udev parser could be a bit smarter about known items that lead to false test failures
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| Checkbox Support Library |
Fix Released
|
Medium
|
Sylvain Pineau | ||
Bug Description
This is probably more an issue on servers, but there are occasionally things that create false test failures that could be prevented by a little more targeted filtering in udev output.
Consider this:
iDRAC: Virtual CD CDROM
iDRAC: Virtual Floppy DISK
On Dell servers, the iDRAC presents virtual devices like this and they appear in udev output as actual devices, thus the cert suite then tries running tests against them (well, maybe not the Floppy but it does the CD).
Here's a sample test result showing where a cd test was attempted, but failed due to the fact that the CD is not real:
https:/
This would happen on most Dell systems with the iDRAC manager. It also happens on pretty much any Blade system where the chassis management system presents a USB or Virtual CD/Floppy to individual blades, as well as Cisco UCS where the UCS Manager does essentially the same thing.
We should start collecting instances of this to improve the device filtering.
So this bug should start by focusing on iDRAC systems.
Related branches
- Paul Larson: Approve
- Sylvain Pineau: Needs Resubmitting
-
Diff: 11116 lines (+11070/-3)3 files modifiedcheckbox-support/checkbox_support/parsers/tests/test_udevadm.py (+14/-3)
checkbox-support/checkbox_support/parsers/tests/udevadm_data/DELL_IDRAC.txt (+11051/-0)
checkbox-support/checkbox_support/parsers/udevadm.py (+5/-0)
| affects: | checkbox → plainbox-provider-checkbox |
| Daniel Manrique (roadmr) wrote | #1 |
| Daniel Manrique (roadmr) wrote | #2 |
| Changed in plainbox-provider-checkbox: | |
| status: | New → Incomplete |
| Jeff Lane (bladernr) wrote | #3 |
| Mark W Wenning (mwenning) wrote | #4 |
| Mark W Wenning (mwenning) wrote | #5 |
| Changed in plainbox-provider-checkbox: | |
| status: | Incomplete → New |
| Mark W Wenning (mwenning) wrote | #6 |
| Changed in plainbox-provider-checkbox: | |
| assignee: | nobody → Sylvain Pineau (sylvain-pineau) |
| importance: | Undecided → Medium |
| affects: | plainbox-provider-checkbox → checkbox-support |
| Changed in checkbox-support: | |
| status: | New → In Progress |
| milestone: | none → 0.24 |
| Changed in checkbox-support: | |
| status: | In Progress → Fix Committed |
| Changed in checkbox-support: | |
| status: | Fix Committed → Fix Released |
Hi Jeff,
Could you please run
udevadm info --export-db > idrac-udevadm.txt
and attach the resulting file? Using this we can look for stuff to use to "weed out" these devices.
I'm not sure we can use something like device == CDROM and not device.type == 'IDRAC' (which fails due to an ancient checkbox bug I think) but it's a starting point. If we decide to change the parser to identify these devices, raw udev data is a first step so we can add this to our test suite to ensure we don't break other stuff.