►
From YouTube: SIG i Weekly Meeting for 20221115
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
B
D
C
All
right
wait
another
minute
to
get
started,
but.
C
Actually
I
guess,
while
we're
getting
started
I'm
seeing
a
lot
of
folks
that
I
are
possibly
new
here,
so
Hey
everybody
welcome
to
the
sick
testing
meeting.
If
you
would
like
to
feel
free
to
raise
hand
and
then
unmute
and
introduce
yourself
like
name
why
you
came
to
successing
what
you're
interested
in
but
yeah
no
need
to
do
this.
If
you
don't
want
to,
but
definitely
feel
free.
A
B
A
A
E
F
Hello,
can
you
hear
me
hopefully,
yes,
awesome
I
I
have
a
new
mic.
First
time
testing
it
out.
So
just
asking
I'm
also
here
for
CS
signal,
so
Antonio
basically
erase
the
hand
to
give
a
short
overview
of
a
test
grid.
So
this
should
happen
in
this
meeting.
So
that's
why
I'm
here
excited
to
listening.
C
Awesome
welcome
Sean.
G
D
Name
is
Sean
and
I
have
been
working
with
crowd
and
test
grid,
mostly
for
the
past
couple
of
years
or
so.
I
don't
often
make
these
meetings,
but
I'm
glad
to
be
here
today.
C
Yeah
welcome
sorry,
I
felt
slightly
awkward
because
you're
on
my
team,
but.
C
All
right,
if
anybody
else
also
wants
to
go,
feel
free,
otherwise
I'm
not
seeing
Oh
a
Tony
go
for
it.
Thanks.
H
Hi
everyone,
so
my
name
is
Tony
and
I'm,
currently
contributing
in
computer
and
the
downstairs
application.
The
operator
SDK
and
I've
been
interested
in
testing
for
by
the
time,
let's
take
part
in
the
coupon
and
have
some
sessions
learning
about
the
test
grid
and
the
test
infer
and
the
I
guess
that
that
names,
the
pro
so
I'm
interested
in
how
we're
going
over
here
and
see
if
any
opportunities
that
I
can
get
into
contributions
and
yeah.
It's
very
cool
to
see
around
here.
B
C
Yeah
I
guess
really
quick
shout
out
for
that
I
think
we
will
have
a
lot
to
go
over
today.
That
is
probably
going
to
be
relevant
for
folks,
but
also
definitely
encourage
people
on
the
slack
to
just
like
asynchronously
plug
anything
that
you
are
interested
in
looking
into
just
so.
We
make
sure
that
it
gets
like
addressed
in
a
more
permanent
way
trigger
we'll
take
a
look
at
that.
C
All
right
cool
anybody
else.
C
Okay,
that
I
think
let's
go
ahead
and
dive
in
we've
got
one
two
three
four
five
folks
on
the
agenda.
I
know
Antonio's
is
probably
going
to
take
a
a
or
sorry
it's
slotted
for
30
minutes.
So
I
guess:
let's
go
ahead
and
get
started
and
I'll
try
and
keep
an
eye
on
the
time.
Sean
go
ahead.
D
Sure
hi
everyone
I
was
hoping
to
have
a
small
demonstration,
but
it
didn't
because
time
is
tricky,
but
I've
been
working
with
test
grids
playing
around
with
test
Grid's
UI
or
creating
an
open
source.
Ui
right
now,
test
grid
exists
just
as
a
serious
components.
I
heard
from
introductions
that
some
people
are
interested
in
learning
more
about
test
grid.
D
D
So
for
now,
I
will
just
start
with
here's
where
the
code
is
I,
think
there's
a
video
you
can
get
to
from
the
readme,
and
we
have
a
slack
Channel
like
there
is
a
test
grid,
slack
channel.
So
if
you
have
any
questions,
I
tend
to
hang
out
there
and
can
answer
whatever
it
may
be.
D
But
the
thing
I
just
wanted
to
bring
up
was
that
I
wrote
a
couple
of
very
simple
uis
one
in
flutter,
which
is
a
UI
tool
that
uses
Dart
as
its
language
and
kind
of
set
up
its
whole
stack,
getting
dart
working,
getting
the
produce
working
with
that
getting
a
user
interface
that
calls
test
grids
API
component
right
now,
you
just
push
a
button
and
it
displays
the
dashboards
and
then
I
wrote
the
same
thing
in
lit
components
which
is
a
web
component
technology.
D
It's
written.
All
of
that
is
written
in
typescript
and
a
lot
of
the
tools
to
set
that
up.
I
used
openwc
to
so
that
includes
that
whole
stack
things
like
husky
and
a
number
of
things
that
I've
only
kind
of
researched
and
looked
at.
D
D
D
C
Awesome
thanks,
Sean
and
yeah
I
can
also
plus
one
I'm
really
looking
forward
to
this
and
Sean's
been
doing
a
lot
of
good
work
on
getting
these
experiments
running
so
far.
C
Well,
anybody
have
any
comments
on
this
real,
quick
I
expect
most
of
this
to
take
place
offline,
but
if
there's
anything
that
people
want
to
comment
on
or
ask
questions
about,
now
is
a
good
time.
C
All
right,
I
think,
with
that
Linus
you've
got
the
floor
for
the
prowl
web
API.
I
Hello,
can
you
hear
me
great
yeah,
so
I
just
wanted
to
give
a
quick
link.
It's
in
the
meeting
notes
already
for
the
agenda
about
the
work.
That's
happening
currently
mostly
offline,
I
suppose,
because
there
have
been
no
PRS
or
anything
for
this
yet,
but
the
design
dock
is
linked
in
there
I
just
added
another
small
section.
Today
you
can
see
at
the
very
bottom
of
the
dock,
for
any
new
updates
I've
been
trying
to
track
them
at
least
the
recent
updates
to
this
doc,
but
basically
yeah.
I
So
there's
been
a
lot
of
thought
that
has
gone
into
this
and
I
think
most
of
the
or
pretty
much
all
the
questions.
At
least
the
big
questions
have
been
answered
in
the
doc.
If
there's
anything
else,
I
missed,
please
have
leave
a
comment.
There.
I
I,
don't
want
to
spend
the
next
10
minutes
talking
about
all
the
different
designs,
and
you
know
things
that
are
in
there
with
all
the
different
diagrams
and
whatnot.
So
I'll
try
to
keep
this
short
and
yeah
as
far
as
new
updates,
I
guess
in
the
dark.
I
There
is
a
link
to
the
Upstream
issue
at
the
very
top.
So
look
in
that
issue
for
any
new
new
updates.
Any
new
PRS
that
I
create
for
this
effort
should
be
like
you
know,
visible
from
there
from
that
issue.
So
yeah
I
think
that's
it
yeah.
It
should
be
open
and
accessible
to
everyone.
So
I
saw
Peter
left
some
comments,
they're
already
I'm
trying
to
adjust
them
yeah.
But
if
you
have
any
issues
like
not
being
able
to
see
the
dog
or
something,
then
please,
let
me
know.
B
C
Awesome
yeah
same
thing:
anybody
have
any
questions
or
comments
they
want
to
give
here.
J
J
C
Yeah
and
thanks
folks,
for
giving
this
a
review
also
thanks
for
all
the
work
done
on
this.
C
Cool
yeah
anything
else
from
folks
or,
if
not
I,
think
Patrick.
If
you
got
the
next
item.
K
Yeah
thanks
for
that,
I
I
do
have
a
kid
that
I
need
to
get
to
bed
so
I'll
make
it
quick
and
I
can't
attend.
Antonio's
talk,
unfortunately,
so
I
have
two
PRS
that
I
want
to
highlight
because
they
were
kind
of-
or
at
least
one
of
them
was
meant
to
go
in
before
test
freeze
and
that's
a
bigger
one
of
the
two.
That's
one
about
context
handling
in
the
end-to-end
framework.
K
It's
based
on
the
observation
that
we
had
lots
of
contacts
to
do
calls
that
didn't
properly
do
anything
and
we
have
a
new
feature
in
Ginkgo
that
allows
us
to
have
per
test
or
per
per
node.
What's
what's
as
it's
called
in
Ginkgo
timeouts
and
interrupts
user
user
visible
benefit
is
that
if
you
run
a
test
Suite
manually
and
you
press
Ctrl
C
it
properly
aborts,
whatever
it's
currently
running,
then
does
the
normal
cleanup
steps
that
belong
to
a
test
and
properly
shuts
down
the
entire
test.
K
It
bubbles
up
all
the
way
to
various
test
cases
and
and
it's
a
huge
PR
with
400
files
modified
and
it's
not
all
automatic,
so
I
really
had
to
go
in
and
look
at
individual
files
to
see
what
it's
doing
with
timeouts,
for
example,
polling
also
needs
to
abort
soon,
and
there
are
various
ways
how
current
code
does
polling
from
a
simple
for
Loop
to
the
weight.
Pol
support
functions
that
we
have
and
all
of
us
need
slide
updates,
and
all
of
that
is
in
webpr.
K
I
showed
that
to
Antonio
and
Aaron
before
code
freeze-
and
we
kind
of
decided
at
that
time
that
it
would
not
be
a
good
idea
to
put
that
in
what
everyone
else
is
trying
it's
busy
trying
to
get
their
features
finalized.
K
So
we
waited
for
for
code,
freeze
and
I
re-based
again,
because
we
were
indeed
quite
a
few
code-
conflicts
that
were
not
hard
to
con
hard
to
resolve,
but
still
a
few
and
I
I
kind
of
had
hope
that
we
can
now
put
it
in
before
test
freeze.
The
advantage
of
doing
it
now
would
be
that
when
people
start
to
develop,
CR
features
seriously
for
127.
K
All
of
these
changes
will
be
in
place.
So
when
the
code
for
the
branch
opens
again
for
127,
we
will
not
have
a
large
number
of
conflicting
PRS
that
all
try
to
go
in
at
the
same
time,
but
rather
prepare
for
for
future
development.
Now,
when
it's
relatively
quiet
risk
is,
of
course,
that
we
break
something
this
can't
be
ruled
out,
and
that's
where
we
are
at
the
moment.
It's
currently
in
a
merchable
state,
but
it's
risky
in
us
in.
L
A
way
it,
it
actually
needs
a
re-rease
again
I'm
a
little
concerned
about
a
14
000
line,
PR
this
late
in
the
release.
How
confident
are
we
that
this
isn't
going
to
break
something
and
then
be
a
massive
revert.
K
There
were
a
few
breakages
where
I
did
things
wrong,
like
adding
a
context
parameter
to
a
context,
callback
that
that
immediately
broke
a
test
Suite,
it
didn't
even
start
other
things
like
using
the
context
from
a
preparer
called
or
a
long-running
background
function
is
another
mistake
that
showed
up.
That
only
appears
when
there
is
actually
a
test
that
exercises
that
code,
but
it
does
show
up
reliably.
So
if
we
merge
it
and
some
test
starts
failing,
it
will
be
fairly
obvious.
What
what
it?
G
My
concern
is
that
that
the
with
the
Cure
data
that
we
accurate
and
another
feature
the
API
server
tracing
and
you
got
the
hit
for
for
that,
they
added
contest
to,
and
we
it
took
two
or
three
days
and
other
people
to
notice
that
the
context
was
the
problem.
That's
my
my
concern.
I
mean:
did
they
ever
wasn't
surface
as
as
clean
as
we
thought?
I
do
remember,
but
today,
yeah
and
and
that's
my
concept
here,
we
and
it
happened
with
sun
cubelet
contest
to
they
prompted
all
the
countries
and
that's
why
I
missed
that.
G
K
L
I
see
that
well
from
our
point
of
view,
I
really
sympathize
with
the
whole
like
we
want
to
get
this
in
without
rebase
hell
and
making
sure
we
catch
everyone
I
think
we
can
also
do
that
if
we
do
it
at
the
very
beginning
of
like
post
freeze
and
from
my
point
of
view,
the
purpose
of
test
freeze
is
to
allow
people
to
finish
merging
the
tests
for
their
features
not
to
make
sweeping
changes
to
the
tests
as
Sig
testing,
like
that's
closer
to
Landing
a
feature,
I
I
think
that
we
want
this
time
window
for
people
to
make
sure
that
the
the
release
is
well
tested
and
we
have
snuck
in
other
changes
that
are
like.
L
K
Well,
that's
fair,
I
I
can
live
with
that.
I
can
certainly
rebase
again.
If,
if
something
sneaks
in
before
me,
when
the
when
the
branch
opens,
my
biggest
concern
is
indeed
that
then
we
will
have
more
people
using
the
current
API
and
that
they
will
be
affected,
but
I
guess
they
can
just
rebase.
It's
not
too
hard
to
figure
out
that
a
function,
change
needs
a
context,
parameter
and
and
then
figure
out
how
to
how
to
add
that
to
your
test
boy.
But
it's
fine
I
just
wanted.
K
G
K
B
L
Can
you
like
drop
a
ping
to
all
of
us
in
in
slack
like
once
the
once
a
branch
opens,
so
we
don't
forget.
K
Well,
prioritize
yeah:
well,
someone
can
approve
it
now
and
then
I
just
need
an
lgdm.
If,
if
it
needs
a
rebase,
we
can
do
Revenue
now,
if,
if
or
well
you
would
have
to
trust
me
to
do
the
rebase
properly
but
I
think
that's
the
minor.
L
K
L
I
think,
let's
follow
up
with
that
I
think
as
well.
That
seems
like
an
oversight
at
this
point,
good.
K
No
well
like,
if
you
want
to
know
about
it,
I
can
I
can
certainly,
but
it's
it's
about
how
Ginkgo
formats
the
text
that
it
puts
into
what
used
to
be
with
a
standard
error
of
a
test
case
that
only
had
the
real
log
messages.
What
you,
what
we
write
to
Ginkgo
writer
and
then
we
had
special
code
to
get
the
failure
message
in
at
the
right
place
where
it
occurred
and
the
same
for
for
skipping,
but
there's
more.
K
That
goes
on
Ginkgo
itself
emits
events
or
encounters
things
like
executing
a
certain
node
certain
code,
certain
callback
and
with
the
the
timeline
feature
as
it's
called
in
Ginkgo.
We
get
all
of
that
in
chronological
order
into
the
text
text
output
for
a
test
which
is
a
lot
more
informative
than
it
used
to
be,
and
we
could
remove
and
we
and
I
removed,
basically
all
our
workarounds,
but
try
to
get
something
like
it
have
working
and
it's
it's
I
think
a
lot,
not
a
lot
nicer,
yeah
I!
K
L
Well,
these
both
sound
like
peers
that
we
we
definitely
want.
Thank
you
for
working
on
them.
K
C
Awesome
yeah
plus
one
thank
you
for
all
the
work
on
this
Patrick.
This
looks
awesome
all
right.
We
have
two
more
items.
I
want
to
ask
really
quick
actually
for
one
more
possible
agenda:
Shuffle
Adam.
How
long
do
you
expect
your
item
to
take
if
it's
relatively
short,
I
think
we
could
actually
tackle
it
now
and
then
have
Antonio
just
freely
take
up
the
rest
of
the
meeting.
G
C
It
I
just
think
the
having
the
shorter
thing
go.
First
will
be
a
little
bit
Freer
for
response.
C
Okay,
cool
I've,
moved
it
around
Adam,
go
ahead
and
take
it.
E
E
E
E
That
possible
bug
reports
kind
of
my
goal:
okay,
if
I
break
up
too
much
okay,
if
I
break
up
too
much
feel
free
to
just
stop
me
to
not
waste
everyone's
time.
Great
nice.
Thank
you.
So
the
long
story
is
that
I
have
been
been
working
with
the
the
cncf
on
writing
a
bunch
of
fossils
for
kubernetes,
and
they
are
currently
in
the
cncf
repository
and
are
running
on
OSS.
E
First,
it
would
be
nice
to
get
those
fossils
merged
into
the
kubernetes
repository
so
that
other
developers
can
get
use
of
them
or
make
modifications.
When
the
code
changes
and
I
spoke
with
some
people
at
kubecon
about
how
to
approach
this
and
the
we
kind
of
just
yeah,
the
consensus
was
that
it
would
make
sense
to
take
a
fossa
at
a
time
to
kind
of
yeah.
Do
do
small.
D
I
think
you're
breaking
up
again.
We,
the
last
thing
we
caught
was
just
the
best
way
to
do.
It
is
to
do
small,
and
that
was
it.
C
C
Yeah
I
I
think
following
up
in
slack,
sounds
like
a
good
idea,
especially
because
I
I
don't
think
that
this
amount
of
time
is
probably
going
to
be
enough
to
go
into
a
significant
discussion.
I
did
see
a
couple
hands
from
folks.
If
you
want
to
give
some
good
comments,
feel
free
yeah.
G
E
G
L
I
think
we
should
probably
move
this
to
an
async
this
more
resilient
to.
G
Okay,
I
just
wanted
to
read
the
contest
because
I
was
checking
on
this
and
and
I
was
checking
on
the
test
on
that
infra
where
were
failing
or
where
I
really
don't
understand
how
that
works.
That's
a
good
thing
and
they
are
failing
and
I,
don't
think
that
we
should
merge
a
tests
that
are
not
going
to
be
green
because
it's
very
noisy.
But
this
is
from
my
ignorance
on
this
fast
intestines
I,
don't
know
if
the
test
is
going
to
succeed
or
not.
L
So
I
I
started
a
thread
in
the
Sig
testing
slack
to
discuss
further
I.
Think,
let's,
like
whatever
comments,
I
haven't,
had
or
whatever,
let's
start
a
conversation
there
or
or
if
not
there,
the
mailing,
the
somewhere,
that
we
can
back
and
forth
async
and
a
little
more
resilient
to
connection
issues.
E
Yes,
I
would
be
able
to
to
do
this
over
slack
and
feel
free
to
add
me
to
as
well
in
case
it's
private.
C
Much
yeah,
sorry
Antonio.
Your
keyboard
is
audible,
yeah.
G
Okay,
that's
this
is
the
other
day
we
were
I
found
in
an
issue
in
the
CI
signal
Channel
they
reported
on
this
one
we
were
discussing
during
this
discussion.
We
said
well,
it
would
be
good
if
we
can
have
some
hints
and
tips
on
how
to
detect
these
problems,
how
to
report
to
make
the
round
trip
easier
for
the
fix,
so
the
the
it
got,
assign
it
quickly
and
and
that's
trigger
it
is
the
initiative
to
present
I
have
some
slices
and
you
give
me
a
presenter
presenter
powers.
C
Yeah,
all
right,
you
should
be
able
to
know
Antonio.
G
G
How
can
I
say
internals
that's
doing
the
development
and
I
feel
that
sometimes
we
have
a
gap
in
communication,
so
people
is
developing,
don't
realize
that
it's
breaking
the
the
external
kubernetes
and
the
standard
is
not
is
that
that
is
not
able
to
to
give
feedback
to
the
people
that
is
developing
to
fix
it
so
I'm
reducing
this
this
slide
a
lot.
This
is
more
or
less
how
kubernetes
testing
works?
Okay,
so
you
can
identify
these
these
steps
in
the
pull
requests
or
in
the
jobs.
G
There
are
verified
jobs,
unit
tests,
jobs,
integration,
test,
jobs,
Italy
jobs
and
scale
testing
job.
Usually
the
people
in
in
release
have
that's
for
withhold
these
things,
but
they
need
to
understand
in
which
Ledger
is
happening
when
they
see
a
Purify
failure
is
because
something
in
the
code
is
not
right.
There
are
LinkedIn
linters
checking
that
I,
don't
know
you
don't
have
typos
this
kind
of
things
when
they
have
annuities
test
failure.
G
They
need
to
know
that
this
is
very
focused
on
a
special
component,
so
usually,
if
they
reach
out
directly
to
the
c
or
to
the
owner
people
in
that
file,
they
are
going
to
have
a
quicker
feedback.
Integration
there
is
is
more
widescope,
but
this
is
still
very
focused
on
on
scientific
or
company.
Italy
is
about
everything,
so
this
is
the
more
harder
to
to
understand
to
to
whom
to
reach
out,
because
the
job
maybe
sign
network,
but
the
failure
is
in
the
schedule,
and
this
are
more
complicated
for
the
scale
testing.
G
You
just
go
to
sleep
like
this
to
see
this
scalability,
because
I
think
that
are
the
only
ones
handling
those
jobs.
Do
you
have
here
some
links?
If
you
want
to
to
understand
better
what
is
going
on
in
which
part
and
how
they
run?
Okay,
this
is
more
for
supporting
and
documentation,
and
this
is
how
how
it
works.
Okay,
we
have
a
kubernetes
when,
when
this
kubernetes
has
all
the
steps
right
and
now
we
need
to
run
this
test
when
this
test
run.
G
So
we
just
submit
the
chains
a
pull
request
or
something
you
you
run
in
prism,
so
presume
it
has
only
a
subset
of
jobs,
some
specific
jobs
in
Italy,
some
specific,
all
the
integration,
all
the
linters
and
everything.
But
it's
not
testing
everything.
Okay,
the
summit
has
to
run
quickly
and
it
may
happen
that
some
code
bit
merge
and
some
tests
start
to
fail
after
merge
because
he's
not
running
in
prison
for
Summit
jobs.
G
I,
don't
think
that
they're
relevant,
but
just
know
that,
once
the
code
merge,
it
can
trigger
something
to
build
an
image
or
something
like
that.
Typically,
it's
used
for
promoting
the
E3
test
images
or
the
conformance
index,
and
then
periodic
periodic
is
the
jobs
that
the
people
in
signal
and
release
are
more
familiar
with,
because
these
are
triggered
a
every
periodically.
B
G
And
they
are
a
lot
of
different
jobs
that
test
different
things
in
this
in
different
environment.
So
right
now,
this
is
how
the
test
is
executed
in
kubernetes,
okay
and
to
analyze
everything
we
have
different
reporting
tools.
So
these
are
the
most
important
test.
I
think
that
everybody
is
using
to
test
with,
but
what
I
found
out
is
that
teskey
has
a
lot
of
features
that
nobody
knows
and,
and
we
should
be
able
to
promote
them,
because
test
kit
allows
you
to
filter.
Let's
give
it
allows
you
to
sort
by
fleakness
test.
G
G
How
can
I
go
back
the
one
in
the
in
the
bottom
right
corner,
so
that
opens
a
form
where
you
can
look
for
job
failures.
It's
like
a
kind
of
a
Google
search
for
for
this
fails
and
that's
very
useful
for
for,
investigating
and
start
to
see,
Trends
and
analyze,
and
the
third
one
that
I
that's
important
is
the
pro
the
pro
the
the
condition
of
pull
requests
or
in
the
the
pro.
What's
the
name.
Youtube
allows
you
to
to
go
to
a
dashboard
that
gives
you
the
history
of
the
pull
request.
G
Okay,
so
you
can
understand,
if
you
you
now
to
kind
of
identify
the
requests
that
created
a
problem,
you
can
go
there
and
you
can
see
the
history
of
the
put
request.
I
mean
we
just
see
that
it
failed
it
six
times
out
of
seven
in
in
the
same
job.
Okay,
that's
something
that
is
important
when
you
try
to
analyze
things.
The
last
tool
is
the
first
task
force.
That's
used
for
scalability,
that's
much
more
advanced.
We
don't
usually
need
to
do
that
and
scalability
team
is,
is
usually
pricing.
G
The
problems
when,
when
we
have
packs
on
a
scale,
okay-
and
these
are
some
tips
that
I
collected
the
the
for
finding
the
root
cause.
Imagine
you
identify
something
in
discrete
something
started
to
fail
and
you
need
to
do
some
little
investigation
to
find
the
root
cause
and
what's
important
is
correlation.
Doesn't
does
not
imply
causality,
okay,
but
it's
the
way
that
we
have
to
start.
We
know
that
something
started
to
fail
on
Monday
morning.
G
Okay,
so
we
start
to
correlate
what
happens
on
Monday
morning,
but
that
the
what
they
want
to
say
is
we
need
to
find
the
root
cause.
I
mean
we
cannot
because
it
started
to
happen
on
Monday.
It
doesn't
mean
that
in
March
on
Saturday,
the
problem
that
kind
of
things
happen,
so
don't
trust
anything
or
anybody
always
verify
everything.
I
mean
you
have
a
hypothesis.
You
think
that
this
started
on
Monday
that
this
coming,
maybe
the
corporate
just
check
what
this
committee
is
doing,
verify,
try
to
reproduce
and
use
the
tools
I.
G
We
went
through
the
tools
before
we
have
GitHub
issues
we
have
test
grid.
We
have
the
the
tools
that
I
I,
so
it
before.
You
need
to
use
these
tools
because
they
are
going
to
give
you
the
sources
for
for
your
hypothesis.
So
you
can
correlate
a
formulate,
your
hypothesis
demonstrate
it
and
find
the
root
cause
and
always
don't
hesitate
to
ask,
but
when
you
are
asking
think
that
the
other
person
is
going
to
be
busy,
so
they
are
not
going
to
have
time
to
walk
with
you
through
all
the
scenario.
G
Just
call
to
ask
concise
question:
is
this
happened
on
Monday
at
this
Market
before
this
other
thing,
I
checked
these
logs,
so
the
other
person
has
the
whole
contest,
and
then
he
can
tell
you
what
I
don't
know,
but
I
know
who
could
be,
then
you
have
a
lead
or
I
was
doing
something
that
time.
Let
me
check-
and
this
is
these-
are
very
good
tips
when
you
start
to
investigate
okay,
how
to
find
the
root
cause
and
then
imagine
you
do
all
the
investigation.
G
You
couldn't
find
anything
you
ask
it
and-
and
you
don't
need
to
to
sweat
it
I
mean
it's.
This
is
not.
We
don't
need
everybody
to
be
able
to
solve
everything
this
community
and
after
spending
some
time
you
say,
I
I
cannot
advance
and
block
just
report.
This
and
Reporting
is
very
important
because
reporting
is
is
asynchronous.
Communication
in
asynchronous
communication
I,
don't
know
right
now
how
many
issues
do
we
have
opened,
but
there
are
a
lot
of
issues
in
kubernetes
and
it's
crazy.
G
Other
thing
important
is:
we
don't
need
a
very
good
narrative
or
grammar
right.
This
is
a
worldwide
Community.
Language
is
a
huge
part.
I
mean
we
are
from
different
countries
with
different
ways
of
speaking.
So
don't
try
to
to
add
a
lot
of
literature
just
be
clear
and
concise
is.
What
is
the
problem?
G
If
you
want,
you
can
formally
hypothesis
or
suggestion,
because
maybe
nobody's
going
to
take
on
the
issue,
but
maybe
somebody
can
give
you
a
hint
and
say
well:
have
you
tried
to
do
this
and
then
maybe
you
open
this
and
you
are
the
same
person
that
are
going
to
cross
it
because
you
have
some
guidance
and
during
that
process
you
are
learning
it's
kind
of
of
mentorship
thing.
G
So
I
want
to
to
use
two
examples,
so
the
first
sample
was
the
the
one
that
that
started
this
initiative
to
present
this
okay
do
I
cannot
pronounce
forgive
me,
so
you
have
this
nice
channels
that
I
always
follow,
and
you
report
the
issues-
and
you
are
very
good
at
this,
because
you
are
faster
than
my
notification
money.
So
I
I
really
like
and
appreciate
your
work.
G
To
solve,
because
this
has
a
clear
trigger.
Okay,
you
see
that
since
one
point
some
specific
test
started
to
fail.
Okay,
then,
what
they
did,
let
me
screens
here:
let's:
let's
do
these
slides?
Okay,
let
me
insert
this
step
okay.
So
now
you
see
my
tab,
so
this
is
this
is
the
the
acting
the
trigger
and
everything
what
I
did
is
I
clicking
in
in
the
tab?
G
Okay
and
now
I'm
going
to
do
this
with
and
you
see
well,
you
see
that
there
is
always
the
same
pattern
of
tests
that
are
failing
and
then
I
can
scroll
to
the
start
to
scroll
back
scroll
back
and
I
can
see.
Well,
there
is
a
point
that
everything
is
going.
I
pointed
everything
right
and
two
important
is
here:
I
have
this
commit
and
I
have
this
other
copy?
Okay,
this
is
very
important.
So
we
have
has
a
really
nice.
G
If
you
know
gate
you
can
play
Bitcoin,
but
usually
you
don't
need
to
go
to
the
command
line
or
everything.
This
Hub
has
a
tool
that
allows
you
to
compare
between
these
two
comments.
Okay,
I
have
this
here
for
simplification.
You
see
the
URL
is
compare
this
last
well,
you
can
Google
it
and
it
gives
you
how
to
compare
between
two
commits
and
then
you
have
here
all
the
changes,
the
all
the
files
that
has
changed
during
this
time.
Then
I
go
to
to
my
control.
G
There
are
several
C
CSI
passing
some
balloons
that
I
don't
know
what
they
are
doing,
but
I
don't
know
that
they
change
it
here.
Okay
and
then
you
can
see
more
test
storage
CSI,
whatever
okay,
then
I
can
click
in
one
of
them
and
and
I
can
okay,
I
wanted
to
go
to
the
change?
Okay,
I!
Don't
know
how
to
do
it
from
here.
G
Okay,
this
is
what
they
want
and
now
here
those
are
the
files,
and
here
is
the
history.
Okay,
I
work
there
and
I
can
see
CSI
I,
just
look
for
CSI.
Just
think
like
you
are
Googling
something
and
trying
to
find
a
travel
to
some
place
or
whatever.
This
is
the
same.
Just
apply
the
same
logic,
look
for
peoples
and
sets
and
I
can
see.
Mark
CSI
driver
okay.
This
is
changing
test.
They
may
be
related
to
the
other.
One
and
I
go
to
the
pull
request
from
here.
G
G
This
is
very
welcoming.
Okay,
you
see
so
the
person
that
comes
here-
oh
sorry,
I'm,
not
selling
what
they
want.
This
is
is
really
well
commented,
so
this
is
here.
This
is
plenty.
This
job
is
failing.
Some
more
tests
in
city
stories
are
failing.
I
can
see
without
doing
anything
that
there
is
a
this
problem
here,
since
this
day
and
I
have
a
link
to
go
there
and
do
this
same
thing
that
I
did
before
and
you
have
the
reason
for
failure.
That's
about
sense.
G
Some
tests
I
did
that
investigation
and
paste
paste
it
here
and
I
tag
it.
The
two
persons
that
were
reviewing
and
everything
on
that
tier
and
it
took
here.
You
cannot
see
it
took
25
minutes
to
the
person
to
acknowledge
that
and
he
submit
appear
the
same
day.
Okay,
it
took
some
more
time
in
in
good
merch,
but
you
see
that
in
one
two
hours
we
were
able
to
fix
a
an
important
failure
that
was
affecting
all
the
release.
Okay,.
G
These
are
the
I
say,
quoted
one.
This
is
one,
so
let's
go
for
the
the
for
the
hardest
one,
and
this
is
are
more
complicated.
Why?
Because
you
see
things
start
failing,
but
to
start
failing
randomly
okay
and
you
can
see
the
trend,
there
are
some
jobs
that
are
flicking
and
okay.
Let
me
I
want
to
show
you
this
feature
of
this
video
because
I
really
love
it.
Let
me
go
there.
G
G
Look
when
you
go
to
test
screen,
you
have
this
option
that
says:
Circle
flick,
yes,
okay,
that
makes
you
go
back.
This
runs
many
times,
so
we
kind
of
we
don't
have
the
whole
history
again.
Well,
we
don't
have
the
cool
history,
but
the
point
is
we
have
a
lot
of
awesome
tools
with
a
lot
of
awesome
features
that
we
should
use
more
when
when
you
go
there,
you
can
Aggregate,
and
you
can
see
that
it
is
a
trend
on
on
fleckiness
and
they
and
it's
not
affecting
the
same
test.
G
G
It
has
so
many
times,
okay,
one
second,
okay.
How
did
I
solve
this
problem?
Okay,
I?
The
thing
is:
I'm
reviewing
I
have
a
lot
of
reviews
and
mentally.
You
start
to
create
some
kind
of
of
training
that
not
useful
this
last
five
six
PR.
It
started
to
fail
in
on
this
job
that
don't
used
to
fail.
Okay,
and
that's
why
I
ended
going
to
this
test
grade.
G
I
went
to
the
discrete
I
started
it
by
flakiness,
as
I
said,
and
as
you've
seen
the
screenshot.
You
can
see
that
actually
it
started
to
fail
more
during
the
last
day.
Okay,
so
what
you
need
to
do?
You
can
do
start
looking
for
this
and
you
can
see
Trends.
Okay,
you
see
this.
You
can
add
the
job
here
and
start
filtering
and
therefore,
let's
see
if
it's
the
network,
only
let's
see
I,
don't
know
whatever.
G
G
G
G
Okay,
I
should
be
more
prepared.
This
is
some
Panic
somewhere
and
I
cannot
find
it.
I,
don't
know
if
I'm
finding
the
right
file
abroad
as
well.
Sorry
for
this
okay,
so
the
point
is
the
middle
level.
Is
the
freaking
days
for
the
faking
test
you
need
to
to
to
follow
Jordan.
Did
it
made
a
great
presentation
here
into
six
testing
two
years
ago
or
something
like
that?
He
he
delivered
this.
G
He
recorded
it
and
he
has
this
kissed
with
all
the
tricky
things
that
you
have
to
do
to
the
freight
test
and
what
the
point
is
for
the
freaking
attack
like
that.
You
need
to
understand
how
the
whole
system
works
and
it's
something
that
takes
a
lot
of
time
to
need
to
know
all
the
components
you
need
to
know
how
they
are
related
and
you
need
to
be
lucky,
because
there
are
those
things
that
are
lucky
and
it
takes
a
lot
of
time.
G
So
don't
go
to
I'm,
going
to
the
flight
test
and
spend
two
hours
and
thinking
that
you
are
going
to
solve
it
in
three
hours.
Maybe
you
can,
but
it
usually
takes
a
lot
of
time,
and
sometimes
you
are
not
able
to
to
sorry
so
don't
hesitate
to
ask
for
questions
for
help
and
good
luck,
I
mean
the
more
people
that
we
can
test
the
better.
G
G
C
L
Thank
you
that
was
really
great
if
folks
on
the
reviews
are
interested
in
coming
another
time,
a
little
more
time,
prep
I
I'd
like
to
also
add
somewhere
a
dive
into
sort
of
just
breaking
down
a
specific
test
result
and
talking
about
some
of
the
specific
entries
in
like
test
grid
as
well.
There's
a
lot
on
the
subject
nice
Tony
mentioned.
We
have
a
lot
of
tooling.
L
C
Yeah
I
didn't
see
what
order
so
I'm
gonna
go
in
the
order
of
my
screen
of
Lana.
N
N
Noise,
like
I
would
expect
the
system
to
be
more
a
bit
more
intelligent
where
it
says:
okay,
it
failed
on
some
day
after
that
I've
had
like
four
or
five
successful
runs,
and
it
should
show
me
as
a
pass
rather
than
play
key,
so
I
just
wanted
to
understand.
What's
the
criteria
based
on
which
the
tests
start
reporting
is
flaky,
and
can
that
be
configured
to
be
a
bit
more
intelligent.
D
Hey
I
think
I
can
feel
that
in
general,
a
lot
of
it
has
to
do
with
a
test
grid
setting
called
num
columns
recent
in
order
to
determine
flakiness.
It
looks
over
the
number
of
tests
equal
to
num
columns.
Recent
I
think
the
default
is
five
and
determines
from
those
flakiness.
So
if
in
within
those
five,
there
is
a
test
that
is
passing
anti-test,
that
is
failing,
then
the
whole
thing
is
summarized
as
flaky.
D
C
I'll
add
really
quick.
That
there's
also
been
some
recent
updates
here,
because
or
were
that,
like
flaky,
for
a
entire
tab,
for
instance,
is
a
little
bit
confusing.
So
I
think
we
can
look
forward
to
like
a
couple
things
to
help
that
in
future,
Brian.
M
Hey
I
got
a
question
on
behalf
of
new
people.
I
know
that
going
to
test
grid.
How
do
you
find
what
you
want
in
that
thing?
It's
crazy
and
once
I
find
something
that
I
want
I'm
afraid
to
leave
that
page,
because
I'll
never
find
it
again.
So
maybe
a
quick
lesson.
I,
don't
know
today
or
some
other
day
of
how
do
we
start
from
that
that
beginning
and
get
to
where
we
need
to
be.
L
D
He
yeah
something
like
that
would
probably
be
very
useful.
I've
been
looking
at
test
grid
for
some
years
now
and
I'm
deeply
cursed
with
knowledge.
D
So
it's
tough
to
say
exactly
what
information
is
useful
to
a
new
person
and
what
is
me
just
spouting
Madness
about
the
the
way
that
test
grid
is
my
top
tip
for
new
people,
though,
is
that
if
you
are
looking
in
Spyglass
and
like
because
your
prow
test
failed,
there
is
usually
if
there
is
a
corresponding
test
grid,
there
should
be
a
link
in
the
top
right
that
says
test
grid.
D
D
Fair
I
think
Ben
just
posted
a
link
because
come
on.
Let
me
look
at
the
link.
Yeah
Ben
posted
a
link
because
it's
easier
to
show
people.
L
D
G
D
All
right,
oh
it
paused
my
audio
strange.
This
is
this,
is
it
this
is
Spyglass.
This
is
what
I'm
talking
about.
This
is
the
link.
It's
a
good
link.
It
takes
you
to
test
grid.
D
That's
my
tip.
Thank
you.
L
Oh
all
of
that
super
briefly,
there
is
a
search
bar
at
the
top,
which
is
somewhat
helpful,
but
it
because
the
test
grid
dashboards
sometimes
have
more
descriptive
names
than
prow
permits,
they're,
not
necessarily
one
to
one.
So
the
the
tip
that
Sean
just
gave
is
super
useful
for
going
from
I
know
which
job
I
want
to
I
want
to
see
it
in
test
grid.
L
Once
you
know
that
you
can
find
other
dashboards,
the
other
thing
is
they're
supposed
to
be
a
top
level
hierarchy.
It's
gotten
a
little
bit
harder
to
navigate
because
it
wasn't
only
used
for
kubernetes,
but
you
should
be
able
to
like
look
for
a
Sig
and
find
things
and
in
most
cases,
if
you're
not
looking
for
a
specific
six
results.
L
You're
looking
for
like
under
Sig
release,
the
release
team
has
the
release,
blocking
release
and
forming
dashboards
and
so
on,
and
we
have
some
documentation
about
what
the
expectations
are
to
to
get
into
those,
but
because
it's
part
of
the
release
process.
That's
actually
over
in
Sig
releases
docs
and
like
some
suggesting
folks,
wrote
it,
but
it
it's
owned
by
Sig
release,
because
it's
part
of
their
release
process.
C
All
right,
yeah
any
other
comments
or
questions
from
folks.
C
Foreign
I
think
with
that
we
can
call
close
so
thanks
everybody
for
coming,
and
also
thanks,
everybody
for
letting
us
around
smoothly
with
subtract
agenda.
Looking
forward
to
this
in
future,
yeah
take
care.