►
From YouTube: Octant Community Meeting - September 30, 2020
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
A
I
think
it's.
I
think
it's
not
unreasonable
to
say
that
we
can
lead
with
a
javascript
plugin
as
our
bleeding
edge
feature
branch
first
and
just
say
if
you
want
features
implemented
or
if
you're
doing
something
that
doesn't
have
to
be
in
go
or
if
you're
just
focused
on
building
a
ui
javascript
is
the
better
way
to
do
it
because
you
get
a
lot
of
nice
features
like
hot
reloading
and
but
if
you
aren't
then,
or
if
you
just
have
to
be
in,
go
for
some
reason.
A
There's
no
reason
why
we
can't
support
that,
and
so
I
think
this
is
a
pretty.
I
think
there
there's
a
pretty
strong
case,
or
at
least
we're
not
sure
that
it's
going
to
be
the
correct
move,
and
so
at
least
we
have
said
that
go
plugins
might
be
deprecated
in
previous
meetings
before
and
I
yeah
that
probably
might
not
be.
B
True
yeah,
I
think
that's
right,
sam.
I
I
think,
as
we
as
we
started,
to
look
at
what
it
means
to
expose
this
api,
we're
realizing
that
the
the
the
core
elements
of
our
plugin
api
are
really
the
the
the
ability
to
make
and
perform
requests
to
the
kubernetes
api.
They
are
the
ability
to
send
actions
to
a
client
and
it's
the
ability
to
render
components
and
views
out
from
your
plugin
right
and
and
the
the.
B
Like
making
a
dynamic
web
component
or
enhancing
our
iframe
experience
so
that
people
can
embed
their
graphs
and
and
charts,
and
things
like
that
so
and
you're
right,
like
supporting
the
go
endpoint
and
the
javascript
runtime,
really
isn't
an
issue.
Now
it
was.
It
was
kind
of
an
issue
before
when
we
were
looking
at
like.
What's
what
like
is
octant
going
to
ever
run
in
cluster,
or
is
that
a
completely
separate
application
right
or
a
completely
separate
piece
of
code
as
there
is?
B
And
since
we're
saying
you
know,
octanes
is
an
application,
we're
going
to
make
it
an
electron,
the
the
the
problems
we
had
before
of
like?
Well,
how
do
you
do
plug-ins
in
cluster
with
go
they're
compiled
binaries
all
this,
like
that?
All
of
those
kind
of
problems
also
go
away
so
yeah.
I
think
I
think
it's
likely
that
we
will
the
the
go.
B
Plugins
will
be
the
slower
channel
for
plugin
development,
but
once
we
get
to
1.0,
everything
like
everything
should
have
feature
parity
between
the
javascript
and
go
plugins,
and
I
think
that's
a
reasonable
thing
to
say.
C
Yeah-
and
I
absolutely
agree
with
that-
I
think
this
is
one
of
the
cases
where
we
definitely
need
both
and
we
may
not
have
100
parity,
but
hopefully
at
some
points
we
will.
C
But
you
know
we
definitely
need
to
support
both
and
because
this
is
a
community
oriented
feature
and
the
more
support
from
him
from
community.
We
get
the
better
and
the
best
way
to
achieve
that
is
to
support
as
many
languages
as
possible.
B
I
would
I
would
probably
I
might
slightly
disagree
with
the
language
as
many
languages
as
possible.
We
will
support,
go
and
we'll
support
the
the
typescript
like
I
don't
want
to
get
into
the
the
the
like,
like
I
know
before
we
were
like.
Oh
hey,
look,
you
can
do
this
in
python
because
it's
grpc
and
all
that
but,
like
I,
don't
think
we're
gonna
we're
not
gonna
formalize
support
for
that
stuff,
like
we're
gonna
leave
that
as
a
if
you
wanna
do
it
and
you
wanna
build
all
of
the
all
of
the
artifacts.
B
A
No,
I
think
we
can
update
the
agenda
doc
with
what
we
just
said.
You
guys
may
be
someone
who
just
doesn't
want
to
watch
the
video
or
isn't
tuned
in
right
now.
B
Okay,
yep,
that
sounds
great,
so
moving
on.
So
this
is,
I
wanted
to
show
so
we
we
have
a
session
tomorrow,
which
you
can
register
for
through
the
agenda
doc.
It
is.
It
is
this
this
session
right
here,
it's
code
connect
2020,
it's
like
we're
currently
on
day.
B
B
It's
a
developer,
centric
conference
that
is
going
to
be
highlighting
some
various
things
and
I'm
doing
a
presentation
tomorrow
on
octane
and
specifically
around
using
octane
as
an
open
source
platform
for
building
kubernetes
uis,
the
I
kind
of
wanted
to
show
what
I
was
going
to
get
into
a
little
bit
as
like
a
teaser
for
people
who
might
be
interested,
I'm
going
to
be
building
out
a
plug-in
using
the
javascript
runtime
and
our
typescript
libraries,
which
were
just
recently
published.
B
They
are
very
much
a
work
in
progress
and
we
will
be
continuing
to
work
on
those
as
we
move
towards
1.0.
But
what
I
wanted
to
show
quickly
was
just
kind
of
like
what
I
was
going
to
be
doing,
which
is
this
this
little
simple
plug-in
here.
So
this
plug-in
works
off
of
the
idea
that
there
is
some
type
of
deployment
service
that
obviously
it's
it's
it's
for
demonstration
purposes,
but
so
people
can
yell
at
me
about
about
stuff
and
that's
fine
I'll,
probably
just
ignore
it.
B
Yes,
you
wouldn't
do
this,
but
there's
a
there's
a
system
somewhere
that
has
some
yaml
and
it
has
some
descriptions
of
that
dml
and
and
in
that
yaml
is
like
a
deployment
and
a
service,
and
things
like
that.
This
plugin
is
pulling
that
information
in
from
from
that
third-party
service,
you
can
go
in
and
you
can
view
right
now,
scaffold
it
out,
but
you
can
go
in
and
you
can
view
the
yaml
and
you
can
get
some
details
about
things.
B
The
deployments
themselves
have
the
ability
to
be
installed.
This
status
field
is
going
to
update,
as
as
things
get
installed,
and
anyway,
that's
that's
like
the
that's
it
this.
This
plugin
is
going
to
be
created
from
from
nothing
from
an
empty
generator
or
scaffolded
out
project,
using
our
generator
to
this
functioning
and
working.
B
This
deployment
will
update
when
things
are
installed,
you'll
be
able
to
click
on
it
and
go
to
that
installed
deployment.
This
install
link
will
change
to
a
delete
link
and
all
of
that's
going
to
be
we're
going
to
code
all
that
up
in
about
20
minutes,
so
that
it'll
be
fun.
I
I
suggest
folks
attend
and
I
have
a.
I
have
a
backup
in
case
that
fully
falls
on
its
face
and
I'll
just
swap
it
out.
B
You
know
like
in
cooking,
shows
where
they
things
things
go
horribly
wrong
and
the
pan
catches
on
fire
and
they
stick
a
lid
on
it,
and
then
they
pull
out
the
the
full
version.
So
there'll
be
a
code
repo
that
people
will
be
able
to
get
to.
If
you
can't
attend
the
session,
live
that
actually,
I'm
I'm
staging
all
the
commits,
so
you'll
see
the
progress
of
it
going
forward,
and
so
the
commit
entries
in
the
repo
line
up
with
the
steps
that
we're
going
to
take
in
in
the
session
itself.
B
B
That's
all
I
have
there.
I
believe
milan,
you
were
going
to
show
some
progress
on.
No,
no!
That's
that
wrong
issue.
Today
I
was
going
to
show
the
progress
on
ephemeral
containers
got
it
yeah.
C
I
I
really
don't
plan
to
show
navigation
today,
it's
we
are
making
a
lot
of
good
progress
and
I'm
hoping
to
show
the
finalized
navigation
next
time.
A
Okay
sure,
then
I
will
show
what
I'm
working
on,
even
though
it's
far
from
complete,
but
I
think
this
will
be
useful
in
case
we
get
some
or
yeah.
We
get
some
feedback
on
this
particular
feature
because
there's
a
lot.
C
A
Sure
this
looks
good
enough.
All
right,
everyone
can
see
the
screen,
okay
cool.
So
let
me
start
off
by
running.
This
is
currently
a
master,
build
of
often
with
something
that
I'm
working
on
currently.
A
But
I
want
to
start
the
conversation
here
with
the
kind
plugin
and
I've
demoed
this
before
in
a
previous
community
meeting.
But
now
it's
a
little
bit
more
fleshed
out
and
it
highlights
some
of
the
things
that
we've
been
working
on
and.
A
Know
from
don
in
the
dot
16
release
like
we
get
like
small
things
like
form
validation
here,
like
you
know,
being
able
to
show
cannot
be
empty
and
you
can
actually
submit
this
data
have
some
basic
validation
and,
of
course,
if
I
want
to
throw
alerts,
it's
possible
to
get
an
alert
here,
but
of
course
it's
bugged
with
the
action
request,
but
I
think
the
print
request
still
works
so
half
of
it's
working,
but
it's
just.
A
This
is
just
a
really
nice
tool
to
be
able
to
create
a
kind
cluster
really
quickly
so
like.
If
I
had
like
test
two,
you
can
select
a
specific
version
and
then,
if
you
hit.
C
A
You
can
also
get
like
a
nice
check,
boxes
of
various
feature
gates
and
before
really
what
I
was
doing
was
kind
of
was
an
example
like
you
would
you'd
have
like.
I
don't
know
like
if
I
have
a
project
I'd
end
up
doing
some
weird
stuff
like
just
having
a
whole
folder
through
these
kind,
configs
that
I'm
using
to
just
boot
up
development
environments
and
just
to
test
out
something
on
kubernetes.
A
So
this
is
just
a
good
way
to
just
create
disposable
clusters
locally
and
not
have
to
think
too
deeply
about
what
you
know.
What
the
gamble
should
look
like
or
just
constantly
going
through
the
docks-
and
I
think
this
was
this
used
to
be
a
problem
in
the
past,
but
it's
better
now.
I
think
it's
this
one
yeah
so
like
adding
feature
gates
like
if
you
do
it,
pre
0.8
you'd
have
to
do
all
these
crazy,
cube
admin
patches
and
now
they
actually
introduced
a
feature
gate
flag.
A
So
it's
I
think,
I'd
test
something
like
this
yeah,
where
you
can
just
list
out
the
feature
gate.
So
it's
a
lot
easier
now,
if
you're
doing
0.9,
but
which
still
doesn't
you
know
which
this
is
nice.
This
is
super
awesome,
but
this
is
still
going
to
involve
me
like
saving
a
bunch
of
ammo.
C
A
A
cluster
and
you
currently
on
this
18.8
kubernetes
version
and
so
that
what
I'm
really
trying
to
drill
into
today
is
this
idea
of
ephemeral
containers
and
there
are
a
lot
of
ideas.
A
It's
very
important
to
understand
that
the
container
image
that
it's
being
deployed
on
is
going
to
limit
you
on
whether
or
not
you
can
actually
create
a
shell.
In
this
case,
we
use
something
called
a
distress
image,
and
this
is
essentially
just
a
really
stripped
down
version
of.
I
think
in
this
case
it's
debian.
I
have
a
dockerfile
here
yeah,
so
it
really
just
copies
a
bunch
of
files
from
base
and
then
it
uses
the
distrolus.
And
what
this
really
just
means
is.
A
So
one
approach
or
one
proposed
approach
to
get
around
this
is
using
a
thermal
container,
which
means
you
just
add
a
second
container
to
the
pod
and
that
secondary
container
will
have
a
lot
of
the
debugging
tools.
You
need
to
actually
do
your
debugging,
so
I
can
show
an
example
of
this
and
I
would
just
copy
that
pod
name
and
I
think
I.
C
A
Have
this
already
pre-baked,
I
do
have
it
pre-baked,
so,
let's
so
the
command
here
using
cube
control,
because
this
is
not
coded
into
kind
yet,
and
I
don't
really
necessarily
know
what
this
might
look
like
from
dewey's
standpoint,
but
if
we
create
this
ephemeral
container
now
we
have
this
extra
container
here
and
I
use
the
command.
I
think
it's
using
busy
and
it's
targeting
the
nginx
container,
and
this
just
creates
this
new
debugging
container,
and
now
this
debugging
container
uses
the
image
busybox
and
now
we
actually
do
have
a
shell
for
it.
A
A
Let
me
get
yes,
it
does
so
we
could
do
like
something
like
we
get
local
hoax
and
the
interesting
pieces
we
just
we
were
able
to
get-
or
I
guess
like
do
a
little
bit
of
we're
able
to
interact
with
the
port
in
an
adjacent
container
in
this
same
pod,
and
we
can
take
a
look
and
or
not
then,
but
maybe
you
can
see
this
is
our
index.html
from
our
just
base
engine
x.
So
this
is
kind
of
cool
right
so
like
if
you
were
trying
to
debug
something
networking
related.
A
This
might
be
a
good
way
to
inject
a
container
with
tools
to
help
you
debug,
but
something
that's
more
interesting
is
that
if
you
enable
this
share
process
namespace
into
your
container,
so
in
other
words
well,
I
can
talk
about
the
implications
of
using
this
later,
but
you
can
actually
do
something
like
this.
You
can
actually
look
up
the
pid
of
your
of
the
process
running
in
a
different
container,
and
then
you
can
actually
do
something
like
root,
and
now
all
of
a
sudden
you
can
go
into
say,
share,
I
don't
know
share.
A
And
if
we
do
something
like
this,
we
can
say
I
don't
know
it's
a
fun
one
hacked,
and
if
we
save
this,
we
can
theoretically
see
something
like,
oh
or
maybe
not,
or
am
I
maybe
not
doing
the
right
pod,
but
we
should
be
able
to
modify
yeah
so
this
if
it
actually
saved
it,
should
show
the
updated
version
or
maybe
not
or
maybe
I'm
doing
the
wrong
one.
A
Oh
there
it
is
yeah.
I
was
looking
at
this
welcome
thing,
so
yeah
say
so
that's
kind
of
the
idea-
or
maybe
it's
not
in
the
big
font.
So
maybe
I'll
do
the
big
font.
If
test
and
yeah
there
we
go
so
you
can
modify
the
file
system
of
something
in
the
file
system
of
a
shape
container.
A
And
so
what
I
want
to
do
in
opt-in
is
to
be
able
to
have
something
like
this,
where,
if
you
have
this
terminal
and
execs
the
shell
by
default-
and
I
think
I
might
have
to
code
here-
yes,
apparently
we
just
parse
command.
We
just
assume
it's
shell
and
what
I
really
want
is
to
be
able
to
say:
if
this
fails,
can
we
do
this
alpha,
debug
and
say
hey?
If
you
have
your
feature
gate
enabled.
Would
you
like
to
create
an
ephemeral
container
and
it'll?
A
Do
it
for
you
and
then,
and
then,
if
you
go
through
the
drop
down,
it'll
show
this
new
container
and
then
also
I
want
to
make
this
share
process.
Namespace
discoverable,
because
there
are
two
pieces
of
this
there's
like
one
is
that
on
a
debugging
cluster
or
maybe
a
staging
cluster,
this
is
acceptable.
I
don't
think
people
might
want
this
in
production,
or
at
least
it
there
are
implications
of
using
this
which
people
might
not
want
in
production.
A
But
that
said
for
debugging,
this
is
kind
of
neat,
but
the
problem
is
that
if
you're
not
aware
this
even
exists
in
the
first
place,
you
might
have
to
push
the
deployment
again
and
then
you
kind
of
what
you
lose
here
is
that
you're
not
able
to
debug
a
currently
running
container
you'd,
have
to
regenerate
and
create
a
new
pod
with
the
shared
process
namespace,
and
then
you
can
access
the
file
system.
A
So
there
are
a
lot
of
gotchas
here,
like
you,
can
always
push
a
debugging
container
to
something
that's
running
if
the
feature
gate
is
enabled,
but
you
can
do
more
if
you
configure
it
slightly
further
and
to
further
complicate
this
a
little
bit
more
checking
whether
or
not
a
feature
gate
is
enabled
is
kind
of
difficult
you
can
using
either
using
the
discovery
client.
A
You
can
check
whether
or
not
this
you
have
this
resource
in
the
api
server,
but
there
isn't
really
a
way
to
check
that
the
feature
data
is
also
enabled
on
the
cubelet,
and
so
what
you'd
end
up
with
is
that
you'd?
You
could
say.
Even
if
we
often
looks
up
and
say
we
found
your
feature
gateway
for
firmware
container
resource
it
still,
this
whole
thing
still
might
not
necessarily
work.
A
So
that's
kind
of
for
the
difficult
piece
about
this,
where
they're
just
a
lot
of
failure
modes
and
to
really
get
the
most
value
out
of
this.
It
needs
a
little
bit
more
configuration
that
we'd
like,
and
it's
hard
to
really
just
make
this
in
button
form
right.
You
can't
it's
not
just
click
buttons,
it's
that,
like
you'd,
have
you'd
have
a
lot
of
ifs
before
you
can
get
to
the
state
and
just
to
move
past
this
a
little
bit
more.
A
A
They
have
this
criteria
for
graduating
a
feature
so,
and
one
of
these
are
saying
that
oh,
it
must
be,
you
must
see
use
in
projects
articles
whatever
right,
so,
in
other
words,
they
want
to
see
adoption
of
these
feature
gates,
and
what
this
allows
people
to
do
is
one:
let
people
know
that
these
feature
gates
actually
exist,
mix
and
discover
it
makes
them
easy
to
set
up,
but
also
it'd,
be
really
cool.
If
often,
all
of
a
sudden
becomes.
A
This
signal
boost
for
a
lot
of
these
feature
gates,
and
so
this
is
something
that
we
can
look
into
as
moving
forward
in
our
future
releases.
To
see
like
what
feature
gates
are
being
pot
or
you
know,
are
popular
and
what
things
that
we
want
to.
Try
supporting
into
often-
and
maybe
we
can
see
if
other
parts
of
the
community
actually
wants
to
work
with
us
on
building
out
a
ui
for
some
of
these
things.
A
So
that's
about
all
I've
got
I'm
definitely
looking
for
input
on
what
the
developer
experience
for
this
is
supposed
to
look
like,
because,
obviously,
if
you
just
see
a
button
that
says,
create
ephemeral,
container
you're
going
to
be
like
what
is
in
a
fumble
container,
what
does
it
do?
Why
do
I
need
one
right
and
it's
hard
to
wrap
up
what
I
just
talked
about
in
the
last
15
minutes
in
a
single
button
right.
A
So
this
is
where
it
gets
tricky
and
I
would
definitely
love
feedback
on
what
this
should
look
like
how
you
know:
how
do
we
introduce
more
feature
gates
in
the
future
and
just
you
know,
just
feedback
in
general
on
how
we
think
often
should
move
forward
with
debugging
tools
like
this.
A
So
that's
about
it.
Thanks
for
listening
to
my
ted
talk.
B
Oh,
that's,
that's
really
cool!
I
really
like
the
there's.
There's
been
many
times
where
I've
wanted
to
like
get
into
a
container
that
I
was
building
or
but
like
that,
I
have
to
go
and
I
have
to
add
a
shell
into
it
and
add
all
these
tools
into
it
right,
and
this
kind
of
lets
me
not
have
to
deal
with
doing
all
of
that.
I
really
like
that.
C
Yeah,
it
looks
pretty
cool.
I
I
really
like
the
idea
of
attempting
to
to
establish
the
connection.
If
that
doesn't
work,
then
prompting
the
user
to
to
you
know,
create
this
new
container
and
maybe
maybe
have
something
like
do
that
only
once
and
have
give
them
the
option
to
do
it
automatically
after
that.
If,
for
example,
I
would
I
would
do
that
every
time,
but
in
the
same
time
I
want
to
be
prompted
with
the
option
to
do
that
first
time,
so.