►
From YouTube: Octant Community Meeting - March 17th, 2021
Description
Octant community meeting is held weekly. We discuss and talk about the current state and future of Octant, demo upcoming features and releases, and preview new ideas we are considering for Octant.
Meeting agenda: https://hackmd.io/CzaPxtmXT_SW8nEpdwvGzw?view
A
Hello
and
welcome
to
the
march
17th
octane
community
meeting
today,
we've
got
a
few
things
to
talk
about
we're
going
to
talk
about
the
dynamic
cache
refactor
that
we
mentioned
last
week
and
is
now
officially
like
up
for
review
this
week.
Some
updates
for
our
example
plugins
some
coverage
on
a
feature
to
allow
metadata
to
be
sent
into
plugins
and
a
resource
viewer
preview
so
which
that's
exciting.
So
with
that
I
will,
I
will
start
with
the
much
faster
dynamic
cache.
A
So
there
was
in
previous
community
meetings.
We
had
discussed
some
cpu
things
and
in
an
effort
to
kind
of
address
some
cpu
stuff,
but
more
generally
just
speed
things
up.
I
felt,
like
things
were,
have
always
been
a
little
slow
and,
like
generally,
the
team
like
has
felt
that,
like
we
would
watch
demos
of
people,
use
octane
and
go
oh
man
that
seems
slow.
So
I
was,
I
was
looking
at
it
with
that
lens
of,
like
maybe
I
can
get
some
cpu
increase.
A
Probably
I
can
get
some
speed
increase
and
I
did
not
get
any
cpu
improvements,
but
I
did
not
make
it
worse
and
so
the
that
is
to
me.
That
is
a
net
win,
because
the
end
result
is
that,
on
average
we're
seeing
a
30
to
50
x,
improvement
in
load
times
for
resources
within
octane,
which
is
which
is
it
still?
It
still
blows
my
mind
a
little
bit
like
I
don't
I
don't
know.
A
Looking
at
the
old
code,
I
spent
some
time
looking
at
the
old
code,
trying
to
figure
out
what
we
were
doing
so
wrong
before,
like
what
I
was
doing
so
wrong,
but
I
wrote
the
old
code,
so
I
was
like
how
did
I
get
this
so
wrong
that
it
is
so
slow
compared
to
the
refactor?
I
did,
but
that's
just
how
it
goes
sometimes
so
I
still
have
not
yet
fully.
A
I
haven't
fully
figured
out
why
the
old
version
was
so
slow.
I
do
have
some
ideas
of
why
the
new
version
is
much
faster.
That
said,
the
there's
a
pr
up
right
now,
2144.,
it's
got
a
couple
outstanding
to
do
tasks.
A
I
also
called
out
just
some
general
notes
around
some
implications
for
this
change.
Nothing
that'll,
really
impact
folks.
There
is
a
slight
there's,
a
slight
change
to
expectations
of
of
what
you
have
access
to
with
verbs
in
octant,
meaning
we'll
we'll
now
expect
you
to
have
watch
to
a
resource
and
if
you
don't
have
watchful
resource,
there's
no
longer.
A
This
attempt
to
fall
back
to
like
this
list
get
polling
semantic,
and
I,
I
suspect
that
that
will
not
impact
many
folks
and
in
fact
it
will
make
a
lot
of
people's
experiences
better,
because,
generally,
what
we're?
Seeing
with
people
who
are
working
in
our
back
restricted
environments
or
where
they
only
have
access
to
a
certain
namespace?
A
Is
they
have
access
to
all
of
the
verbs
for
a
set
of
resources?
Maybe
not
like
the
ability
to
create,
but
like
they
get
the
standard
set
of
list
get
watch?
And
so
following
trying
this
like
to
be
smart
and
like
fall
back
to
list
and
get
when
we
fail
to
watch
it
was.
I
think
I
think
there
was
times
where
that
just
wasn't
working
correctly
like
the
watch
was
starting
and
it
just
hadn't
synced.
A
Yet
like
we
didn't
wait
a
couple
milliseconds
longer
for
the
cache
to
sync,
so
we
went
ahead
and
did
a
manual
list
call
which
is
like
anyway.
I
think
I
think
we
were.
We
were
overloading
some
some
api
calls
that
we
didn't
need
to
and
we
were
like
trying
to
trying
to
be
like
too
smart
and
like
like
too
fancy.
They
were
just
like.
Oh
we'll
do
everything
and
make
sure
that
you
never
see
a
failure.
A
A
If
you
can,
we
we've
tested
it
where
it's
like.
I
am
in
a
namespace
and
all
I
can
do
is
watch
pods
and
if
you
go
to
the
pod
view,
you
can
see
and
list
the
pods.
There
is
some
breakage
that
exists
today.
Even
so,
it's
not
it's
not
a
regression
where,
if
like
you,
only
have
the
ability
to
list
and
get
pods
and
you
go
try
to
view
the
details.
A
Sometimes
that
details
view
will
break
because
we're
trying
to
like
render
events-
and
you
might
not
have
permission
to
events
so
there's
going
to
be
an
effort
to
trap
errors
where
they
happen
and
then
respond
appropriately
and
let
the
components
kind
of
render
their
own
error
messages.
So,
instead
of
the
pod
summary
being
this
gross
like.
A
Oh,
you
have
you
can't
watch
events
or
you
can't
see.
Events
it'll
render
all
the
pieces
for
the
pod
and
then
the
events
area
that
table
will
say
you
don't
have
the
ability
to
watch
events.
A
If
you
want
to
populate
this
table,
you
know,
add
that
permission
and
so
it'll
be
a
much
much
better
experience
for
users
and
just
generally
will
allow
us
to
handle
those
error
cases
a
lot
better.
This
also
fixes
the
issues
of
inconsistency
with
the
api.
A
I
wrote
that
too,
it's
very
embarrassing
when
I
say
it
out
loud,
but
it's
fixed
in
this,
and
so
I
was
going
to
just
show
a
short
little
demo
stop
of
this
running
under
electron.
A
A
So
I
launched
this
and
then
shared
it,
so
I
could
actually
attach
to
the
window
and
you
can
see
the
it
was
already
populated
now.
This
is
just
a
mini
cube
with
very
little
stuff
in
it,
but
you
can
see,
as
I
switch
to
cube
system
the
quickness
in
which,
if
you've
used
octane
before
the
speed
at
which
this
change
happens,
is
it
it's
it's
great.
A
It
is
really
great,
I'm
very
pleased
with
it
and
then,
as
you
as
you
drill
down
into
the
various
items
right,
they
all
load
at
that
same
speed,
where
the
the
limitation
on
how
fast
things
are.
Loading
now
is
actually
on
our
internal
worker
event:
loop.
A
If
you
catch
it
at
the
beginning
of
the
loop
you'll,
have
to
wait
the
entire
one
second
to
get
the
next
content
response,
but
if
you
catch
it
at
the
right
point,
you
you
like
sometimes
the
the
pages
load
almost
instantly,
so
it's
it's
pretty
great
and
then
switching
if
you
ever
switched
context
in
octant
before
you
know
how
long
that
can
take,
and
so
here
I've
just
switched
to
context.
It's
pulled
crds.
Now
I
have
all
deployments.
A
I
have
pods
and
then
I
can
start
to
browse
these
things
just
right
away,
and
you
can
see
my
crd
list
is
here.
I
can
go
look
at
my
crds
and
then
I
can
switch
right
back
to
the
mini
cube
and
you
see
this
is.
This
is
part
of
that
graceful
like
hey,
you
switched
namespaces.
You
were
looking
at
a
custom
resource
that
custom
resource
doesn't
exist
in
this
namespace,
so
we'll
just
give
you
a
knockdown
page.
A
That
generally
didn't
happen
very
well
before,
because
it
would
like
try
to
do
some
error
or
or
it
would
return
a
nil
response
because
you
like
it,
couldn't
find
it
so
now,
it's
I
don't
think,
there's
anything
in
this
namespace
but
switch
to
cube
system
and
so
yeah
if
you've
used
octane
previously
and
you're
seeing
this
now,
you
know
why
this
is
a
great
improvement.
If
you've
not
used
doctrine,
it's
your
first
time
seeing
it.
A
This
is
probably
how
you
expect
it
to
work
so
very
pleased
with
the
outcome
of
this,
and
there
was
one
there
was
one
issue
that
came
up
and
I
call
it
out
in
the
pr
which
is
occasionally
you'll
get
these
messages.
It's
really
it's!
It's!
This
it's
out
of
the
reflector.go.
A
There
are
these
warnings
about
how
it
was
unable
to
decode
meta
v1
event.
Now,
if
you
used
octane
previously,
we
used
to
get
some
warnings
about
this
list
and
watch
that
would
sometimes
come
into
the
console.
A
I
and-
and
I
don't
know
what
ever
generated
those
either,
but
I
asked
some
folks
about
this
error
specifically
and
if
I
should
worry
about
it,
if
it
was
gonna
cause
memory,
links
or
cause
like
stray
api
calls
or
workers
to
be
spun
off
and
and
the
consensus
was
no
also.
No
one
could
give
me
a
clear
answer
as
to
why
my
informers
were
generating
this
error,
but
the
consensus
was
you
don't
have
to
worry
about
it?
I
was
like
fair
enough.
A
We
also
did
some
non-scientific
testing
by
I
just
let
octet
run
for
like
three
days
straight
and
watched
the
memory
and
it
didn't
grow
in
an
insane
amount.
My
cluster
didn't
get
spammed
and
and
and
fall
down
funny.
This
is
funny
when
sam
and
I
were
initially
testing
this
pr
gke
decided
to
do
an
upgrade
of
the
clusters
and
was
like
rolling
nodes
and
doing
this
weird
upgrade,
and
I
I
thought
that
my
pr
had
crashed
the
cluster.
I
was
like.
A
B
Yeah
I
tested
it
briefly.
I
I
think
this
is
incredible.
You
know
just
overall
responsibleness
is
just
I
probably
order
a
magnitude
better
than
it
used
to
be
so
I
think
everybody
will
will
see
huge
speed
improvement.
As
you
said,
this
is
the
way
it
should
work.
So
so
I'm
really
excited
to
try
this
on
on
master
the
lighthouse
you
didn't
mention
that
wayne,
but
on
I
did
some
testing
with
the
lighthouse
performance
index
and
on
my
mac
it
jumped
from
5
to
46.
B
A
A
Cool,
so
moving
on
to
the
next
item:
I'll,
let
sam
take
this
one,
since
it's
all
work
that
sam
did
go
ahead.
C
Yeah
sure
so
this
one
is
well
if
you
could
click
the
link
to
the
example,
plugins
repo,
it's
still
work
in
progress,
but
essentially
we
discussed
this
in
the
last
community
meeting
where
the
sample
plug-in
we
have
is
capturing
a
ton
of
patterns,
and
it
can
be
a
little
unwieldy
to
even
just
over
time
as
we
continue
to
just
add
things
on
there.
It
could
just
be
unwieldy
to
understand
all
of
the
possible
features
of
a
plug-in,
and
so
this
just
breaks
them
down
into
small,
smaller
repos.
C
It
does
only
one
thing,
specifically
one
thing
and
effectively
give
a
at
a
glance
documentation
for
a
very
specific
thing
that
you're
trying
to
do
so.
Most
of
this
stuff
is
boilerplate,
but
we
just
a
lot
of
the
comments
within
the
code
was
just
rewritten
to
pinpoint
to
show
like
this
is
the
thing
that
is
different
about
this
particular
plug-ins.
To
help
isolate
that
it's
still
missing
a
lot
of
the
typescript
equivalents.
It
should
be
possible.
C
I
think
we
just
held
off
on
that,
because
we're
really
trying
to
hit
that
march
24th
release
date,
so
community
contributions
also
accepted,
because
I
think
these
are
just
the
easy
first,
you
know
first
time
issues
or
and
or
you
have
interest
in
building
out
plug-ins
but
don't
know
quite
know
what
to
build.
Yet
some
of
this
stuff
is
also
kind
of
nice
because
we
know
exactly
what
to
build
and
it's
already
scoped
to
a
very
specific
feature,
and
so
we'll
we'll
keep
adding
on
to
this
over
time.
C
I
I
already
have
like
another
four
or
five
more
patterns
off
the
top
of
my
head
that
I'd
like
to
see
added
in
here,
so
it's
yeah.
So
hopefully
you
know
this
will
be
a
resource
that
we
can
start
pushing
more
formally
and
moving
forward.
Hopefully
clear
up
some
of
the
confusion
around
what
plugins
can
or
cannot
do.
A
A
A
Commit
that
right
directly
commit
it
directly
there
that's
a
popular
that
one
comes
up.
A
lot
like
I
want
to
write
a
plugin,
but
I
don't
want
my
thing
to
be
called
every
second,
and
so
I
think
that
and
then
I
think,
a
probably
like
a
storage
storage,
one
is
probably
a
good
one
too.
A
So
yeah
thanks
for
this
work
sam.
This
is
great.
This
is
already
more
useful
than
than
it
originally
was,
and
we
yeah
we'll
continue
to
push
changes
here
and
just
to
reiterate,
this
is
a
great
place
for
first-time
contributions
and
and
just
like
getting
started,
and
it's
like
a
lot
of
bite-sized
chunks
here
that
can
that
are
easy
to
to
get
get
into.
A
So
all
right
this
next
one
is
the.
I
think
it's
the
one
I
think
yeah.
So
this
one
was
added
by
bikram.
Do
you
want
to
talk
about
this
one,
a
bit
vikram.
D
Sure
so
this
story
and
its
corresponding
pr
that
relates
to
js
plugins
they're,
all
addressing
the
use
case
where
people
with
their
own
custom
store
implementation,
might
want
to
use
additional
data
from
the
plugin
and
the
use
case
that
we
are
looking
for
in
this
specific
scenario
is
where
we
have
a
store
implementation
which,
where
we
use
a
proxy
which
needs
access
to
some
tokens,
that
we
can
pass
to
the
store,
and
it's
currently
not
possible.
D
I
mean
it
was
not
possible
with
the
plugins,
but
with
this
change
you
could
add
an
extra
argument
to
all
the
calls
to
the
store
methods
and
all
these
random
key
values
will
be
passed
over
to
the
context
and
the
store
can
use
the
the
key
set
in
the
context.
You
know
to
address
the
issues
for
the
missing
values.
A
Awesome
yeah
yeah.
This
is
a.
This
is
a
pretty
cool
change
for
for
people
who
are
looking
to
like
extend
octane
or
do
custom
things
with
octant.
One
of
the
one
of
the
other
things
that
came
up
out
of
this
was
the
idea
of,
like
you
have
some
data
set.
A
That
represents
maybe
like
what
a
cluster
used
to
look
like
or
what
a
cluster
is
going
to
look
like,
and
you
create
an
object
story
under
like
an
object,
store
implementation
that
satisfies
this
interface
and
then
the
metadata
can
be
used
to
like
fetch
the
specific
like
either
a
time
slice
or
a
series
or
like
you
can
use
that
metadata
to
kind
of
do
whatever
you
want,
and
then
you
could
essentially
have
an
octant
running
on
like
a
cluster
that
doesn't
exist
right,
it's
just
like
it
might
exist
or
it
used
to
exist
or-
and
you
could
you
could,
instead
of
having
to
like
mock
and
fake.
A
All
of
these
api
calls
through
like
in
like
fork,
octane
you
can
use
metadata
so
that
your
object,
store
implementation,
can
just
inspect
metadata
and
then
do
short
circuits
to
go
fetch
what
it
needs
to
fetch
or
do
what
it
needs
to
do.
A
It's
pretty
it's
pretty
neat,
because
we
there's
another
pr,
that's
kind
of
related
to
this,
where
you
can
inject
the
object
store
for
those
who
aren't
familiar
with
how
octan
starts
up,
there's
a
runner
that
you
use
when
you're
extending
octant,
and
there
was
an
a
set
of
work
that
let
you
inject
the
set
of
options
into
that
runner
and
those
options
can
include
your
your
store
now.
A
So
you
combine
the
ability
to
inject
a
store
with
the
ability
to
add
context,
specific
metadata
to
those
store
methods,
and
it
provides
a
much
better
interface
for
people
extending
octant
to
do
embedded
data
or
data
that
requires
extra
access
tokens.
Things
like
that
and
thanks
to
vikram
this
is
supported
both
in
the
go
for
for
go
plugins,
making
object,
store,
calls
and
javascript
plugins
making.
A
B
D
B
B
B
D
B
Maybe
I
will
not
give
it
that
much
today,
zoom
is
not
working
with
me.
Yeah
I
had
to
reopen
the
zone.
I'm
sorry.
E
E
A
So
while
we
wait
for
milan
to
get
back,
we
can
give
a
little
progress
update.
Oh
wait!
He's
back
around.
B
That's
much
better.
Can
you
hear
me
yep
all
right,
so
yeah
we've
been
talking
about
the
resource
viewer
quite
a
bit
recently,
thinking
about
it
doing
different
approaches,
trying
different
things
and
basically
what
what
I'm
trying
to
do
right
now
is
just
for
this
next
release
kind
of
do
the
first
step
in
in
in
what
we
think
it's
going
to
be
a
little
longer
road
in
improving
the
how
the
resource
viewer
works
and
how
it
we
interact
with
it.
And
the
starting
point
was
this
design,
which.
B
Cool,
so
the
starting
point
was
this
research
and
design
that
esha
did
a
little
while
back
and
and
basically
based
on
all
that
knowledge
that
she
collected.
This
is
kind
of
initial
approach.
We
wanted
to
take
and
I've
been
working
on
this,
and
this
is
where
I
am
right
now,
so
I
created
a
bunch
of
stories
in
a
storybook
just
to
cover
all
the
different
use
cases
and
differences.
You
can
see
it's
it.
It's
pretty
much,
just
the
changes
in
style,
so
the
boxes
look
a
little
better.
B
The
text
is
a
little
easier
to
read.
We
have
the
icons
inside
each
box
indicating
the
status
also
selection
is
indicated.
Better
and
selection
is
also
set
initially,
when
you
open
a
new
view.
I'll
show
that
in
action
later
so
so,
small
improvements,
just
in
in
visual
appearance
of
the
resource
viewer,
is
kind
of
a
first
step
we
want
to
take,
and
then
I
can
show
some
other
layouts
and
how
they
look.
B
So
this
is
pretty
similar
to
the
what
isha
designed,
except
we
don't
still
don't
have
this
expanded
collapse,
feature
that
that's
coming
a
little
later,
but
actually
I
cheated
a
little
bit
in
this
and
I'll
explain
that
later,
when
I
show
the
octant,
but
it
it's
kind
of
attempt
to
show
the
deployments
at
the
top
and
then
replica
sets
parts
and
it's
a
top-down
view
and
the
way
some
other
layouts
look
as
so
for
views
that
have
ingress.
This
is
a
how
they
look.
B
Currently,
you
can
still
see
that
in
this
case,
there's
five
boxes
in
a
row
you
can
still
see,
even
even
if
you
have
only
like
a
half
screen.
Let
me
update
that,
because
this
doesn't
update
you
can
still
read
every
every
pretty
much,
even
even
the
kind
and
and
type
so
this
is
kind
of
what
what's.
This
situation
has
been
my
gold
just
to
show
it
at
least
five
boxes
inside
the
row
initially
without
in
this
side
of
screen,
which
I
think
is
pretty
typical.
B
It's
probably
1200
pixels
wide
if
you
have
so
this
is
a
very
small
layout.
If
you
get
to
the
other
extreme,
you
still
need
to
work
on
this.
Obviously,
but
when
you
use
I
mean
there's
no,
not
much
improvement
here,
but
we
have
some
ideas:
how
to
do
that
and
I'll
show
that
here.
B
So
this
is
a
little
less
extreme
crd
view
that
has
12
boxes,
which
is
a
I
think,
lower
limit
when
when
things
get
a
little
more
complicated,
but
if
you,
if
you
look
at
this
view
and
analyze
it
some
interesting
things
start
to
show
up.
B
For
example,
something
like
this,
you
can
see
that
there's
a
huge
amount
of
clustering
here
going
on.
So
if
you
can
display
this
initially
something
like
this
and
then
allow
user
to
expand,
then
we'll
give
enough
information
initially
without
overwhelming
user,
with
the
with
the
you
know,
the
whole
image,
so
something
like
that.
B
B
So
if
I
go
back
to
the
where's
that
I'm
sorry
echo,
so
this
is
exactly
the
same
deployment
as
the
one
I
showed
previously
here.
A
I
can
tell
you're
using
the
the
old
version
of
octane
by
the
yes.
B
It's
obvious
so
see.
This
is
how
the
same
deployment
looks
by
default.
But
if
you
make
just
a
couple
of
changes
that
I
did
in.
B
B
A
So
no,
this
is
great.
I
I,
the
only
initial
feedback
I
have
immediately
is
that
I
have
I'm
having
a
really
difficult
time,
distinguishing
the
icon
from
the
background
right
and
then
in
I
felt
like
in
pia
or
in
isha's,
yes,
design,
yeah.
It
was
like
the
white.
I
think
I
like
that.
A
Yeah,
I
think
I
think
I
I
think
I
like
that
better
right,
because
the
green
is
it's
already
like
yeah.
It's
already
gonna
have
the
green
check
showing
through
the
the
card,
and
so
but
other
than
that
this
is
yeah.
I
I
mean,
I
know
it's
just
a
style
change
but
like
there's
a
lot
to
style
like
it
goes
a
long
way.
Oh.
A
B
B
B
Right
into
for
some
reason
any
except
pod,
because,
as
you
can
see,
there
is
no
link
here
either.
I
don't
know
why
we
don't
do
link,
but
we
can
definitely
change
that.
A
Yeah,
I
think
we
would
want
that
to
work
universally,
so
whatever
whatever
that
ends
up
being,
that
we
need
to
fix
to
make
it
so
that
you
can
double
click
and
follow
that
link.
That
would
be.
That
would
be
great
because
I
really
like
that
feature.
I
I
find
it.
I
guess
pod,
because
what
what,
if
there's
a
bunch
of
them,
which.
B
A
So
yeah,
I
think
that's
probably
why
pod
is
like
that,
because
normally
it
would
show
the
the
the
list
of
all
the
pods,
but
there
might
be
a
good
way
to
you
know:
here's
some
no
yeah!
I
don't
know
a
good
way
to
do
it.
A
B
B
It's
just
your
opacity
is
different.
I
see
yeah,
so
I
I
can
increase
the
opacity
in
boxes
too,
but
I
thought
it
looks
nice
with
a
little
bit
of
capacity,
especially
with
the
border.
You
know,
if
you
look,
and
so
this
is
their
color
yeah
icon
is
the
same
color,
but
then
the
background
has
a
I
think,
capacity
of
zero
six
or
something
like
that.
I
don't
remember:
what's
the
last
one.
A
Yeah,
maybe
just
making
the
sh
the
the
the
strike
the
border,
the
a
little
thicker
would
would
tie
the
the
like.
I
see
what
you're
saying
I
agree
like
having
it
kind
of
softened
is
nice
but
like
in
this
view
in
particular,
where
you
have
two
instances
of
the
other
green.
It
feels
like
it
just
looks
completely
different
right
yeah,
but
no,
this
is
great
yeah.
This
is,
and
I
like
it
any
other
who
else
has
feedback?
Why
am
I
talking.
C
Yeah
well,
I
get
you
mentioned
this
very
briefly
like
there
was
one
case
where
you
mentioned,
that
we
could
actually
cluster
a
few
of
these
nodes
right
and
it's
like
it
reminds
me
of
like,
like
I
wonder,
like
one
it's
like
whether
or
not
we
can
actually
define
these
relationships
as
a
directed
graph
and
then,
like.
I
remember
if
there's
like
oh
there's
a
long
time
ago,
there's
like
this
idea
of
like
a
strongly
connected
component
in
graphs,
where
you
can.
C
Out
like
a
graph
into
essentially
groups
based
on
whether
or
not
every
single
node
in
that
particular
group
can
reach
each
other,
and
if
we
can
do
that,
we
can
kind
of
algorithmically
generate
these.
Like
larger
subgroups.
B
C
Depending
on
how
relationships
are
defined
like
it
might
not
be
like,
the
directions
might
not
be
in
the
way
that
we
think
they
are
because
sometimes
right
like
if
we
think
about
like
between
like
say,
a
deployment
in
a
pod,
it's
pretty
straightforward
or,
like
maybe
between,
like
say,
storage,
but
like
something
like
like,
for
example,
this
cluster
api.
Like
I
don't
know,
I
don't
know
like
if
they.
C
What
would
they
be
yeah.
A
A
This
is
the
the
thing
that
you're
looking
at
is
the
center
and
then
it's
just
kind
of
we're
not
trying
to
tell
you
the
direction,
we're
just
showing
you
the
clusters
that
are
connected
to
it
and
then
like
yeah,
instead
of
trying
to
guess
at
the
directionality
of
things,
just
getting
them
group,
even
just
even
if
they
were
just
grouped
like
this,
that
is
to
me
a
better
view
than
the
default.
B
Rights,
no,
that's
that
that's
exactly
the
idea,
I
think,
but
I
agree
with
both
of
you
and
actually
stethoscope.
The
the
library
we
use
supports
has
a
support
for
detecting
it
has
a
lot
of
different
apis
and,
and
some
of
them
can
be
used
just
to
detect
situations
like
this.
I
have
not
tried
them
yet,
but
I
think
that
that's
gonna
it
based
on
a
first
look.
B
It
looks
really
promising
that
can
solve
situations
like
this
one
and-
and
especially,
I
would
be
really
curious
to
run
that
on
something
like
this,
but
here
you
can
see
the
same
thing.
Actually,
this
one
is
a
little
more
specific,
because
if
I
can
find
a
secret
here,
it's
even
hard
to
find
because
oh
yeah,
that
must
be
secret
or
service,
so
secrets
and
services
are
used
by
a
lot
of.
B
Other
resources,
so
this
is
really
hard.
You
know
this
is,
and
you
don't
want
your
secrets
or
service
to
be
like
a
center
they're
more
like
attached
to
the
to
whatever
object.
They
use
them.
Yep.
A
B
Misleading
so
secret
service
accounts
they
should
almost
be
like
outside
of
the
chart
or
maybe
not
connected
or
connected
with
the
beta.
A
Yeah
or
like
just
thinking
about
this,
a
little
more
like
if
you
if
the
secret
and
service
account
work
weren't
connected
and
then
they
were,
but
they
were
on
there
like
at
the
bottom,
like
you
have
them
right
and
they
had
a.
B
A
And
then,
when
you
clicked
on
it
object
status
showed
a
list
of
all
the
things
that
were
using
that
secret
right.
So
the
secret's
never
actually
even
part
of
the
the
graph
that
we
draw
it's
just.
But
but
you
know
it's
there,
you
know
things
are
using
it
and
you
can
click
to
see
a
list
of
all
the
things
that
are
using
it
and
then
maybe
from
there.
You
could
click
and
say.
Show
me
like
get
like.
A
A
E
A
C
A
C
One
final
point:
before
we
switch
topics,
do
you
think
it'd
be
helpful
here
like
if
we
click
the,
for
example,
this
one
elasticsearch
replica
set?
If
we
highlight
the
first
degree
nodes
that
it
was
connected
to.
B
Might
be
yeah
like
I
think
you
get
to
that
idea
when
I
was
playing
here.
It
would
really
help
like
I
was
playing
with
this
pod
and
trying
to
see
what's
connected
to
it,
and
it's
kind
of
hard
yeah
might
be
a
different
like
mode
of
operation
when
you
click
some
check
box
here
or
maybe
even
a
global
settings,
or
something
like
that.
A
Like
make
the
lines
bold
that
are
coming
out
right,
yeah
yeah,
that
would
be
just
that,
like
you,
could
click
on
that,
and
it
would
bold
those
lines
that
that
is
also
good.
That
yeah.
I
want
that.
Only.
B
Lines
or
the
connecting
nodes
as
well.
A
B
Yeah,
I
think
we'll
get
there
in,
but
I'm
really
excited
that
it's
and
right
now
we
are
kind
of
going
step
by
step
and
doing
iteration
by
iteration,
which
is
much
easier
than
you
know,
trying
to
solve
the
whole
problem.
So
so
I
mean
I'm
excited
that
this
is
going.
C
B
Right
right,
yeah,
there's
there's.
Actually
I
think
it's
a
right-click
extension.
C
C
A
Excellent
yeah:
this
is
great
any
other
comments
before
we
move
on.
Actually
that
was
the
last
thing
we
had
so
is
there
anything
for
the
open
q
a
before
we
end
the
meeting.