~apw/+git/kteam-tools:swm/swm-publishing--track-builds-better

Last commit made on 2021-07-30
Get this branch:
git clone -b swm/swm-publishing--track-builds-better https://git.launchpad.net/~apw/+git/kteam-tools
Only Andy Whitcroft can upload to this branch. If you are Andy Whitcroft please log in for upload directions.

Branch merges

Branch information

Name:
swm/swm-publishing--track-builds-better
Repository:
lp:~apw/+git/kteam-tools

Recent commits

fb13de0... by Andy Whitcroft

swm-maintenance: elide positive records by default

Now that we are showing postive records (currently building etc) we need
to hide positive records by default. Also check the actual status not
the reported one when checking. Use --all to include them.

Signed-off-by: Andy Whitcroft <email address hidden>

f5a188d... by Andy Whitcroft

swm-publishing: check for changes in buildstate and trigger

Now that we keep maintenance records for all builds, check them against
the current state in launchpad and trigger a scan when they do not match.
This lets us detect the end of builds which result in failure.

Signed-off-by: Andy Whitcroft <email address hidden>

3f7fed4... by Andy Whitcroft

swm: package -- include all builds which are not successful in maintenance record

Include all builds which are not succesfully complete in the maintenance
records. This allows swm-publishing to probe building jobs moving failed
so it can trigger those to be scanned.

Signed-off-by: Andy Whitcroft <email address hidden>

3e3a6d2... by Andy Whitcroft

info/kernel-series: ppa2 is not a devel configured PPA

Signed-off-by: Andy Whitcroft <email address hidden>

be6fb53... by Andy Whitcroft

swm: scan performnce improvements

Scan performance is horrible. One major contributions is enumerating
build state for every package in every route (especially now we have
additional signing routes). Another is keeping the on-disk status.yaml
in sync between overlapping runs. Attack both of these.

Move to an on-demand instantiation of build information. Switch to
caching the status and validating the cached copy under the lock.

Acked-by: Stefan Bader <email address hidden>
Signed-off-by: Andy Whitcroft <email address hidden>

96c930f... by Andy Whitcroft

swm: report component check waits

When publishing to proposed we wait for full publication before we can
check components. When we are waiting for that report it so we know why
we are in that state.

Acked-by: Stefan Bader <email address hidden>
Signed-off-by: Andy Whitcroft <email address hidden>

98906c2... by Andy Whitcroft

swm: update state flags when manipulating the workflow task

We are not dropping the -live tags from trackers because we do not
reevaluate is_closed at al after changing the workflow task status.
Recheck those flags when manipulating the workflow task status.

Acked-by: Stefan Bader <email address hidden>
Signed-off-by: Andy Whitcroft <email address hidden>

3600652... by Andy Whitcroft

swm: bug,wft/workflow -- ensure we update is_closed et al on closure

We are not dropping the -live tags from trackers because we do not
re-evaluate is_closed at al after changing the workflow task status.

As things stand check_is_valid has a strange API, it is a class member
but is passed s.bug and returns the various status flags which then must
be assigned. Switch this function to a proper class function picking
s.bug up itself and setting the various state flags directly.

Now that check_is_valid is reusable call it to update these state flags
when we manipulate the workflow task status.

Signed-off-by: Andy Whitcroft <email address hidden>

ef9c37b... by Andy Whitcroft

swm: follow duplicate chain for invalid master bugs

Acked-by: Stefan Bader <email address hidden>
Signed-off-by: Andy Whitcroft <email address hidden>

96c0e40... by Andy Whitcroft

swm: add support for new gke artifact naming in the gcp bucket

Acked-by: Stefan Bader <email address hidden>
Signed-off-by: Andy Whitcroft <email address hidden>