* New machines running in arm64 servers
* Fix for tests and workflow update to support a different backend for arm
* Use spread-arm when the system is arm64
* Fix matrix.system comparison
* Fix save results step in github worflow
* fix error in mount-order-regression test
* Skip upgrade-from-release on arm machines
* skip preseed-core20 test on arm
The model is for amd64, so it fails to prepare the image in arm
* Adding more logging to the run tests workflow job
* Fix uc20-create-partitions for pc-amd64 arch
* Update spread run command
This is because it failed to re-execute failed
* Run just on focal
This was discussed to reduce the number of machines we create on each
workflow execution
* Add executions for Ubuntu Jammy
There are some tests that are not been executed in Jammy
This change adds ubuntu 2* to supported systems on spread tests
* Fixes for tests
* revert change in test interfaces-opengl-nvidia
This needs some changes in the code and it will be done in a following
pr
* revert change to fake-netplan-apply
Co-authored-by: Michael Vogt <mvo@ubuntu.com>
The test is failing after it has been moved to focal, the change
includes:
. Add new dependency for focal
. Use the new group libvirt instead of the old one libvirtd
This test was moved to run on `20.04` but there are at least two
problems:
1. The `libvirtd` group got renamed to `libvirt`
2. The <kernel>test_ping_serve.virtio</kernel> errors when it
comes from a read-only medium like a snap
To unblock us short term this commit just reverts back to 16.04 to
unbreak master and then we can work on a proper fix.
MATCH is a wrapper around grep -E -q, thus MATCH -v ends up being grep -E -q -v.
However, grep -v is peculiar as far as exit codes are concerned, and returns 0
when lines not matching the pattern were found in the input.
For example, this command returns 1:
$ cat <<EOF | grep -E -v foo
foo
foo
foo
EOF
but this one returns 0, even though the pattern we don't want is in the input:
cat <<EOF | grep -E -v foo; echo $?
foo
bar
baz
EOF
In the tests, MATCH -v is used to check that the pattern does not appear in the
input, which is incorrect.
Signed-off-by: Maciej Borzecki <maciej.zenon.borzecki@canonical.com>
When default umask is strict, eg. 0077, the
/home/gopath/src/github.com/snapcore/snapd hierarchy will have 0700 permissions,
meaning attempts to write to a place under that path as a `test` user will fail.
This has come up with Fedora 29 where the default umask was changed to a more
strict one.
Signed-off-by: Maciej Borzecki <maciej.zenon.borzecki@canonical.com>
This is done in the tests which are checking the interfaces information
in order to make them more robusts, making the test independent of other
plug:slots declared such it is happening in gadgets.
This change includes:
. Remove all the checks that are done after a connection and
disconnection (this is already checked as part of the snap_connect and
snap_disconnect tests).
. install_local function used to pack and install local snaps
. Remove not needed echo
. Change the check of interfaces default behaviour for the
autoconnection to support the scenario when there are other snaps also
connected to the interface (that happens when testing caracalla gadget)
. Add new lines between task tags
The objective of this change is to provide an environment for all the
tests as similar as possible, so all the dependencies are installed
during the project setup.
The packages are not removed at the end of the project, so it will help
in case the machine needs to be used to debug.
Openvswitch test is moved to manual due to the dependencies install is
stuck trying to install openvswitch-switch package.
apt update is hanging on security.ubuntu.com with IPv6, we are also getting timeouts on other apt operations that might be related to the same problem.
Instead of disabling it and enabling after concrete apt calls, with these changes we give weight to IPv4 over IPv6 at the prepare stage of the suite and this setting is kept during the whole execution.
