Based on the comment that this was observed on servers, I tried this on a server installation of Ubuntu, though the problem should still occur with a client installation. First, with checkbox 0.13.7 installed: $ dpkg --list 'checkbox*' Desired=Unknown/Install/Remove/Purge/Hold | Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend |/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad) ||/ Name Version Description +++-==============-==============-============================================ ii checkbox 0.13.7 System testing application ii checkbox-cli 0.13.7 Command line interface for checkbox un checkbox-gtk (no description available) I had to change the test command to use checkbox-cli instead of checkbox-qt. Otherwise the way it works is exactly the same. I did the test run, the fwts_wakealarm test did NOT cause the system to suspend. The xml report shows the test as passed: pass fail pass uninitiated unresolved unsupported untested but then the fwts_wakealarm.log says it didn't actually do anything: 00012 wakealarm Test ACPI Wakealarm. 00013 wakealarm ----------------------------------------------------------------------------- 00014 wakealarm Must be run as root or sudo to be able to read system information. 00015 wakealarm Aborted test, insufficient privilege. 00016 wakealarm ============================================================================= 00017 wakealarm 0 passed, 0 failed, 0 warnings, 4 aborted, 0 skipped, 0 info only. 00018 wakealarm ============================================================================= OK, then I remove the .cache/checkbox directory, upgrade to checkbox 0.13.8 from proposed: $ dpkg --list 'checkbox*' Desired=Unknown/Install/Remove/Purge/Hold | Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend |/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad) ||/ Name Version Description +++-==============-==============-============================================ ii checkbox 0.13.8 System testing application ii checkbox-cli 0.13.8 Command line interface for checkbox un checkbox-gtk (no description available) I ran the tests using the given testing command. I noticed the system did *not* suspend, though it *did* take longer than the previous try. The test is now running as root, and the proof is that the fwts-wakealarm.log file is getting created in /. This is not entirely desirable but the behavior is at least better than it was previously (i.e. non-running test). Here's the result from the fwts-wakealarm.log: 00012 wakealarm Test ACPI Wakealarm. 00013 wakealarm ----------------------------------------------------------------------------- 00014 wakealarm Test 1 of 4: Check existance of /sys/class/rtc/rtc0/wakealarm. 00015 wakealarm PASSED: Test 1, /sys/class/rtc/rtc0/wakealarm found. 00016 wakealarm 00017 wakealarm Test 2 of 4: Trigger wakealarm for 1 seconds in the future. 00018 wakealarm Trigger wakealarm for 1 seconds in the future. 00019 wakealarm PASSED: Test 2, RTC wakealarm was triggered successfully. 00020 wakealarm 00021 wakealarm Test 3 of 4: Check if wakealarm is fired. 00022 wakealarm PASSED: Test 3, RTC wakealarm triggered and fired successfully. 00023 wakealarm 00024 wakealarm Test 4 of 4: Multiple wakealarm firing tests. 00025 wakealarm Trigger wakealarm for 1 seconds in the future. 00026 wakealarm Trigger wakealarm for 2 seconds in the future. 00027 wakealarm Trigger wakealarm for 3 seconds in the future. 00028 wakealarm Trigger wakealarm for 4 seconds in the future. 00029 wakealarm PASSED: Test 4, RTC wakealarm triggered and fired successfully. 00030 wakealarm 00031 wakealarm ============================================================================= 00032 wakealarm 4 passed, 0 failed, 0 warnings, 0 aborted, 0 skipped, 0 info only. 00033 wakealarm ============================================================================= So it's now passing because it was able to run. The test case mentions the system should suspend, but since we're just testing the wakealarm functionality, it's not really necessary to suspend, just to ensure that the wakealarm is set and triggers correctly. I think this is good enough, I'll set the bug to verification-done since at least the test is now running.