►
From YouTube: Meshery Development Meeting (Apr 21st, 2021)
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
So
so
elizabeth
and
khan
and
miami,
I
think
the
three
of
you
may
not
be
aware
of
one
of
by
the
way
am
I
sharing
the
whole
screen
or
just
my.
B
A
Got
you
that's
the
problem
with
that
is.
I
was
talking
bad
about
novendo
a
second
ago
and
so
he's
gonna
see
that
that's
gonna
be.
A
I
should
just
talk
bad
about
him
in
front
of
his
face,
then
we'll
but
anyway,
bad
jokes,
miami
khan
and
elizabeth,
just
forewarning,
there's
a
standing
rule
in
the
community
that
you
know
that
you
should
record
your
record
your
attendance,
which
drop
your
name
in
and
that,
if
you
so
happen
to
not
put
your
last
name,
someone
might
make
one
up
for
you
and
no
guarantee
that
you
will
like
or
take
to
it
so
be
warned.
A
Actually,
speaking
of
nicknames,
so
we're
about
to
get
going,
I'm
sure
is
going
to
interrupt
me
and
and
start
the
meeting.
But
I'll
say
that
husseina
goes
by
a
couple
of
different
nicknames.
Actually
she
between
well
who's,
saying
to
remind
me
so
there's
it's
there's
hex
dump,
but
then
there's
also
it
was
it.
The
the
piscean
was
that
it.
D
Yes,
yeah,
but
I
renamed
myself
as
hussain
on
the
channel.
People
are
finding
it's
hard
to
reach
to
me
so
yeah.
A
E
So
with
all
those
jokes,
let's
get
started,
I
guess
this
is
the
material
development
call
and
today
is
april
21st
and
we
are
six
minutes
in
and
we
have
ten
people
on
the
call
and
I
guess
we
will
started
with
khan.
Would
you
like
to
talk
about
the
pr
your
recent
pr.
F
Sure,
I
think
the
reason
pr
just
to
use
the
api
for
the
import
github.
F
And
support
it
through
measuring
cuddle.
I
think
I
misunderstood
it
at
first
and
I
think
I
only
implemented
importing,
but
I
think
it
was
pointed
out
that
this
will
still
need
to
be
deployed,
so
I'm
currently
working
on
it
but
yeah.
I
think
the
way
I
do
is
just
use
like
strings
package
and
going,
I
think
it's
not
one
of
the
person
suggests
using
viper,
but
I
think
string
does
split
and
things
like
that,
a
little
bit
more
straightforward.
F
I
look
into
viper
and
I
wasn't
sure
exactly
like
how
they
do
the
parsing
so
yeah.
I
think
another
option
is
to
use
net
url
package
going,
but
even
that
it
still
pose
some
problem
because
we
have
to
extract
specific
sub
path
but
yeah
so.
A
Yeah,
do
others
have
feedback
utkar,
she
might
be
the
closest
to
this.
Is
it
with
caution.
A
A
A
A
spec
on
what
this
is
intended
to
be
from
a
bigger
vision.
I
think
it
might
intrigue
one
two.
It
will
help
with,
as
you
think
on
the
problem,
it
will
help
with
part
of
where
we're
better
understanding
where
we're
trying
to
go
generally.
This
is
quite
well
aligned.
A
There
is,
I
think,
specific,
like
leaving
a
side,
viper
or
strings
or
net
url,
or
what
have
you,
but
but
coming
back
to
it
in
a
moment
that
there's
a
couple
of
nuances
around
measuring
ctl
and
it
as
a
user
experience,
and
at
this
point
I
think
most
people
in
the
call
are
sick
of
me
talking
about
it
as
a
user
experience.
But
but
the
the
point
there
is
that
there's
a
lot
of.
A
A
You
know
easy
to
easy
to
update
bearing
more
of
the
what's
more
meaningful
and
more
impactful,
I
think,
is
the
behaviors
behind
those
end
points
and
we
we
had
gotten
into
a
pretty
decent
routine
of
having
design
specs
written
up
first
on
in
a
number
of
areas,
and
I
don't
know
this
might
be
one
where
we
were
moving
more
quickly
without
a
spec
but
having
multiple
folks
looking
at
it,
trying
to
help
advance
it
good.
A
It's
a
good
reminder
for
us
to
go
back
into
a
spec
kind
of
get
some
stuff
written
down
sort
of
hash
out
the
the
nuances
of
mastery
cdl
commands
and
you
know
hash
out
all
of
what
we're
looking
to
accomplish,
which
is
it's
helpful.
A
A
E
So
the
answer
would
be
no,
because
meshkit
is
a
mastery.
Cdl
has
not
yet
using
config
packages
of
meshkit.
Yet
in
this
scenario,
in
this
particular
player,
I'm
not
clear
about
what
exactly
the
use
case
is.
But
here
I
have
only
one
suggestion,
which
is
basically,
instead
of
doing
a
strings
parts,
you
can
do
use
the
net
url
package
to
do
the
parsing
of
the
urls
that
that's
one
feedback
that
I
can
give.
D
D
D
E
So,
basically,
if
file
fetch
is
the
only
is
the
exact
use
case,
then
there
is
a
utility
function
in
meshkit
which
would
do
that,
like,
basically,
you
don't
need
to
write
code.
I
send
I
posted
the
link
on
the
chat.
E
D
F
Yeah,
I
think
the
euro
is
a
good
choice.
I
just
wasn't
sure,
because
there
is
like
two
different
lengths
that
we
can
support.
Of
course
we
choose
only
one
over
the
other
like
url
is
fine,
but
we
still
have
to
extract
like
the
owner,
the
repository
and
then
the
path,
and
then
within
those
they're
like
the
blob
and
then
that
master
or
like
trio
master,
which
I
think
the
api
hasn't
support.
F
Yet
so
I
just
wasn't
sure
like
using
their
url,
will
give
the
right
case,
because
eventually
we
still
have
to
extract
individual
like
sub
paths
out
of
it.
So
I
was
thinking
like
using
strings
or
it
might
be
more
straightforward,
but
yeah
after
I
look
into
like
the
utils
go
and
see
whether
or
not
that
can
help,
but.
F
F
F
Currently,
I
don't
think
I
check
it.
I
wasn't
sure
I
would
assume
that
if
it's
private
and
the
api
should
return
an
error,
but
I
will
I
will
look
more
into
it,
but
I
think
that
because
the
because
the
measuring
api
should
does
it
important.
So
if
it
couldn't
fetch
the
file,
it
should
report
an
error,
and
so
because
we
rely
on
that
api,
like
in
the
case
of
a
private
repository,
it
should
return
an
error
so.
G
G
A
I
think
well
almost
would
cost
you
want
to
help
us
get
organized
a
little
bit
with
respect
to,
I
think
you're,
maybe
mid-flight,
on
a
small
design,
spec.
C
C
C
I'm
not
quite
sure,
because
I
joined
this
quietly,
but
basically
I
joined
the
discussion,
but
but
I
am
I'm
changing
the
implementation,
the
current
implementation
of
remote
imports,
and
that
is
importing
current
patents
from
remote
places.
It
could
be
an
http
endpoint
or
it
could
be
a
github
repository
or
it
could
basically
anything
I.
The
current
goal
is
to
make
it
a
bit
more
flexible.
So
if
you
would
see
her
the
import,
I
think,
must
be
familiar.
How
the
api
looks
like
right
now.
C
It
has
something
like
this:
that
is,
we
take
in
the
type
of
of
the
provider,
it
could
be
github
or
it
could
be
http.
Those
two
are
supported
right
now,
but
this
is
going
to
change
in
the
with
a
new
implementation
and
it's
moving
to
a
post
type
request.
That
is
the
different.
The
different
requirements
of
the
request
would
be
in
the
post
body
and
those
would
be.
The
first
thing
would
be
time.
That
is
what
type
of
remote
provider
is.
Is
it
github
or
http
or
gitlab?
C
It
could
be
anything,
as
I
mentioned
in
here,
that
machine
server
won't
be
taking
the
responsibility
of
verifying.
If
a
particular
provider
file
provider
is
supported
or
not,
it
would
be
on
the
on
the
remote
provider.
Actually,
I'm
using
provider
n2
context
here.
The
remote
right
is
basically
any
remote
provider,
for
instance,
it
could
be
mystery
mode
provider
or
it
could
be
some
other
one
and
the
second
one.
The
second
context
here
is
the
files
provider,
that
is,
it
could
be
engineering.
C
It
should
be
important
or
github
or
gitlab,
or
with
bucket,
or
something
like
that.
So
so
the
implementation
will
be
heading
towards
that
machine.
Server
won't
be
actually
checking
if
a
certain
provider
is
supported
or
not
certain
files
providers
supported
or
not.
It
will
just
offload
the
request
if
it
sees
that
this
particular
remote
provider
actually
supports
importing
files
from
remote
places.
The
request
would
change
from
get
to
post
and
it
will
contain
a
few
different
param,
basically
fields.
The
first
would
be
type,
as
I
mentioned.
Second
would
be
path.
C
It
is
also
present
right
now.
The
third
thing
that
is,
that
that
I
am
adding
that
was
not
included
in
the
previous
implementation,
was
some
metadata
feed.
This
metadata
data
field
will
actually
vary.
That
is,
it.
C
Will
I've
not
yet
mentioned
in
here,
so
basically,
it
will
account
for
the
things
that
could
be
provide
a
files
provider
dependency,
for
example,
if
it's
github
it
could
be
that
you
wanted
to
take
in
the
maybe
maybe
the
branch
name,
as
I
think
khan
recently
pointed
out
in
a
in
a
github
issue,
but
actually
definitely,
as
mentioned
in
the
issue.
Also.
I
guess
that
this
backward
functionality
is
definitely
needed,
but
it
would
be
implemented
in
a
bit
different
way.
C
A
C
Yeah,
I
I
can't
share
this
thing,
because
this
is
because
this
is
the
original
patterns.
This
is
the
original
patterns
spec.
So
this,
basically
because
this
feature
is
also
an
extension
of
machine
guard,
so
this
will
also
be
our.
C
C
Also
to
answer
one
more
question:
I
think
it
was
about
private
repositories
so
again
quickly.
It
is
that
the
thing
is
that
if
it's
a
private
repository
and
if
you
are
not
already
authenticated
with
github,
so
if
you
are,
if
you
look
so
when
you
choose
misery
as
your
provider
mission
remote
provider
as
a
provider,
and
if
you
choose
to
log
in
with
github,
then
you
will
be
able
to
access
your
private
repositories.
C
But
if
you,
if
you
choose,
maybe
the
nun
provider
that
is
missing
server
only
and
no
other
remote
providers,
or
maybe
some
other
remote
provided
in
the
future,
then
if,
if
it's
none,
then
private
depository
importing
private
positive
repositories
won't
be
possible
and
you
will
get
an
error
from
api
and
that
basically,
you
will
get
affordable,
because
that's
what
we'll
be
getting
from
github,
it
won't
basically
won't
tell
that
it's
unauthorized.
It
will
tell
us
that
that
doesn't
exist.
That's
how
you
have
actually
responses,
so
so
yeah.
C
F
C
Yeah,
actually
so
so,
even
though
the
implementation
would
change,
I
will
try
to
make
sure
that
it
is
so
if
mystery
ctl
is
right
now
using
the
previous
version
of
the
idea
I'll
try
to
make
sure
that
that
isn't
breaks,
and-
and
this
thing
like
basically
these
changes-
these
will
come
in
really
quickly.
So
you
won't
have
to
wait
for
a
really
long
time.
C
So,
basically,
I'm
trying
to
say
is
that
even
if
it
comes,
this
thing
comes
within
six
hours
or
something,
but
we
I'll
try
to
make
sure
that
the
previous
implementation
doesn't
break,
and-
and
the
second
thing
that
I
was
kind
of
promising-
was
that
this
will
drop
really
quickly,
and
that
is
the
changes
will
be
dropping
really
quickly.
F
Yeah,
definitely
that
definitely
helps
yeah.
I
can,
I
think,
right
now,
I'm
only
supporting
like
get,
but
I
think
once
post
out,
I
can
try
to
implement
that
as
well.
Yeah
yeah.
Definitely
I
think
it
shouldn't
yeah.
I
think
it
should
be
doable
but
yeah
I
can.
I
can
try
it
out
and
see
how
it
goes.
A
Yep,
you
guys
might
have
spoken
to
this.
I
might
have
missed
it,
but
yeah
at
some
point
in
a
new
car
shift.
This
is
already
well
yeah.
I
guess
what
I'm
about
to
say
is
probably
whatever
it
isn't
necessarily
directly
within
measure
ctl,
but
would
be
part
of
the
end
point
probably
for
the
local
provider
and
if
a
remote
provider
chose
to
do
the
same.
A
That
makes
sense
that
that
we
we
would
recurse
the
url
that
someone
is
given
that
we
would
go
through
any
number
of
by
default,
probably
just
all
the
sub-levels
of
the
directories
in
search
of
additional
that
we
would
be
greedy,
so
to
speak,
that
we
would
make
it
convenient
for
users
to
import.
Do
a
bulk
import.
C
Yeah,
actually,
one
thing
that
I
forgot
to
write
in
here
is
that
yeah.
So
in
the
previous
implementation,
it
was
like
you
are
supposed
to
give
the
part
of
the
file.
So
if
you
give,
if
you
so
right
now,
what
I'm
heading
towards
is,
if
you
give
the
path
of
a
file,
you
will
just
import
file.
C
But
if
you
don't
give
anything
we'll
just
reverse
the
entire
repository
to
find
all
of
the
parent
files
and-
and
if
you
give
a
part
of
a
sub
directory,
then
we'll
just
go
and
look
into
that
sub
directory
for
the
for
all
of
the
pattern.
Files
as.
I
B
Hey
hi
guys,
so
let
me
take
all
the
screen
from
doing
this.
This
updates
well
shown
in
yesterday's
call,
but
I
guess
this
is
a
to
show
the
community
what
we
have
been
working
on,
since
we
don't
give
up.
So
for
people
who
don't
know
this
is
mesh
map.
We
are
sorting.
We
are
sort
of
working
on
a
visualizer
for
make
sure
you
can
see
what
all
things
are
going
in
your
cluster
and
you
can
filter
through
them.
You
can
select
different
views
to
work
on
it.
This
is
in
very
early
early
stage.
B
B
Yeah,
so
I
guess
I
can
fetch
them
from
over
there.
We
have
different
views
from
which
we
can
select
and
stuff
like
that.
So,
if
I
go
like
few
of
them
are
then
still
being
implemented.
Few
of
them
are
implemented
like
this
one
is
not
implemented
for
now,
so
you
won't
get
any
filters
for
this
particular
kind
of
thing
now.
Multi
cluster
is
right
now
being
worked
on,
so
you
can
see
it
has
different
kind
of
filters
for
different
kind
of
views
which
you
can
see.
You
can
see
what
all
is
happening.
B
So,
let's
see
what
all
needs
this
is.
I
have
and
let's
query
all
of
the
new
stresses
so
for
now,
if
you
don't
select
anything,
it
queries
all
of
them,
oh
so
yeah.
So
these
are
basically
the
name
switches
which
we
have
and
I
have
a
lot
of
namespaces
because
I'm
I
was
testing
out
whether
it
was
updating
in
real
time
data
and
name
just
were
the.
A
B
Is
from
meshing
itself,
like
my
meshing,
is
continuously
looking
at
my
clusterway
and
I
am
getting
logs
for
each
of
the
changes
which
I'm
using
so
like.
If
I
add
or
delete
any
namespaces,
it
will
just
reflect
over.
B
B
So
if
I
do
that,
we
get
updated
think
over
here.
So
basically
what
what
is
happening
over
here
for
the
people
who
don't
know
mesh
sync
is
basically
they're
in
my
cluster
and
it
is
watching
for
any
changes
over
here.
So
if
you
can,
if
you
want
to
look
through
the
locks,
you
will
see
one
lock,
which
said
I
had
created
a
name
face
in
itself,
so
it
detected
a
new
resource
in
my
cluster.
It
added
to
the
meshing
database
and
we
are
using
that
particular
database
to
show
you
the
presentation.
A
G
Yeah,
so
is
this
like
supposed
to
show
how
stuff
in
your
cluster
related.
B
It
is
yeah
the
next
step
would
be
to
show
how
each
of
them
are
related
to
each
other
and
like
this
is
part
of
this
particular
name
space.
So
it
would
come
inside
this.
Currently,
we
don't
have
any
edges
to
show
that
which
particular
resource
is
related
to
which
one
and
how
and
say
but
yeah.
That
is
one
of
the
places
where
people
woke
up.
B
B
B
G
G
B
Yeah,
that's
a
good
input,
because
that's
that's
also
in
our
plan
and
yeah,
I
feel
lackluster
now
that
we
don't
have
it
currently,
we
show
you
guys,
but,
as
you
can
see
over
here
this
panel,
I
was
not
hesitant
to
show
it
because
still
there
are
some
bugs
over
here,
but
probably
what
the
use
case
next
would
be
to
would
be
over
here.
B
B
B
B
Each
of
the
components
like
you
would
have
get
in
a
interview
that,
like
light,
green
and
squares
or
name
spaces
and
apollo
pentagon
is
for
deployments
and
stuff
like
that,
but
they
are
properly
defined
in
figma
for
now
in
future,
once
we
launch
this
to
public,
we
will
probably
have
a
legend
which
will
come
from
like
right
or
left
side.
Well,
it
will
show
that
this
particular
shape
and
this
particular
color
resonance
to
that
particular
resource
type.
So
if
you
want
to
get
more
detail
about
that,
you
can
go
to
figma.
A
If
I
drop
the
link
in
the
chat,
okay,
thank
you
and
actually
joy,
joy,
great
question.
The
end
so
dhruv
gave
the
answer,
which
is
well,
it's
a
it's
a
poor
man's
answer,
or
it's
not
a
good
answer
like
the
the
the
re
and
not
that's
no
reflection
of
dhruv.
Well.
Actually,
in
this
case,
it
directly
is
true
joey.
The
right
answer
should
be.
A
I
don't
know
why
you're
asking
that
question.
It's
so
obvious
what
the
shapes
mean,
but
that's
not
the
case
right,
it's
not
obvious
what
they
mean
and
what
they
do
and
what's
a
rounded
rectangle
versus
a
hexagon
versus
a
pentagon
versus
well,
I
don't
know
nothing.
You
have
to
go
to
the
legend
jeez.
Well,
I
guess
it's
not
very
intuitive,
then
well
so
joy
like
what
a
great
question
there
should
be
a
much
better
answer
we
should
work
to.
A
A
There
are
certain
things
that
can
really
that
can
really
help.
There's
certain
things
that
like,
if
there's
a
node
on
the
map,
that
is
in
fact
a
container
well,
you
might
be
able
to
show
a
docker
logo
or
a
whale
that
kind
of
looks
like
a
docker
logo
that
would
probably
intimate
to
the
user
that
it's
a
container
all
right.
Okay,
fine,
okay!
Maybe
if
you
know
there's
a
bunch
of
kubernetes
flavors
out
there
distros,
I
don't
know
if
there's
like
100
something
of
them.
A
It
might
have
multiple
nodes
and
it's
got
the
different
services
and
name
spaces,
and
it's
got
all
kinds
of
objects
and
things
within
that
cluster,
but
that
cluster
it
might
be
a
an
open
shift,
cluster
or
or
an
eks
cluster,
or
it
might
be
of
a
certain
type
of
system.
Okay.
Well,
that
look.
A
I
guess
what
I'm
getting
at
is
like
logos
are
quite
communicative,
because
it's
that
that's
the
thing
you
know
like
the
logo
is
the
thing
is
the
and
you
either
know
the
logo
or
you
don't,
but
that
in
my
mind,
that's
one
of
the
closest
ways
that
you
can
precisely
communicate
to
someone
like
what
this
thing
is,
but
that's
not
very
creative.
What
I've
just
said
so
to
jubril's
point.
There
are
quite
a
number
of
other
projects
out
here
that
are
out
out
there
that
are
similar
to
this.
They
are
visual
topologies.
A
There
are
competing
projects,
projects
that
kind
of
compete
with
mesheri.
Most
of
them
are
closed
source
that
have
similar
user
interfaces,
and
why
did
I
bring
them
up?
Oh,
is
to
say
you
know,
there's
some
inspiration
to
be
gotten
from
some
of
those
I've
had
teams
in
the
past
create
some
of
these.
Maybe
I'll
I'll
share
some
screenshots
of
what
they've
they've
created.
A
But
I'm
disappointed
with
the
answer
that
I
just
gave
like
I,
I
feel
like
oh
what
I
just
said
like
a.
We
need
to
make
sure
that
we're
using
the
logos
following
the
terms,
the
logos
terms
like
which
I
would
think
that
we
would
and
most
logos
are
available
as
an
svg,
which
is
great
because
we
need
for
things
to
scale
zoom
in
and
zoom
out
scale.
Well,
so
you
know
anything,
that's
rasterized
isn't
gonna,
do
it
for
us,
okay,
fine!
This
is
part
of
the
question
part
of
the
answer.
A
One
of
the
things
that
all
of
us
might
notice
is
that
these
entities
that
are
on
the
map-
they're.
Well,
I
guess
they're
related
because,
like
there's
a
group
of
blue
ones
and
there's
a
group
of
green
ones,
so
they're
probably
kind
of
like
of
the
same
thing
but
are
is
any
are
any
of
them
talking
to
each
other
or
is
there
any
interchange
between
them?
A
I
don't
know
we
should
probably
at
some
point.
We
should
probably
see
that
interchange
how
real
time
would
that
be
if
there's
traffic
going
across
them.
There's
like
a
year's
worth
of
work-
and
I
don't
say
that
I
say
that
literally
a
year's
worth
of
work
to
do
to
build
out
like
very
rich
set
of
interactions
to
be
done
here.
If
you
think
about
it
very
briefly,
any
operations
any
features
that
you
can
invoke
in
other
areas
of
measuring
they're.
A
All
of
those
can
be
done
from
with
in
this
visual
topology
as
well,
and
some
of
them
are
a
lot
easier
to
be
done.
Some
of
them
might
be
slightly
harder,
maybe
or
it'd
be
about
the
same,
drew
if
you
right
click
on
these
anything
going
to
happen
for
us.
B
B
A
Dhruv
always
good
to
get
an
update
on
this.
I
think
I
think
this
that
this
will
be
eventually
really
helpful,
for
people
really
take
the
blinders
off.
E
A
Oh
sorry,
I'm
getting
hammered
with
other
messages.
Oh
the
measuring
preflight
design,
spec
yeah.
I
think
we
should
dig
into
it
because
oh.
E
I'm
in
the
air
gap
support,
oh
isoko
is
not
called
today.
I
guess
he
won't
join.
H
E
So,
to
give
some
context
measuring
detail
currently
like
we
support
multiple
kubernetes
platforms
like
gk,
aka,
cks
and
mini
cube,
so
we
leverage
their
usually,
their
native
cli
is
to
deploy
measuring
onto
that
cluster.
So
what
usana
is
trying
to
do
is
to
potentially
remove
those
dependencies
and
she
has
set
up
a
design
spec
for
it.
Yeah.
D
The
ask,
is
to
actually
get
rid
of
any
scripts
to
deploy
mystery
onto
these
platforms
and
use
golang
sdks
or
enhance
our
mesh
kit
like
currently
meshkit
already
has
apis
to
detect,
cube
config
and
a
few
other
things.
So
the
requirement
for
bringing
up
mystery
varies
for
each
cloud
platform,
so
some
cloud
providers
they
already
have
cli,
which
can
get
us
the
cube,
config
file
with
a
single
command
and
they
do
have.
D
So
in
case
of
eks,
I
think
there
are
two
apis
which
we
can
use
like,
which
says
list
clusters
or
disk,
describe
clusters
from
where
we
can
get
all
the
we
might
get
a
lot
of
information
again.
We
need
to
filter
for
the
given
region,
name
or
cluster
name,
so
there
would
be
some
work
to
be
done
there.
I
guess
so,
even
for
gke
like
we
have
a
complicated
set
of
steps,
but
recently
I
came
across
one
more
article
where
it
can
be
done
with
a
single
g
cloud
command.
D
D
Hopefully,
I
will
do
it
in
the
coming
weeks
and
another
thing
which
came
in
recently
was
about
docker
compose
so
earlier
it
it
used
to
be
a
separate
command,
docker
hyphen
compose,
so
the
docker
has
made
it
an
implicit
command
like
so
darker
space
compose
so-
and
this
also
has
golang
sdk
support,
so
we
we
do
use
docker
compose
command.
If
it
is
not
available,
we
also
go
and
fetch
it
from
github,
so
that
dependency
also,
we
have
to
remove.
So
this
is
another
requirement,
so.
A
So,
in
terms
of
in
sequence
of
priority,
like
the
compose
one
is,
is
perhaps
lower
on
the
totem
pole,
so
to
speak
or
is
perhaps
of
not
of
of
the
lower
of
the
lowest
priority.
Maybe
because
it's
fairly
well
taken
care
of
today,
like
it's
just
generally
present
on
individual
systems
and
when
it
isn't,
we
go
grab
it
and
and
the
one
change
that
might
be
anyway.
There's
that
they're
the
more
popular
systems
that
we
do
manage
kubernetes
systems
that
we
see
people
using
are
well.
A
Actually,
surprisingly,
I
guess
by
by
count
it
has
been
gke
more
than
it
has
been
eks,
but
between
those
two
gke
eks
and
then
aks
has
been
much
further
in
between,
but
hussein
have
you
like
in
an
ideal
world.
Maybe
that
would
be
the
sequence
by
which
we
would
chase
after
these.
But
have
you
already
started
to
get
traction
with
one
of
these
before
the
other.
A
A
Who's
saying,
are
you
still
there?
I
interrupted.
A
Oh
nice,
okay,
hey
of
the
you
were
saying:
you've
done,
you
know
you've
done
research
across
all
these.
Obviously
you
got
the
links
to
each
of
the
sdks
and
but
then
you'd
said,
while
you're
researching
gke,
and
I
think
it's
the
g.
Is
it
g
cloud?
That's
the
name
of
the
cli.
Yes,.
H
D
A
D
The
sdk
way
of
doing
gke
config
right
now
it
has
a
set
of
steps,
and
I
think
this
g
cloud
cli
also
does
all
of
that.
So
it's
like
a.
D
And
if
we
can
explore
around
the
second
cli
single
c
energy
cloud
cli,
so
that
would
be
simpler,
even
if
we
have
to
do
it
the
sdk.
So
I
was
thinking
maybe
I'll,
explore
that.
A
Or
the
refactored
docker
compose,
I
might
reinforce
it
coming
a
little
bit
later.
The
the
singular
exception
to
that
is
that
we
were
asked
last
year
to
speak
at
dockercon
on
webassembly,
and
it
had
been
briefly
discussed
that
we
might
be
asked
to
speak
again
this
year
and
the
topic
being
on
docker
compose
and
doing
doing
that
exactly
like
moving
off
of
the
cli
and
moving
into
use
of
their
new
goaling
implementation.
A
E
A
D
Yes,
I
was
thinking
of
doing
that.
I
I
shall
do
that
actually,
so
there
is
one
already
open
to
do
this
I'll,
open
two
more
for
pks
and
aks
and
darker
compose
anyway.
We
have
one
separate
field.
A
Nice,
so
in
terms
of
comfortable
timing,
not
that
we're
on
not
that
we've
been
able
to
bring
ourselves
to
like
a
strict
cadence
of
when
releases
happen,
we're
still
far
from
that,
but
we're,
but
as
we
work
toward
that,
if
we,
if
we
think
about
just
if
we
think
real
high
level
like
so
we're
we're,
I
don't
know
about
a
third
of
the
way
into
or
half
of
the
way
into
the
6-0
release.
A
So
what
from
sri
ctl,
it
was
getting
kubernetes
supported
through
all
the
system
commands
and
we're
mostly
on
the
other
side
of
that
the
vendo.
It
might
make
sense
as
soon
as
I'm
done
looking
at
this
here
for
us
to
share
the
mastery
ctl
spreadsheet,
we're
also
going
to
create
create
support
so
jubril
and
khan
and
and
piyush
and
other
people
that
I'm
not
going
to
name.
I've
been
working
on,
oh
any,
been
working
on
patterns
from
sri
ctl.
A
E
D
Yeah
sure
gk
we
can
target
0.6
okay.
A
E
Here,
uh-huh
just
to
do
a
quick
look
into
the
measurement
spreadsheet.
So
what
we
was
mentioning
about
was
the
compatibility
between
windows,
so
we
haven't
actually
confirmed
or
tested
all
these
commands
on
windows,
machine
so
yeah
it
might
be
best
if
if
anyone
can
volunteer
and
try
this
out
on
a
windows,
machine
and.
E
Yeah,
I
guess
that's
it
and
about
the
perf
command.
We
are
still
in
the
design,
specs
kind
of
thing,
and
we
still
need
to
complete
the
spec
before
we
create
it
into
issues.
E
I
I
have
left
left
the
link
in
the
chat
you
can
check
this
out.
I
think
it
would
be
best
if
you
we
just
look
into
the
another
agenda.
We
have
to
use
with
measuring
preflight
checks.
E
E
If
we
are
able
to
initialize
a
client
and
we
are
able
to
query
it
and
then
we
are
checking
if
the
kubernetes
version
is
actually
the
minimum
required
version
for
the
deployment
okay,
so
I
actually
created
a
spec
spec
here.
Let
me
share
it.
E
E
E
Oh
yes,
yes,
yes,
so,
like
the
first
part
was
to
verify
the
docker
presence
and
kubernetes,
and
I
told
you
about
the
minimum
version
check
so
verifying
docker
and
docker
compose
was
like.
I
was
trying
to
do
it
with
executing
the
command
docker,
ps
and
docker
compose
version.
If
they
are
not
returning
any
errors,
that
means
they
are
available
there
and
for
the
part
of
kubernetes
api.
E
I
was
trying
to
initialize
a
client
with
the
help
of
mesquite
cube
and
then
query
it
to
get
some
ports
or
some
data
about
pods
if
they
are
available
there.
That
means
I
if
this
does
not
return
an
error.
That
means
I
am
able
to
carry
the
kubernetes
away,
kubernetes
api
and
the
last
part
was.
There
was
some
a
third
part
also
included
in
kubernetes
api,
which
was
to
check
if
meschery
has
the
required
level
of
privileges.
E
That
is,
if
measurey
is
having
the
cluster
admin
role
assigned
in
the
kubernetes
kubernetes.
So
I
was
trying
to
get
like
how
we
can
actually
check
about
it.
So
abhishek
told
that
we
can
do
it
by
checking
measures
manifest
file,
but
I
was
quite
confused
if
we
are
already
declaring
it
as
a
cluster
admin,
then
do
we
actually
need
to
check
it
like
to
like
check
it
again.
E
So
I
haven't
like
I'm
not
sure
about
that,
so
I
skipped
it
for
a
while,
and
the
last
part
was
to
check
the
kubernetes
version,
so
I
am
also
not
aware
of
what
are
the
minimum
versions
for
these
like
for
a
measuring
deployment.
E
A
Great
nice
yeah
one
piece
of
feedback
with
respect
to
confirming
cluster
admin
role
is
there's
that
I
don't
know
what
the
the
go
client
calls
this,
but
from
the
command
line
from
cubectl
there's
this.
Can
I
so
it's
a
cubectl
auth.
A
Can
I
and
I'll
I
I'll,
put
yeah,
that's
just
probably
yeah.
Actually
what
you've
got
is
probably
good
enough
cannot
can
hyphen,
I
and,
as
you
go
to
explore
and
familiarize
with
that
yeah.
E
I
A
Compatibility
matrix
and,
and
actually
the
compatibility
matrix,
would
be
really
important
to
the
project
as
it
go
as
it
goes
forth.
It
is
not
well
is
not
well
maintained,
it's
not
programmatically,
updated
and
needs
to
be.
A
A
A
So
so
yeah
right
now,
you
might
end
up
hard
coding.
Something
like
it
has
to
be
this
for
actually
for
the
time
being,
just
the
retrieval
of
that
is
is
helpful.
The
the
challenge
that
you
will
have
in
this
command
is
that
you
know
ideally
what
you're,
showing
as
output
you're
saying.
Okay,
I'm
checking
on
this.
Oh
that
looks
good
check
mark
or
I'm
checking
on
this
that,
oh,
that
failed
x.
A
The
problem
with
my
answer
is
that
you're
like
well,
I'm
not
sure
if
I
should
put
a
check
or
an
x
and
so
for
the
time
being,
we'll
just
put
a
check,
because.
A
I
think
one
of
the
things
for
me
to
rationalize
in
my
own
head
and
for
everyone
else
that
considers
this
is
that
well,
there
are
the
value
of
such
a
check
is
both
prior
to
installing
meshri
is
to
go
just
like
you
install
a
piece
of
software
on
your
your
local
machine,
sometimes
you're
installing
it
and
it
says:
oh
well,
you
actually
don't
have
this
other
dependency.
A
Then
there's
value
in
it
post
deployment
as
well
like
ongoing,
is
measuring
healthy
and
there's
some
amount
of
consideration
for
if
we
get
this
done
fairly
well
and
so
p,
you
should
miss
some
of
what
I
said.
I'll
put
this
into
the
spec,
but
some
of
this
might
be
that
when
someone
runs
mastery,
ctl
systems
start,
it
might
be
appropriate
to
implicitly
do
a
pre-flight
check
depending
upon
how
involved
it
is
and
how
long
it
takes,
but
it
might
be
something
to
run
in
the
I
don't
know.
A
A
You
know
another
discussion
around
well
around
helping
troubleshoot
users,
environments
that
are
having
problems
if
they're
having
problems,
running
measuring
and
they're
in
the
slack
or
they're,
where
they're,
wherever
in
the
discuss
forum,
saying
they're
having
challenges
well,
there'll,
be
this
standard
set
of
questions
that
you'll
ask
them
you'll,
be
like
okay!
Well,
hey
what
version
of
mesh?
Are
you
running
what
version
of
measuring
ctl
what
version
of
the
adapters?
A
What's
your?
What
type
of
kubernetes
system
what
and
it's
just
like
agnosium
it
goes
on
for
a
while
when
you're
running
you're
running
istio.
What
version
of
that
whatever
and
it's
like
you
might
not
need
all
that
info
in
every
every
time
that
you're
helping
a
user,
but
you
may,
and
it's
pretty
commonplace
and
to
put
that
burden
on
the
user.
It
means
that
they
have
to
be
committed
and
really
wanting
to
get
a
solution,
because,
most
of
the
time
most
of
us
will
use
a
piece
of
software.
A
It
crashed
something
happened
if
you're
like
me,
you
say
a
curse
word
outside
of
your
mouth
and
then
keep
going,
and
maybe
you
don't
use
that
software
anymore
and
that
would
be
terrible
like
so.
My
point
is,
you
know
you
use
some
software
today,
it
crashes
and
it
says:
oh
wait.
We
detected
that
we
crashed.
Would
you
like
to
send
a
crash
report?
A
Oh
that's
great!
That's
the
type
of
thing
that
mastery
should
have,
whether
it's
automated
like
that
or
it's
just
a
manual
command-
that
people
can
run
to
collect
up
their
diagnostics
info
and
make
it
easy
for
them
to
open
up
a
ticket
to
paste
that
either
into
a
github
issue
or
into
a
disqus
forum
or
something.
I
G
G
A
I
almost
typed
it
out
as
fast
as
I
could
do.
Part
of
the
problem
is
that
you
know
it's
not
infrequent.
Well,
it's
less
and
less
frequent
over
time,
but
it
used
to
be
the
case
that
people
try
to
execute.
Docker
has
a
command
and
their
user
didn't
wasn't
permissioned
to
actually
run
docker,
and
so
that
would
be
an
issue
so
like,
as
as
an
example
of
like
why
you
want
to
it
kind
of
goes
back
to
what
push
was
saying
is
like
hey.
A
He
could
just
look
at
the
manifest
and
say
yeah,
but
it's
in
there
we're
good
well,
maybe,
but
to
actually
go
check
on
it
and
like
invoke
the
thing:
do
a
dry
run
or
do
a
f,
a
p,
an
actual
ping,
an
actual
so
to
actually
do
a
docker,
something
something
command
that
that
would
take
you
a
step
further
you're
right,
though
I
mean
you
could
have
a
series
of
levels
of
depth
of
interrogation.
Of
like
hey,
do
you
just
want
to
do
a
cursory
environment
check,
or
do
you
want
to
do
a
full
diagnostics?
A
And
so
maybe
you
take
them
through
the
rest
in
this
case,
that
would
be
over
engineering
for
us.
We
just
want
to
like
get
to
the
end
point
like,
but
that
type
of
a
concept
that
I
just
said
about
a
tiered
interrogation,
a
tiered,
like
a
full
discovery
versus
a
cursory,
a
light
check.