disk/hdd-parking failed on non-hdd system and cause the system enters "degraded"

[jeremyszu]

Here are two problems.

1. disk/hdd-parking needs to check the system has HDD(s) as extra "requires:"
2. the hdapds should not be installed in non-HDD system OR the hdapds.service should be disabled if there is no HDD.

* plainbox-provider-certification-client installed hdapsd as Recommends by default.[1]
* hdapds will create a systemd service "hdapds.service" and if there is no hdd in the system then the system will enter "degraded" mode. [2]
* disk/hdd-parking use the command "command: hdd_parking.py" and it use
```
    try:
        hdapsd = Popen(
            ['/usr/sbin/hdapsd'], stdout=PIPE, stderr=PIPE,
            universal_newlines=True)
    except OSError as err:
```
without checking the hdd, it cause the test case failed[3].
* The DUT is only has nvme[4]

As least, the "disk/hdd-parking" needs to be executed only if the HDD presents.
I suggest to disable "hdapds.service" when there is no HDD in the system if the "plainbox-provider-certification-client" dependencies could not be checked in runtime.

[1]
$ apt show plainbox-provider-certification-client
Package: plainbox-provider-certification-client
...
Recommends: ..., hdapsd, ...

[2]
● hdapsd.service - hdapsd hard drive active protection system daemon
     Loaded: loaded (/lib/systemd/system/hdapsd.service; enabled; vendor preset: enabled)
     Active: failed (Result: exit-code) since Wed 2021-01-06 18:32:25 CST; 22h ago
       Docs: man:hdapsd(8)
    Process: 760 ExecStart=/usr/sbin/hdapsd (code=exited, status=1/FAILURE)
   Main PID: 760 (code=exited, status=1/FAILURE)

-- Logs begin at Wed 2020-12-30 20:24:13 CST, end at Thu 2021-01-07 17:35:58 CST. --
systemd[1]: Started hdapsd hard drive active protection system daemon.
hdapsd[760]: Wed Jan 6 18:32:25 2021: Could not detect any devices.
hdapsd[760]: Wed Jan 6 18:32:25 2021: Starting hdapsd
hdapsd[760]: Wed Jan 6 18:32:25 2021: WARNING: You did not supply any devices to protect, trying autodetection.
hdapsd[760]: Usage: hdapsd [OPTIONS]
hdapsd[760]: -c --cfgfile=<cfgfile> Load configuration from <cfgfile>.
hdapsd[760]: By default, configuration is read from /etc/hdapsd.conf
hdapsd[760]: -d --device=<device> <device> is likely to be hda or sda.
hdapsd[760]: Can be given multiple times
hdapsd[760]: to protect multiple devices.
hdapsd[760]: -f --force Force unloading heads, even if kernel thinks
hdapsd[760]: differently (on pre ATA7 drives).
hdapsd[760]: This only works when adding devices by hand (-d).
hdapsd[760]: -r --force-rotational Autodetect drives as rotational, even if
hdapsd[760]: kernel thinks they are not.
hdapsd[760]: -s --sensitivity=<sensitivity> How sensitive hdapsd should be to movements.
hdapsd[760]: Defaults to 15, higher value means less
hdapsd[760]: sensitive.
hdapsd[760]: -a --adaptive Adaptive threshold (automatic increase
hdapsd[760]: when the built-in keyboard/mouse are used).
hdapsd[760]: -v --verbose Get verbose statistics.
hdapsd[760]: -b --background Run the process in the background.
hdapsd[760]: -p --pidfile[=<pidfile>] Create a pid file when running
hdapsd[760]: in background.
hdapsd[760]: If <pidfile> is not specified,
hdapsd[760]: it's set to /var/run/hdapsd.pid.
hdapsd[760]: -t --dry-run Don't actually park the drive.
hdapsd[760]: -y --poll-sysfs Force use of sysfs interface to
hdapsd[760]: accelerometer.
hdapsd[760]: -H --hardware-logic Use the hardware fall detection logic instead of
hdapsd[760]: the software one (-s/--sensitivity and -a/--adaptive
hdapsd[760]: have no effect in this mode).
hdapsd[760]: -S --software-logic Use the software fall detection logic even if the
hdapsd[760]: hardware one is available.
hdapsd[760]: -L --no-leds Don't blink the LEDs.
hdapsd[760]: -l --syslog Log to syslog instead of stdout/stderr.
hdapsd[760]: -V --version Display version information and exit.
hdapsd[760]: -h --help Display this message and exit.
hdapsd[760]: You can send SIGUSR1 to deactivate hdapsd for 8 seconds.
hdapsd[760]: Send bugs, comments and suggestions to <email address hidden>
systemd[1]: hdapsd.service: Main process exited, code=exited, status=1/FAILURE
systemd[1]: hdapsd.service: Failed with result 'exit-code'.

[3]
$ cat com.canonical.certification__disk_hdd-parking.stdout
Starting HDD protection test - move the system around on all axis. No particular force should be required.
Tue Jan 5 01:52:02 2021: Starting hdapsd
Tue Jan 5 01:52:02 2021: WARNING: You did not supply any devices to protect, trying autodetection.
Usage: hdapsd [OPTIONS]

   -c --cfgfile=<cfgfile> Load configuration from <cfgfile>.
                                     By default, configuration is read from /etc/hdapsd.conf
   -d --device=<device> <device> is likely to be hda or sda.
                                     Can be given multiple times
                                     to protect multiple devices.
   -f --force Force unloading heads, even if kernel thinks
                                     differently (on pre ATA7 drives).
                                     This only works when adding devices by hand (-d).
   -r --force-rotational Autodetect drives as rotational, even if
                                     kernel thinks they are not.
   -s --sensitivity=<sensitivity> How sensitive hdapsd should be to movements.
                                     Defaults to 15, higher value means less
                                     sensitive.
   -a --adaptive Adaptive threshold (automatic increase
                                     when the built-in keyboard/mouse are used).
   -v --verbose Get verbose statistics.
   -b --background Run the process in the background.
   -p --pidfile[=<pidfile>] Create a pid file when running
                                     in background.
                                     If <pidfile> is not specified,
                                     it's set to /var/run/hdapsd.pid.
   -t --dry-run Don't actually park the drive.
   -y --poll-sysfs Force use of sysfs interface to
                                     accelerometer.
   -H --hardware-logic Use the hardware fall detection logic instead of
                                     the software one (-s/--sensitivity and -a/--adaptive
                                     have no effect in this mode).
   -S --software-logic Use the software fall detection logic even if the
                                     hardware one is available.
   -L --no-leds Don't blink the LEDs.
   -l --syslog Log to syslog instead of stdout/stderr.

   -V --version Display version information and exit.
   -h --help Display this message and exit.

You can send SIGUSR1 to deactivate hdapsd for 8 seconds.

Send bugs, comments and suggestions to <email address hidden>
[4]
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
...
nvme0n1 259:0 0 953.9G 0 disk
|-nvme0n1p1 259:1 0 512.1M 0 part /var/lib/snapd/hostfs/boot/efi
|-nvme0n1p2 259:2 0 4.3G 0 part
`-nvme0n1p3 259:3 0 949G 0 part /var/lib/snapd/hostfs

[jeremyszu (os369510)]

Hi Pierre,

could you please help to take a look at this issue?

[Pierre Equoy (pieq)]

Thanks for filing this up! I had a similar issue in another project and I thought it originated in the image content... Turns out it's a problem with Checkbox!

Thanks for spending time to investigate this as well. I'll see what we can do about this.

[Pierre Equoy (pieq)]

@Kent, given that there is less and less devices using spinning hard drives, and that this job (disk/hdd-parking) is not a cert-blocker, we are thinking of retiring this job from the certification test plan[1] (and removing the needs for hdapsd to be packaged in our Debian release).

What do you think from a QA point of view?

[1] https://git.launchpad.net/plainbox-provider-checkbox/tree/units/disk/test-plan.pxu#n7

[Maksim Beliaev (beliaev-maksim)]

Bug was migrated to GitHub: https://github.com/canonical/checkbox/issues/125.
Bug is no more monitored here.