas" ]; then
			install -d -m 0770 -o "ripe-atlas" -g "ripe-atlas" "/etc/ripe-atlas" 1>/dev/null 2>&1
		fi
		migrate_file "/var/atlas-probe/etc/probe_key"     "/etc/ripe-atlas/probe_key"                  0600 \
		             "ripe-atlas"       "ripe-atlas"
		migrate_file "/var/atlas-probe/probe_key.pub" "/etc/ripe-atlas/probe_key.pub"              0644 \
		             "ripe-atlas"       "ripe-atlas"
		migrate_file "/var/atlas-probe/state/mode"    "/etc/ripe-atlas/mode.atlasswprobe"    0644 \
		             "ripe-atlas"       "ripe-atlas"
		migrate_file "/var/atlas-probe/state/config.txt"  "/etc/ripe-atlas/config.txt"               0644 \
		             "ripe-atlas"       "ripe-atlas"

;;

It is important to note that I have also expanded the variables to facilitate readability of the code snippet.

On a minimal, clean Debian 12 test system, I further executed the following steps:

- Compiled and installed a `5080` software probe.

- Downloaded the `common` and `probe` packages for the Debian 12 distribution from the FTP server.

- Executed `dpkg -i ripe-atlas-common_5100_amd64.deb ripe-atlas-probe_5100_all.deb` with superuser privileges.

Before installing the `5100` package, the state directory for `5080` contained the following items

(output of `ls -la /var/atlas-probe/state`):

total 16

drwxr-xr-x 2 atlas root 4096 Feb 25 12:08 .

drwxr-xr-x 9 atlas root 4096 Feb 25 12:07 ..

lrwxrwxrwx 1 atlas nogroup 44 Feb 25 12:07 FIRMWARE_APPS_VERSION ->

/usr/local/atlas/state/FIRMWARE_APPS_VERSION

-rw-r--r-- 1 atlas root 11 Feb 25 12:08 config.txt

-rw-r--r-- 1 atlas root 5 Feb 25 10:08 mode

The configuration directory for `5080` contained the following (output of `ls -la /var/atlas-probe/etc`):

total 16

drwxr-xr-x 2 atlas nogroup 4096 Feb 25 12:07 .

drwxr-xr-x 9 atlas root 4096 Feb 25 12:07 ..

-rw------- 1 atlas nogroup 1823 Feb 25 12:07 probe_key

-rw-r--r-- 1 atlas nogroup 394 Feb 25 12:07 probe_key.pu

After installation as described above, the post- FHS configuration directory for `5100` contained the following items

(output of `ls -la /etc/ripe-atlas/):

total 24

drwxrwx--- 2 ripe-atlas ripe-atlas 4096 Feb 25 12:09 .

drwxr-xr-x 74 root root 4096 Feb 25 12:09 ..

-rw-r--r-- 1 ripe-atlas ripe-atlas 11 Feb 25 12:08 config.txt

-rw-r--r-- 1 ripe-atlas ripe-atlas 5 Feb 25 10:08 mode

-rw------- 1 ripe-atlas ripe-atlas 1823 Feb 25 12:07 probe_key

-rw-r--r-- 1 ripe-atlas ripe-atlas 394 Feb 25 12:07 probe_key.pub

It is important to note that the RIPE Atlas service is now disabled and stopped by default post-installation. As such,

the final step would be to execute `systemctl enable ripe-atlas.service && systemctl start ripe-atlas.service` with

superuser privileges in order to start the probe up.

Based on the outcome of this approach, it appears that executing the steps described above resulted in successful

migration of the configuration files. Therefore, executing the aforementioned steps would be the best practice

recommended from our end in this case.

We also thank you for highlighting the need for more detailed documentation on this matter. We will update the

documentation to list this process for all Debian distributions (currently Debian 11 and 12) in an upcoming release.

Best regards,

Marios Mavropoulos Papoudas

Software Engineer @ RIPE Atlas

RIPE NCC