Comment 37 for bug 2022915

The autopkgtest failures on focal are expected, and not an issue.

422s Summary
422s
422s failures true time 3 Size 0.00MiB List plugins
422s failures true time 3 Size 0.00MiB List presets
422s failures true time 2 Size 0.00MiB List profiles
422s failures true time 3 Size 11.69KiB --no-report --label TEST -o hardware
422s failures true time 11 Size 3.15MiB --log-size 0 --no-env-vars -z gzip
422s failures true time 29 Size 2.31MiB --mask
422s
422s Tests Failed: 9
422s Failures within each test:
422s
422s run_expecting_success: test generated stderr output, see above
422s run_expecting_success: test generated stderr output, see above
422s run_expecting_success: test generated stderr output, see above
422s run_expecting_success: test generated stderr output, see above
422s run_expecting_success: test generated stderr output, see above
422s run_expecting_success: test generated stderr output, see above
422s run_expecting_success: test generated stderr output, see above
422s run_expecting_success: test generated stderr output, see above
422s run_expecting_success: test generated stderr output, see above

As an example,

343s !!! TEST FAILED: test generated stderr output, see above !!!
343s ### start stderr
343s
343s WARNING: Failed to load 'magic' module version >= 0.4.20 which sos aims
343s to use for detecting binary files. A less effective method will be used.
343s It is recommended to install proper python3-magic package with the
343s module.
343s
343s ### end stderr

These failures are expected, and are a result of https://github.com/sosreport/sos/pull/3025 which makes python3-magic optional. On focal we do not have the required python3-magic version, so we default back to the older way of detecting binaries in sosreport. Unfortunately this then spews that message on stderr.

Even though we have

Restrictions: needs-root, allow-stderr, isolation-machine

in the d/tests/control file, the tests fail because the test returns 1 when it sees stderr output.

The simple.sh file which has the sosreport tests has the following code,

run_expecting_success () {
    #$1 - is command options
    #$2 - kind of check to do, so far only extract
    FAIL=false

   <snip>

    if [ -s /dev/shm/stderr ]; then
       add_failure "test generated stderr output, see above"
       echo "### start stderr"
       cat /dev/shm/stderr
       echo "### end stderr"
    fi

The add_failure then does

add_failure () {
    FAIL=true <--
    echo "!!! TEST FAILED: $1 !!!"
    FAIL_LIST="${FAIL_LIST}\n \t ${FUNCNAME[1]}: \t\t ${1}"
}

So the test returns 1 not the expected 0,

    if $FAIL; then
      NUMOFFAILURES=$(($NUMOFFAILURES + 1))
      return 1 <---
    else
      return 0
    fi