►
From YouTube: Testgrid - New Contributor Workshop Seattle 2018
Description
Speaker: Michelle Shepardson
A
All
right
can
everyone
hear
pretty
well:
okay,
cool
hi,
everyone,
I'm
Michelle
I'll,
go
ahead
and
go
to
my
very
green
slide,
so
yeah
I'm,
a
software
engineering
tools
and
infrastructure
at
oh
cool
I'm,
a
software
engineering
tools,
an
infrastructure
at
Google,
I
work
on
the
Google
kubernetes,
inter
prod
team.
You
might
know
some
of
my
co-workers.
They
work
on
prowl
and
whatnot,
but
yeah.
So
today,
I'll
just
give
like
a
brief
whirlwind
tour
of
test
period:
how
to
find
things
on
it
and
some
of
the
cool
features.
So
first,
what
is
it?
A
I
think
you
saw
a
little
bit
earlier,
but
it's
basically
just
a
tool
we
use
to
take
test
grid,
their
test
results,
be
them
and
grid
and
then
be
able
to
see
like
patterns
and
failures
over
time.
So
here,
for
instance,
we
could
see
easily
that,
like
there
are
some
things
that
were
failing
consistently
a
couple,
things
that
were
flaking
and
a
lot
of
tests.
They're
agreeing
and
with
that
that's
the
end
of
my
slides.
A
Awesome,
thank
you.
Alright.
So
first
things.
First,
this
is
the
main
index
for
a
test
grid.
This
has
all
of
the
currently
collected
kubernetes.
Dashboards
gray
is
like
a
single
dashboard
that
has
a
collection
of
related
tests.
Underneath
it
blue,
is
something
that
has
a
set
of
dashboards
underneath
it.
So,
for
instance,
you
could
see
for
sig
testing.
We
have
a
few
things
for
like
Canaries
miscellaneous
some
stuff
for
kind,
specifically
some
stuff
around
parable
offs.
A
A
There
is
a
particular
dashboard
I
wanted
nope
you
see,
I
did
you
see
there
yeah
and
then
the
the
search
bar
can
help
like
find
a
lot
of
the
smaller
tabs
that
are
like
not
as
obvious
from
this
view,
so
I
think
I,
remember,
like
a
co-worker
said
that
this
had
interesting
results,
so
I'm
going
here
and
checking
there
are
a
few
tests
that
look
kind
of
flaky
a
lot
of
green
results.
So
generally
the
dashboard
looks
pretty
good.
Well,
let's
say
I
want
to
view
some
like
more
specific
things
as
well.
A
A
There
we
go
okay,
so
that's
sort
of
my
name.
We
can
do
a
few
other
things
as
well.
So
let's
say
that
I
specifically
want
to
I.
Think
I.
Remember
there
was
some,
no,
no,
no
sick,
storage
stuff
that
I
wanted
to
be
cool,
so
you
can
do
things
like
filtered
test
by
what
you
want
to
see
or
otherwise
what
you
like
don't
want
to
see
a
for
instance,
if
I
didn't
want
to
view
a
sick
storage,
I
could
put
that
in
the
exclude
filter
by
regex.
A
A
A
A
So
yeah,
like
touch
screen,
has
a
lot
of
different
things
in
order
to
customize
the
view
at
the
dashboard
itself.
There's
also
a
couple
of
things
that
I
can
show
over
here
as
well,
so
at
the
top
there's
all
of
the
group's
dashboards
as
well
as
all
of
the
tabs
under
them.
You
can
kind
of
see
that
there's
different
colors
to
the
names
red
for
viewing
at
a
glance
like
this
has
consistently
failing
tests,
and
we
should
probably
check
some
of
those
out
versus
blue,
which
is
working.
Fine
or
black.
A
You
can
see
up
here,
means
that
there
haven't
been
test
results
in
some
period
of
time,
usually
24
hours.
The
summary
tab
also
provides
a
nice
easy
summary
of
all
of
the
tabs
under
a
dashboard
where
you
can
see
things
like
this
was
flaky
where
it
doesn't
have
consistent
failures,
but
we
noticed
some
failures
in
it.
A
A
Yeah
I
think
that's
a
lot
of
it.
I
think
the
only
thing
I
wanted
to
point
out.
My
super
apology
is
for
Mac
usage,
there's
also
something
with
I'm
going
to
click
about
results.
First,
things
like
graphs
just
may
take
a
slight
amount,
just
because
full
of
points
but
graphs
allow
you
to
view
things
like
by
default.
How
long
a
test
has
been
running,
but
also
there
are
some
other
things
where,
if
your
test
results
are
reporting,
other
metrics,
you
can
view
other
metrics
on
them.
A
So,
for
instance,
some
people
have
started
reporting
coverage
as
a
thing
for
like
their
code
coverage
results
and
also
displaying
them
on
cells.
So,
like
these
areas
have
this
much
code
coverage.
These
areas
are
like
a
hundred
percent
versus
80
percent
and
similar
things.
You
can
do
that
with
basically
any
numeric
property.
You
can
think
of
reporting
as
a
metric.
A
A
Okay,
cool
kubernetes
tests,
infra
holds
all
the
stuff
that,
like
me,
and
my
co-workers
on
or
work
on,
including
the
test
grid
config.
So
you
go
to
kubernetes,
infra
and
then
test
grade
itself.
There
is
a
bunch
of
stuff
on
the
config
itself,
as
well
as
a
handy
readme,
for
a
lot
of
the
basic
things
you
want
to
do
for
configuration
as
well
as
some
of
the
test
grade
concepts
and
some
stuff
on
more
advanced.
A
Things
like
you
need
to
specify
some
other
configurations,
because
your
dashboard
works
a
little
differently
or
you
want
things
like
email
alerts
when
things
fail
consistently.
That's
all
explained
here
and
if
it's
not
then
feel
happy
or
feel
free
to
contact
me
and
ask
questions
and
with
that
test
grade
decade
say:
I/o
has
the
community's
test
grade
itself
I'm
available
on
slack
as
Emory
at
under
the
SiC
testing
channel
generally,
where
I
lurk
very
occasionally
every
couple
days
and
then
criminais
us
in
for
a
test
score.
Folder
has
all
of
these
stuff
for
the
config.
A
You
can
feel
free
to
file
issues
against
tes,
infra
and
ping
arietis
grid
or
ping
me
directly
for
things
like.
I
want
some
improvements
to
test
grid.
It's
unfortunately
not
open
source,
except
for
the
config
right
now.
I
will
let
people
know
when
that
starts
to
change.
It's
a
thing.
We've
been
wanting
to
do
for
quite
a
while
yeah
time
for
questions
or
probably
just
do.
A
So
test
results
can
or
sorry.
The
question
was:
where
do
you
get
test
results
for
test
results?
It
kind
of
depends.
A
lot
of
the
test
results
in
tes
infra
are
from
neighboring
tool
called
prowl,
which
will
run
jobs
and
will
also
upload
those
results
into
a
google
cloud
storage
in
a
public
bucket.
So
those
ones
like
proud
will
automatically
format
them
in
a
format
that
test
grade
likes
to
read.
A
Yeah
pearl
stuff
is
also
under
test
intro.
If
you
want
to
take
a
look
Prabhas
like
their
own
config
for
running
the
jobs
there
and
then
it's
similarly
pretty
easy
like
another
couple,
three
to
four
lines
to
add
it
to
the
test
grade
config
and
see
the
results
from
their
secretary,
you
can
do
custom
results
as
well
as
long
as
it's
in
a
public
bucket
and
in
an
expected
format,
then,
which
would
be
linked
to
from
the
test
code
read
me,
then
it
should
be
able
to
be
picked
up
on
the
dashboard
as
well.