►
From YouTube: Meshery Development Meeting (Dec 1st, 2021)
Description
Meshery Development Meeting - December 1st, 2021
Join the community at https://layer5.io/community
Find Layer5 on:
GitHub: https://github.com/layer5io
Twitter: https://twitter.com/layer5
LinkedIn: https://www.linkedin.com/company/layer5
Docker Hub: https://hub.docker.com/u/layer5/
A
Hey
everyone:
let's
give
it
a
minute
to
let
others
join
in
on
the
call.
B
I
can
I
can.
I
can
say
that
there
is
a
problem,
but
I
don't
know
like
means.
I
will
investigate
more
like
on
that
thing
and
then
I
I
know
that
I
can
reproduce
the
problem
in
my
setup
so
with
the
latest
master
build
only
so,
but
I
need
to
because
I
am
new
to
this
helm
way
of
installing
right.
So
that's
that's
the
problem
with
me.
So
I'm
just
checking.
What's
the
problem
there,
maybe
yeah,
I
showed
it
to
ya.
A
B
A
C
B
A
A
All
right,
we
are
almost
three
after
on
the
call,
so,
let's,
let's
get
started
so
before
we
get
a
weekly
update
from
nitisha
or
let
me
welcome
all
of
you
to
the
measuring
development.
Call
today
is
the
1st
of
december
and
we
are
just
on
the
verge
of
the
v0.6
release.
So
v0.7
is
already
in
the
cards
here,
so
it's
pretty
exciting.
So
all
right
without
further
ado,
an
update
on
the
refactoring
machine,
yeah.
D
Okay,
so
for
for
people
who
are
new
to
this
call
so
actually
currently
like,
we
are
doing
architectural
overhaul
to
measure
ui,
so
we
are
bringing
in
redux
and
redux
observable
and
redux
and
all
those
stuff
so
yeah,
this
particular
dock
over
here.
It
explains
what
are
the
things
that
we're
bringing
in
and
what
are
the
reasons
behind
those
and
all
this
stuff.
So
if
you
want
to
dig
more
into
that,
you
should
definitely
look
into
this
particular
dock.
D
It's
it's
shared
in
the
meeting
minutes
as
well,
so
yeah
as
a
yeah
so
towards
that,
like
we
have
been,
people
have
been
working
on
that
and
yeah.
So
so
there
are
some
things
that
we
are
taking
it
step
by
step
and
first
step
of
that
is
to
create
a
globally
reusable
components
that
will
be
using
all
over
and
measure
ui
and
yeah.
We
have
a
list
of
components
that
are
to
be,
you
know,
created
so,
and
a
lot
of
people
have
been
working
on
it,
namely
aditya's.
D
So,
if
you're
into
ui
and
you're
into
react
and
you're,
if
you
think
comfortable
with
this,
then
you
should
probably
write
your
name
over
here
and
create
a
component
yeah
if
you
actually
to
get
what
exactly
has
done.
You
can
actually
in
this
in
this
particular
branch
over
here.
You
can
see
that
there
is
a
readme
file
which
actually
actually
explains
what
has
to
be
done
and
what
are
the
procedures?
What
are
the
procedures
that
we
are
following.
E
F
D
Yep
so
yeah
over
here
in
the
readme
file,
you
can
see
the
steps
that
actually
instruct
us
to
how
to
proceed
and
all
those
stuff,
so
you
can
look
into
it
and
you
can
also
ping
me
or
anybody
else
in
the
slack
channel
as
well
or
in
my
specifically
and
yeah
as
a
so
actually,
there
has
been
an
open
pr
that
that
was
my
scientin
that
addressed
some
of
the
that
actually
created
some
of
the
components
that
I
listed
over
here.
So
you
can
also
feel
free
to
comment.
D
I
mean
yeah
review
the
peers
that
are
open,
so
the
peers
that
are
concerning
this
will
be
will
have
a
redux
label
to
it.
So
you
can
actually
go,
go
to
the
pr
and
add
your
gum.
Add
your
thoughts
or
whatever
you
feel
like
to
it
so
yep.
There
are
some
progress
and
we
definitely
needed
some
more
people.
Working
on
that.
E
G
Just
a
quick
note
from
me
that
the
ui
re-architecture
and
some
redesigns
are,
as
it
sits
currently
for
the
upcoming
road
map.
It's
a
significant
portion
of
the
road
map,
and
so
so
it'll
be
a
consistent
theme
going
forward
and,
like
nation
said
it's
a
great
time
to
get
on
the
ground
floor
of
what's
more
or
less
a
rewrite
so
and
hopefully
a
few
new
designs
will
emerge.
G
A
Sayanthan
had
a
question
in
chat
that
he
he
finished
the
app
bar
and
loading
spinner.
So
yeah
is
there
anything
else
done
on
that
particular
area.
D
Yep
actually
like
yeah,
we,
the
actually
the
need,
was
quite
different.
That
is
the
loading
spinner
that
that
was
supposed
to
be
custom
to
misery
and
yeah.
Right
now
we
do
have
a
loading
spinner
that
was
created
in
the
pr,
but
yeah
they
asked
was
for
to
create
a
loading
spin
that
is
custom
to
misery,
but
that
was
actually
taken
directly
from
mystery
way.
So
that's
that
probably
involves
some
design.
D
So
that's
what
I
that's
why
I
actually
didn't
consider
it
so
and
the
app
bar,
I
guess,
like
okay,
I
didn't
actually
look
into
the
app
stuff,
so
I
might
have
to
take
a
look
into
it
again
and
I'll
get
back
to
you.
A
All
right,
thank
you.
Nitish!
Next
up
we
have
a
demo
of
seed
content
in
mashari
from
ashish,
so
ashish.
D
Okay,
so
in
the
latest
release
of
meshi,
we
have,
if
you
login,
as
a
nun
provider
or
if
then,
you
will
have
some
content
out
of
the
box
by
content,
I
mean
application
filters
and
patents
so,
instead
of
having
them
instead
of
having
nothing
here,
if
you
log
in
as
an
unprovider
you'll,
have
these
seed
basically
seed
content
basically
means
that
we
are
seeding
this
at
build
time
and
also
at
run
time.
Actually.
So
these
patterns
are
fetched
from
the
central
repository,
which
is
this
github
samples.
D
Basically,
this
sample
subdirectory
anything.
Any
pattern
that
is
added
here
will
be
reflected
on
the
ui
when
the
mesh
server
starts.
The
filters
are
fetched.
Currently,
the
filters
are
fetched
from
the
oh
from
from
an
older
release
of
where
we
have
all
the
filters
as
assets
we
do
need.
There
is
an
open
issue
in
washington
to
have
basically
to
have
a
workflow
that
does
this
for
us
that
basically
be
these
all
these
directories
that
you're
seeing
are
rust
code.
D
These
are
these
are
awesome,
filter
codes
we
need
to.
We
need
a
workflow
which,
basically,
when
we
make
a
release,
this
goes
ahead
and
compiles
them
into
wasn't
binaries
and
upload
them
as
assets.
So
we
can
have
a
bit
more
dynamicism,
so
at
every
new
release
we
can
have
new
awesome
filters.
Currently,
these
are
kind
of
the
open,
1.0
version
and
applications
are.
There
is
no
central
repo
for
applications.
D
These
are
faced
from
various
places,
for
example,
this
emoji
is,
for
example,
I
guess
the
image
service
the
book
info
is
from
steve
and
the
others
are
faced
from
different
places.
So
when
you
log
in
as
local
producer,
instead
of
having
nothing
you'll,
have
something
to
play
around
with
so
yeah,
that's
kind
of
the
final
update
on
this
seating
in
android,
where
seeding
basically
means
having
something
out
of
the
box.
That's.
G
Nice
so
ashish
for
those
that,
for
the
contributors
that
are
building
mescheri
locally.
D
That
will
have
it
too.
That's
why
there
is
logic
for
the
runtime
also.
So,
if
you
have,
if
you
basically
clone
messaging
and
because
we
do
the
seeding
at
two
places,
build
time
and
runtime
instead
of
measuring
image,
we
already
have
these
present.
But
if
you
are,
if
you
are
in
dev
environment,
you
have
cloned
meshri
and
you
you
of
course
do
not
have
you
do
not
have
content
in
dot
mesh
content
folder.
D
So
what
will
happen
is
that
when
message
server
starts,
there
are
different
recipes
right
now,
one
of
which
is,
if
you
run
simple,
make
run
fast,
which
is,
I
think,
what
most
people
run
when
they
are
running
this
server
locally
it
will,
it
will
on
runtime
it
will
download
all
all
of
these
things
and
see
them
here.
There
is
also
another
recipe
to
skip
these
downloads
and
there
is
a
recipe
to
completely
skip
the
seeding
at
all
times,
so
you
can,
depending
upon
what
your
dev
environment
is.
You
can
play
around.
G
Nice,
okay
yeah,
I
just
started
a
measuring
build
and
the
reason
I'm
asking
is
because
I
don't
think
that
I've
seen
a
build
that
I've
done
not
produce
errors
about
missing
content
that
might
be
expected
with
respect
to
filters
or,
but
it
kind
of
sounds
like
it's
not
that
that,
for
me,
are
you
on
upstream
yeah?
Well,
I
I'm
building
right
now,
so
we
will.
If
I
don't
get
any
errors,
then
this
is
a
red
herring.
G
G
Yeah,
so
yes,
here,
let
me
let
me
share
how
rude
of
me
not
to
share
so
okay,
so
I
just
I
just
built
registered
mastery
local
capabilities.
G
Mastery
server
current
channels
edge,
so
I
just
did
a
make,
run
local
which,
by
the
way
yeah
just
seeing
all
these,
can
you
do
maker
and
fast
yeah?
I
I
will,
although
I
wonder
if
we
shouldn't
get
rid
of
that
like
if
we're
not
being
consistent
and
building.
G
So
these
aren't
errors
like
compilation
errors.
These
are
intentional
errors
that
we're
we're
generating
right.
I'm
kind
of
explaining
this
for
some
of
the
others
on
the
call
we're
intentionally
generating
these
errors,
because
mastery
is
well
has
it
has
an
error
code,
documentation
framework
and
so
mesherie's
code
base
is
self-introspecting
and
it
looks
for.
G
G
But
but
yeah
so
not
necessarily
something
we
have
to
figure
out
in
this
moment,
but
just
something
to
be
aware
of.
If
it
sounds
like
this
isn't
supposed
to
be
there,
then
then,
then
we
probably
still
have
an
ongoing
issue.
G
Just
more
info
more
fuel
to
the
fire
more
info,
you
might
already
be
aware
of
this.
Actually
I
take
it
back.
I
saw
a
helm
chart
and
I
thought,
like:
oh
it's
a
it's
one
of
those.
It's
an
one
of
those
kinks
that
we're
still
working
through,
but
it's
not
I'm
not
running
kubernetes.
So
so
I
think
that's
an
expected
issue
so
cancel
that
red
rocks,
but
okay.
G
But
she's
just
fyi
I'll
send
this
over.
We
could
maybe
take
a
look
after
the
call
or.
D
This
is
not
expected
I'll
I'll
drop
in
I'll
drop
in
the
chat.
I'm
looking
into
it
right
now
sounds
good
sounds
good.
A
Yep,
let
me
just
share
my
screen
all
right.
A
So
push
is
on
the
call,
so
peo
was
fixing
the
missouri
ctl
dashboard
command
to
ensure
that
we
we
returned
the
correct,
endpoint
so
pusha
or
to
you,
could
you
could
you
brief
over
the
recent
update
the
recent
pr
that
you
made.
A
A
I
get
wrong
endpoint
for
some
reason
and
like
it
is
like
I
talked
to
with
kersh
about
this.
He
said
that
it
is
an
issue
from
musket
logic,
but
I
checked
but
like
it
wasn't
there
like.
There
was
no
issue
from
my
skate
side,
so
I
thought
why
it
is
doing
that,
so
I
thought
it
should.
It
might
be
an
issue
from
like,
while
we
are
deploying
things,
might
not
get
ready
as
quick
as
it
should
be.
So
that
is
why
we
get
wrong
endpoint,
maybe,
and
what
I
did
was
like.
A
The
main
purpose
of
dashboard
is
to
get
the
exact
end
point
of
measuring.
A
So
if
the
the
platform
is
darker,
we
can
actually
try
to
fetch
the
exact
endpoint
again
like
we
can
repeat
the
process
and
get
the
correct
endpoint,
and
I
did
it
so
I
just
added
the
same
logic
that
was
working
in
messy
systems
system
start
that
actually
fetches
the
like.
It
starts
the
cube
client.
Then
it
starts
to
get
the
answer
service
endpoint.
A
It
makes
some
tcp
checks
and
we
then
we
register
like
we
don't
register.
I
mean
we
change
the
endpoint
in
the
mesh
config,
so
it
it
works.
Now,
at
least
for
me,
other
people
who
work
on
linux
and
use
mini
cube.
They
can
also
check
it
will
always
return
the
correct
endpoint
if
system
start
fails
to
get
the
end
point
correct,
yeah.
I
use
the
word
word
correct
multiple
times.
Sorry,
if
it
was
confusing,
did
I
make
sense
any
like
I'm
not
sure.
G
Yeah,
so
there's
a
related,
so
hussein
can't
be
on
audio
at
the
moment.
I
don't,
I
don't
think,
but
she
had,
you
know
been
able.
She
hadn't
been
able
to
give
a
lot
of
time
to
this
in
the
past,
but
but
part
of
the
concern
here,
I
think
piyush
is
exactly
what
you're
saying,
which
is
there's
a
timing
issue,
which
is,
if
you're
deploying
measuring
on
kubernetes
there's
a
few
things
that
spin
up
it
can
take
up.
You
know
it
can
take
a
beat
before
mesh.
G
Reserver
is
ready
for
use
and
its
endpoint.
You
know
its
service
is
exposed
outside
of
your
kubernetes
cluster
and
right
now
I
if
memory
serves,
I
think,
we're.
G
G
So,
ideally,
what
we're
doing
is
you
know
it
doesn't
have
to
be
very
robust
logic,
but
it's
really
something
more
like
just
you
know
just
polling
every
couple
of
seconds
for
endpoint
readiness
before
we
exit
this
system
start
command
and
then
we
can,
you
know,
hopefully
most
of
the
time
say:
here's
your
you
know:
here's
where
mesri
is
at
here's,
the
url
and
actually
from
there.
We
can
actually
open
up
their
browser
to
the
right
place.
G
You
know,
assuming
it's
not
a
headless
environment
at
some
point
you
know
we'll
need
to
probably
let
the
user
know
that
they
could
do
a
control,
c
or
press.
Some
key
to
you
know,
stop
waiting
and
just
continue
at
some
point.
It
will
also
run
into
the
case
where
mastery,
the
system,
if
we
sit
there
polling,
the
endpoint,
may
never
get
successfully
exposed.
G
That
can
be,
you
know,
that's
because
there's
like
a
hundred
different
distributions
of
kubernetes
and
they
all
work
pretty
darn.
Similarly,
but
depending
upon
the
environment
in
which
kubernetes
is
deployed,
the
type
of
networking
that
occurs
around
exposing
a
service
external
to
the
cluster
is
a
bit
different,
and
so
it
might
be
the
case
that,
in
a
mini
cube
environment,
for
example,
maybe
we'll
have
some
amount
of
trouble
getting
the
endpoint
exposed
in
the
way
that
we
want
to.
We
should
keep
chipping
away
at
those
issues,
but
separate
from
that
you.
G
What
you're
doing
here
is
is
good,
it's
also
a
workaround
to
a
band-aid,
and
we
should
go
address
the
root
cause
the
root
cause
of.
Why
you're
having
to
do
this
is
because
we're
time
we're
not
we're,
we
need
to
improve
measuring
ctl
systems
start
it's
also
nice
to
have
this
so
that
you,
if
you've,
got
mastery
running
for
a
long
time,
you're
on
the
command
line.
G
Like
hey
where's,
that
end
point
again,
you
can
you
know
you
can
use
measure.
Ctl
system
commands
to
look
at
your
current
context
or
switch
context,
and
and
look
at
that
contexts.
G
Mesh
reserver,
you
know
mesh
re-endpoint,
it
might
have.
It
might
not
be
up
to
speed,
though,
because
there's
a
couple
of
race
conditions.
I
think
that
we
experience
as
people
go
to
deploy
that
you
know
as
they
make
their
mess
redeployment.
G
So
there's
from
my
perspective,
there's
certainly
merit
in
a
mastery
ctl
system,
dashboard
command,
and
I
think
this
is
fabulous
and
great
and
there's
a
couple
of
use
cases
for
it.
It's
one
of
the
use
cases
of
what
I
just
said.
Your
systems
are
running
for
a
month.
You
come
over
and
you
want
to
know
where
it's
at
you
might
check
the
config,
but
the
config
might
be
out
of
whack
out
of
sync,
and
so
an
external
command
like
this
is
great.
G
G
A
As
you
said,
the
main
issue
is
the
timing
issue.
However,
we
have
added
four
to
five
seconds:
wait
to
get
the
ports
pods
up.
Sometimes
it
takes
a
lot
more
time
than
that,
like
parts
still
gets
in
the
state
of
downloading
the
image
and
running
it,
so
I
feel
that
there
is
not
much
room
to
improve
on
the
system.
Starts
system
starts
part,
but
like
this
is
more
like.
A
We
can
inform
the
user
that
if
system
start
phase
to
get
the
exact
end
point,
you
can
run
this
command.
Just
as
like
you
said,
bandage.
G
Yep,
but
you
can
you,
but
internally
you
can
run
that
command
in
system
start
like,
like.
I
hear
where
you're
coming
from
so.
A
A
So
I
am
not
doing
anything
like
some
other
magic.
It's
just
the
same
code
from
the
system
slide.
A
Yeah
yeah,
so
I
like,
I
guess,
like
what
we
mentioned
didn't
come
across.
So
lee
was
mentioning
that
in
system
start
instead
of
us
having
to
sleep
a
certain
amount
of
time
and
then
check.
If
machine
reports
are
up,
we
can
just
keep
polling
like
something
similar
with
the
system
status.
So
we
can
call
the
system
status
command
internally
and
then
check
and
see
if
measuring
is
up
and
the
components
are
up.
A
Okay,
it's
it's
like
a
loop.
That's
that's
what
we
are
thinking!
Yep,
okay
and
maybe
we
should
also
have
a
condition
where,
like
the
users
can
exit
out
of
that
check,.
A
Okay,
one
thing
that
happens
sometimes
is
that
when
we
are
executing
this
part
get
service
endpoint,
we
do
get
an
end
point,
but
that
doesn't
have
the
exact
ip
it
has
local
host.
For
some
reason,
if
we
are
running
mini
cube
tunnel
before
deploying
machinery,
then
we
will
get
the
exact
ip
and
the
port
the
the
exact
ibm
report.
But
if
we
are
not
running
mini
cube
tunnel
and
then
it
will
return
a
local
host
and
the
port
for
some
reason.
A
I
So
what
happens
is
that,
if,
if
it
doesn't
figures
out
where
to
reach
it
to
it
defaults
to
localhost
and
that
that
does
work,
sometimes
like
in
docker
desktop
it's
a
it's
a
it's
a
fair
assumption
in
that
case,
but
when
it
does
not,
work
is
when,
when,
when
the
load,
balancer
pool
is
not
binded
to
the
to
the
to
the
local
host
or
something
that's
not
the
right
term,
but
basically
to
the
default
interface
of
your
system.
I
If
it's
not
binded
to
that
one,
then
that's
not
a
fair
assumption,
but
that
does
work
out
in
case
of
mini
cube
or
mini
cube
tunnel
or
docker
desktop
when
it
does
not
work
out
is
when
you
have
when,
when
you
are
using
another
load,
balancer
or
you
are
using
something
like
kind
with
metal,
lb
or
mini
cube
with
metal
lb.
I
In
that
case,
the
pool
would
belong
to
a
different
interface,
so
yeah,
that's
actually
the
bug
I
was
sort
of
referring
to
that
is
in
michigan
and
the
fix
to
that
would
be
to
fall
back
to
a
notepod.
But
actually
that
reason.
So
if
we
know
the
fix
and
why
it's
not
present
in
machine,
the
reason
is
actually
figuring
out,
the
node
id
is
sort
of
is
is
also
a
bit
tricky
because
it
kind
of
depends
how
a
node
was
created
and
what
the
node
is
all
about.
A
B
G
G
Let
me
direct
you
to
an
architecture
slide,
because
I
think
you
are
that's
a
good
suggestion.
That's
a
great
suggestion
and
thanks
for
mentioning
it,
it
is
wrong
or
like
what
I
mean
to
say
is
like
when
you
see
the
architecture
side
you'll
see
that
the
the
discussion
is
is
coming
from
a
different
vector,
okay,
which
is
which
is
the
ctl
and
how
it
communicates
with
pods
but
you're
right.
G
But
I
said
that
really
jokingly
about
being
wrong.
What
like
the
the
notion
that
there's
an
event-driven
aspect
to
this,
that's
a
beautiful.
What
a
beautiful
thing
like
like
hey!
How
long
is
it
gonna
take
to
come
up?
I
don't
know
like
depends
on
your
system
and
what
you're
doing
and
like,
and
you
know,
a
bunch
of
other
factors
and
whether
it's
raining
outside
or
not
and
so
like.
G
Wouldn't
it
be
nice
if
mastery
server
sent
a
signal
saying
I'm
ready,
you
know,
and
so
part
of
that,
like
you
know,
part
of
that
might
be,
I
we
don't
er
the
project.
I
don't
think
we
don't
necessarily
have
a
number.
We
don't
make
ex
geez.
G
A
G
We
can
use
some
standard
signals
to
say
yeah.
This
pod
is
ready
in
this
case.
It's
mesh
reservoir
is
ready
and
if
okay
so
yeah
one
of
the
things
to
consider
like
you've
done,
you
know
quite
a
bit
of
work
around
measuring
ctl
system
check
to
both
you
know
kind
of
do.
Pre-Fro,
pre-flight
readiness,
checking
of
like.
G
B
G
Routine
as
part
of
like
a
post
deployment
health
check,
if
there's
a
you
know,
basically
a
measuring
ctl
system
version
that
check
to
check
for
a
version
that
that
version
is
a
great
api
to
call
it's
a
very
simple
kind
of
a
ping
test
that
if
you
were
to
take
that
same
routine
and
reuse,
it
in
system
start
it.
This
is
a.
This
is
a
a
better
band-aid.
G
What
I'm
about
to
suggest
like
it
is
to
do
polling
of
you
know,
instead
of
what
we're
currently
doing
right
now,
which
is
sleeping
change
it
to
polling,
sit
there
and
and
look
to
see
when
we
can
access
the
thing
it
might
be.
In
some
cases,
like
you
just
said,
well,
we
would
never
exit
because
of
the
environment
and
certain
configurations
that
we
need
to
work
through.
We
need
to
work
on
over
time
figuring
those
out
and
just
identifying
that,
like
maybe
we
have
an.
Ultimately,
we
have
like
a
10-minute
timer.
G
G
Ultimately,
if
there
are
things
that
we
can
do
in
people's
environments
to
help
get
naturally
exposed
great,
if
you
look
at
other,
if
you
take
inspiration
from
other
commands,
other
ctls
in
this
ecosystem,
linkid
has
a
dashboard,
linkedin
dashboard
and
what
they
do
is
poor
forwarding,
and
I
I
don't
know,
that's
desirous,
that's
not
exactly
what
we're
that!
That's.
G
Actually,
I
think
what
I
that's
actually
what
I
had
in
mind
when
we,
I
suggested
a
mystery
ctl
system
dashboard
some
long
time
ago,
yeah
that
might
be
a
nice
flag
to
have
as
a
way
of
circumventing
of
doing
like
a
mini
cube
tunnel.
But
anyway
we
should
get
a
design
down
on
paper.
If
we
end
up
pulling,
I
think
that'll
that
suffices,
at
least
in
my
book.
If
we
can
get
an
event
driven
mechanism,
you
know
all
the
better.
We
should
account
for
health
checks
and
probes
as.
A
Anyone
else
have
comments
on
this
before
views
talks
about
this
is
the
other
topic
I
guess
like
yeah,
okay,
yeah.
Let's
talk
about
the
smp
smi,
okay,
pull
request,
which
should
be
for
requests.
A
A
Path,
url
path:
we
will
get
it
and
we
will
make
a
request
to
message
provider
and
it
will
return
it.
I
haven't
attached
screenshot
here,
but
it
works
okay
and
and
in
case
of
default
default
provider.
I
also
like
took
inspiration
from
pattern
as
a
and
that
works
from
the
default
site.
It
should.
A
I
have
to
check
the
default
one
and
the
local
provider,
the
local
provider
function,
because
I
was
not
able
to
test
it
out,
as
smi
conformance
was
not
working
for
me
on
my
system.
I
tried
everything,
but
it
was
not
working.
So
it
was
like.
I
will
check
when
a
new
machine
release
will
be
there,
but
yeah
remote
provider
works
and
the
local
provider
has
to
be
just
tested
out.
I
need
to
test
that.
I.
A
Mean
yeah
why
this
was
implemented.
This
was
implemented
because
of
this
issue.
A
H
C
A
E
A
Yeah
measuring
shader
mesh
view
yeah,
so
that
is
that
should
be
the
command
that
will
actually
display
all
the
conformance
result,
and
we
have
already
url
that
handles
that
returns
a
set
of
results,
but
there
was
none
and
there
was
not
any
which
could
return
only
one
result.
That's
why
I
raised
the
dpr
for
that.
A
Comments
here
early,
you
mentioned
something
with
respect
to
graphql
here.
G
Yeah,
I
think
we
just
made
another
rest
api
great
in
general,
we're
kind
of
leaning
away
from
rest
or
so
so.
There's
another
rest
api
made
here
for
a
very
valid
reason,
which
is
to
which
is
to
make
measuring
ctl,
more
robust,
and
so
that's
great.
So
push's
effort
is
fantastic.
G
It's
also
part
of
the
dot
seven
goals
to
try
to
get
entirely
away
from
rest
using
the
mesher
ui
client
in
order
to
graphql
using
mastery,
ctl
client
still
rest
and
for
the
foreseeable
future.
I
think
there's
like
a
post
1.0
suggestion
that
maybe
that
would
go
to
graphql
or
some
other
bi-directional.
I
Yeah,
actually,
we
are,
I
think,
the
reason
it
hasn't
been
completely
replaced
is
because
measured
logger
comes
so
you
need
to
create
an
instance
and
then
you
can
actually
start
using
it.
So
it's
available
in
the
hard
disk,
so
yeah
that's
a
place
where
we
can't
completely
replace
logos,
bloggers,
but
the
places
where
we,
for
example,
utils
function.
They
do
not
have
access
because
they
have
just
all
functions
mostly.
I
So
in
that
case
we
don't
have
access
to
instances
and
then
we
cannot
use
the
the
handler
and
in
the
center
we
create
an
another
instance
of
the
logger,
while
in
logarithms
it's
just
a
bit
easier.
It's
a
global
method,
it's
available
everywhere.
You
can
just
import
and
use
it.
Actually,
that's
the
only
reason
it
hasn't
been
completely
replaced.
Yet
in
machine
server.
G
I
So
replacing
it
in
the
hundreds
is
quite
easy:
it's
a
drop
in
replacement
up
mostly,
unless
until
some
someone
is
not
using
logos
or
tracing
something
which
we
don't
actually
so
it's
a
droplet
replacement.
So
it's
easy
to
replace
and
handles.
We
can.
We
should
start
doing
it
today,
while
in
utils
functions
like
the
helper
functions,
that's
a
place
where,
basically
that's
where
we
will
have
to
actually
think
what
we
actually
do.
I
Should
we
create
new
instances
in
every
function,
which
would
be
maybe
not
ideal
or
and
should
we
should
mesh
kit
start
exposing
some
global
loggers
which
is
not
binded
to
it
will
always
be
binded
to
our
instance,
but
not
bind
it
to
an
instance
that
is
local
to
the
function
or
something
so
yeah.
That's
something
that
we
will
have
to
figure
out.
G
G
Yeah
sure,
no,
no,
no
sweat,
we'll
we'll
take
it
to
slack
or
or
send
another
message.
A
Thank
you,
pierre,
before
before
we
move
on
to
the
next
topic,
a
quick
note
that
maybe
we
should
follow
the
ux
of
measuring
ctl
perf,
similar
ux
to
measuring
ctl
buff
within
the
measuring
ctl
mesh
command,
like
also
commands
like
results,
could
work.
I
guess
that
spec,
which
you
showed
before
was
really
outdated,
like
it
was
way
before
the
we
changed
the
ux
of
the
performance,
so
maybe
like
you
could
look
into
it
from
a
perspective
of
how
we
designed
puff
command.
That
would
be
really
good.
I
guess.
C
G
G
And
so,
if
you
would,
will
you
think
about
opening
up
an
epic
or
maybe
a
series
of
issues
just
to
make
sure
that
we're
tracking
toward
consistency.
I
So
I
okay,
let
me
first
share
the
context
because
probably
it
hasn't
been
discussed
yet
so
history
recently
added
support
for
wasn't
plug-in.
What
was
in
plug-in
is
wasn't
plug-in
is
basically
awesome.
Plug-In
is
basically
the
ability
to
extend
functionality
of
istio's
data
plane
by
dropping
in
either
oci
or
some
file
file
is
actually
binary
or
or
loading
wasm
files
from
wasn't
files
from
hdb
or
https
based
endpoints.
I
So
this
is
basically
wasn't
plugin,
because
stills
drop
support
has
basically
added
the
support
into
sdo
version
1.12
in
greater,
so
it's
mystery
also
has
support
for
the
same,
because
mystery
is
designed
and
way
that
it
is
completely
dynamic.
So,
unfortunately,
we
didn't
have
to
do
anything
to
add
this
particular
support.
It
is
just
there
as
soon
as
it
it
was
dropped
in.
I
If
you
are
running
mystery,
adapter,
which
is
not
more
than
a
month
hold
or
something
you
will
also
have
support
for
the
same,
you
won't
have
to
do
anything
new.
So
if
you,
for
example,
if
you
are
going
to
edit
a
editor
pattern,
you
should
be
able
to
access
that.
I
Okay,
thanks,
you
should
be
able
to
access
that
particular
component
in
here
under
1.12,
because
mission
supports
multiple
versions.
1.12
is
one
of
them.
This
is
the
one
and
you
would
be
able
to
access
that
particular
component
in
here
wasn't
plugin.
This
is
the
content
you
can
configure
it,
so
basically
no
code
was
added,
and
yet
we
have
support
for
this.
What
exactly
this
is.
Let
me
show
examples.
Function
show
demo,
so
this
is
my
keyboards
cluster.
I
I
have
some
issue
components
running.
There
is
no
sd
installed.
There
is
no
other
things
running
if
I
show
the
pattern.
This
is
the
button.
This
is
a
design.
If
you
will.
What
I'm
trying
to
do
here
is
I'm
actually
creating
a
namespace
namespace
is
going,
is
called
test,
I'm
also
creating
I'm
also
crea.
I'm
also
asking
mystery
to
deploy
seo
service
mesh.
For
me,
I'm
I'm
asking
it
to
deploy
version
1.12.0
because
that's
the
version
it
works
on,
I
mean
wasn't.
I
Plugin
is
supposed
to
be
working
on.
I'm
also
asking
mystery
to
ensure
that
automatic
cycle
injection
is
enabled
on
certain
image
spaces
what
those
name
spaces
are.
So
this
is
a
filler
expression.
What
I'm
saying
is
that
hey
go
ahead
and
and
basically
replace
this
particular
string
with
it
with
whatever
is
present
here.
That
is
what
is
going
to
resolve
to
ref.
Is
identifier
sort
of
I'm
saying,
go
to
services
and
then
go
to
ns.sample
and
then
go
to
name
whatever
there
is
please
replace
that
here.
This
is.
I
This
is
probably
like,
although
this
is
quite
long,
the
reason
you
may
want
to
do
it
is
because
first,
you
can
just
see
so.
Basically
this
is
a
variable.
You
can
change
this
thing
and
the
every
place
where
this
was
this
is
being
used.
That
will
automatically
be
changed
and
the
second
is
because
you
can't
have
typos
you
can
type
in
test
here,
and
maybe
you
will
type
in
maybe
ax.
Maybe
you
will
have
some
kind
of
typo
in
here.
This
is
completely
avoided.
I
If
you
use
for
the
expressions,
the
reason
is
that
it
is
statically
checked.
You
cannot
have
undefined
references
in
here
so
yeah.
This
is
a
phil
expression.
Just
was
talking
about
it
now.
The
third
step
is
actually
I'm
saying
that
hey,
please
deploy
an
application
for
me.
What
that
application
is
that
application
is
simple,
indirect
server
running
on
port
80,
I'm
adding
a
label
to
it
that
this
label
is
going
to
be
later
on
used.
I
So
this
is.
This
is
all
that
you
actually
require
to
run
an
application
using
mystery.
So
if
someone
has
actually
deployed
up
any
application
to
kubernetes,
you
may
be
formulated
how
long
those
yamas
actually
get
your
configure
deployment
inside
deployment.
You
have
a
template,
spec
and
you
you
basically
configure
your
ports.
I
A
lot
of
things
are
actually
duplicates
like
duplicate
selectors,
because,
obviously,
if
you
most
of
the
time,
if
you
are
actually
creating
a
deployment
and
that
deployment
is
going
to
create
a
replica
that
replicas,
it
is
actually
supposed
to
point
to
the
port.
So
that's
basically
there's
a
lot
of
duplication.
That's
something
we
sort
of
come
to
with
this
particular
custom
construct.
That
is,
you
can
just
write
this
thing
in
it
will
deploy
application.
I
I'm
also
asking
that
this,
but
I'm
also
specifying
that
this
particular
thing
actually
depends
on
two
things
that
is.sample
in
sdo.
What
this
is
is
basically,
basically
I'm
saying
that
please
ensure
that
you
actually
execute
this
particular
step.
If,
and
only
if
this
particular
thing,
this
particular
step
in
this
particular
step
has
succeeded
successfully.
Basically,
this
tip
completed
successfully
without
any
errors,
so
this
step
is
going
to
fade
for
these
two
to
finish,
these
two
steps
would
be
basically
will
be
executed
partly,
and
then
there
is
this
particular
step.
I
This
is
this
step
in
this
step
is
also
going
to
be
executed,
properly
line,
15
and
932.
These
steps
are
also
going
to
be
executed,
partly
because
they
do
not
have
any
interdependency,
while
both
of
them
are
going
to
wait
for
these
two
to
finish,
because
both
of
them
have
these
two
as
listed
their
dependencies.
What
I'm
doing
in
here
this
is
the
washington
thing.
I
This
is
the
this
is
the
new
thing
that
is
still
wrong,
and
that
is
I'm
saying
that
hey
the
type
is
wasn't
plugin
dot
istio
the
name
is
actually
sample.
You
can
actually
call
it
anything,
the
name
spaces
again,
I'm
using
the
same
filler
expression,
because
I
want
it
to
be
deployed.
I
I
wanted
it
to
be
provisioned
in
the
same
space
that
that
my
rest
of
the
things
are
like
like
automatic
cycle
injection
is
enabled
and
enabling
automatic
cycle
injection
for
this
particular
thing
is
extremely
important,
because
if
you
don't
have
data
plane,
then
where
you
would
pro
how
you
would
drop
the
version,
and
that
is
where
automatic
cycle
injection
is
being
enabled
on
that
namespace.
I'm
saying
that
deploy
this
particular
thing
in
this
name
space.
I'm
saying
that
please
put
in
this
particular
vosm
file.
That
is
a
basic
auth.
I
It
wasn't
file,
there's
a
piece
of
author
version
file
and
I'm
saying
that
drop
it
in
every
every
data
plane
where
drop
it
in
every
data
plane
where
this
particular
label
is
attached,
and
that
is
this
one.
If
you
want
this
better
thing
to
be
dropped
in
other
places,
also,
you
can
actually
create
you
can
actually
copy
it
paste
it
and
add
the
same
label
and
it
will
be
available
there.
This
is
not
something
you
need
to
measure.
Obviously,
then,
there's
plug-in
config.
This
is
basic
config.
I
I'm
saying
that
enable
basic
authentication
on
this
prefix
then,
basically,
when
there
is
when
endpoint
is
just
slash
and
the
request
method
is
git
and
the
credentials
are
actually
this.
This
is
nothing
but
if
you
were
decoded,
this
is
admin
three
as
the
username
admin
three
as
the
password.
So
let's
go
ahead
and
try
to
apply
this
button.
I
Okay,
I
don't
think
it's
failing.
So,
let's
watch
the
state
of
the
cluster.
I
Actually,
instead
of
watching
the
boring
serial
cluster,
you
can
actually
also
watch
things
coming
up
here.
The
reason
I
have
multiple
of
these
listed
is
because,
if
you
actually
go
ahead
and
see
the
status
in
here,
you
would
also
see
that
this
particular
thing
was
dominated.
So
the
older
data
is
not
deleted.
Yet
at
the
moment,
probably,
we
should
show
it.
We
should
not
show
it
the
reason.
I
The
the
usefulness
of
this
video
thing,
the
steel
data
is
that
you
can
actually
check
that
how
this
particular
thing
died
and
you
can
actually
see.
I
hope
the
font
is
as
well
it's
quite
small
but
saying
that
it
was
terminated,
and
this
is
what
exactly
happened.
So,
let's
see
if
things
are
ready,
so
I
asked
for
istio
deployment
and
yeah
apologies.
G
Just
to
hit
you
time
check,
maybe
we've
got.
Hopefully
we've
got
to
help.
We've
got
a
helm
issue
that
kind
of
is
affecting
people
about,
so
apologies
would
crush.
I
didn't
mean
to
cut
you
off
cold
and
don't
please.
I
Yep,
so
if
you
have
run
engineering
server
on
your
system,
you
definitely
don't
get
authentication.
The
reason
this
authentication
was
automatically
added
was
because
wasn't
plugin
was
added
to
this
particular
thing.
Now,
if
you
would
type
in
admin
three
admin,
three,
that's
a
password.
I
said
you
will
actually
be
able
to
access
the
and
access
the
particular
service
so
yeah.
Basically,
this
was
the
support
for
wasn't
plugin
in
mystery,
without
pretty
much
doing
anything
because
of
the
way
we
have
designed
a
system
so
yeah.
A
Thank
you
utkar,
so
I
just
want
to
point
out
that
we
discuss
a
lot
of
awesome
related
things
on
the
awesome
filters
call
on
monday,
so
that
is
a
good
place
to
jump
in
and
discuss
on
everything
related
to
awesome
and
all
the
initiatives
we
have
here
in
missouri
so
before
we,
since
we
are
very
long
time,
let's
move
on
to
the
next
topic
from
praveena,
so
there
is
a
bug
when
measuring
server
tries
to
install
the
measuring
operator,
which
we
recently
switched
to
helm,
charts
so
praveen
like.
B
B
B
B
Yeah,
so
if
I
go
here
and
try
to
like
like
upload
the
file
right,
aml
file,
so
basically
yeah,
so
basically
what
it
is.
It
is
detecting
the
kubernetes
cluster.
So
basically,
if
I
say
yes,
so
I
can
able
to
ping
to
the
pin
to
my
cluster
like,
which
is,
which
is
all
out
of
cluster
right
so
out
of
cluster
deployment.
What
I'm
doing
so!
This
is
a
machine
like
oca
instance
where
I'm
I
want
to
run
my
operators
there.
B
So
I'm
just
experimenting
with
that,
but
when
I
see
like
when
I
want
to
turn
on
the
operator
so
that
time
what
happens
this
like
whenever
we
do
a
helm,
install
right
so
operator
comes
when
we
do
a
help
install
for
this
operators
that
time
and
seeing
this
problem
certificate
signed
by
unknown
authority,
something
it
is
showing
I
for
just
to
confirm
what
I
did
was.
I
replaced
that
comment
with
the
yaml
file
commit.
So
then
it
worked
fine
and
to
prove
that
this
is
not
the
setup
problem.
B
B
Yeah,
I'm
just
I'm
new
to
this
helm
also,
so
I'm
just
looking
and
trying
to
find
out
like
how
to
solve
this
problem
so
because
I
want
to,
I
want
to
see
the
out
of
cluster
like
how
it
works
and
all
those
things,
therefore,
I'm
just
setting
up,
so
I
found
this
problem
yeah.
This
is
what
I
wanted
to
show.
C
E
B
A
Praveen,
you
have
already
uploaded
the
cube
config
right.
B
A
A
Okay,
I'm
not
sure.
E
A
Can
you
run
messy
city
system
login?
First,
let's
try
if
we
can
reach
the
imagery
adapters.
B
No
measuring
ctl,
I
how
I
ran
was
I
maybe
I'm
new
like
correct
me
if
I'm
wrong
here
also.
So
what
I
did
was,
I
just
did
a
git
clone
of
a
measuring
server
and
I
installed
like
a
measuring
local
run
right,
so
normally
so
I
do.
I
did
that
that
and
then
I
installed
my
git
like
steve,
adapter
and
git
clone
that
ht
adapter
and
followed
some
procedure
to
install
the
hta
doctor.
That's
my
setup!
Actually
I
don't
have
anything
else
on
that.
A
No,
I
don't.
I
don't
think
that
that
requires
that,
like
there
is
no
need
to
do
that,
I
was
just
like
the
main
goal
was
if
we
were
able
to
check
the
queue
when
it's
api,
we
can.
B
B
No,
but
you
can
just
click
if
you
do
this
right,
so
it's
a
ping
command
right.
You
are
sending
a
tcp
probe
to
your
cluster
right
so
which
is
working.
G
Yep
great
good-
maybe
I
should
jump
in
I
so
the
the
challenge,
that's
so
prevent
to
kind
of
repeat
the
state
of
the
your
system
is
you're
using
oracle
cloud
infrastructure.
You've
got
the
kubernetes
this
cluster.
That
you're
running
there
is
is
that
running?
Is
that
a
managed
service
that
you're
using
or
is
that
just
in
vms
over
there.
G
An
adapter
you
built
mashery
server,
you're
running
as
you
built
mesh
reservoir
and
the
adapter.
You
did
so
locally
on
your
local
machine,
the
kubernetes,
the
cube
config
that
you
have
on
your
local
machine
points
to
your
managed
kubernetes
service
from
you
know,
oci,
correct
and
so
mesri
server
is
up
and
running
successfully
locally.
The
ui
client
is
present.
You're
able
to
log
in
you've
got
your
master
ctl
client
as
well
you're
able
to
run
a
few
commands
there.
G
The
challenge,
though,
is
that
your
message,
server,
that's
sitting
locally
on
your
laptop
or
your
local
dev
system
is
not
in
you
know
like
in
accordance
in
accordance
with
what
you're,
seeing
here,
it's
like.
Well,
hey
the
operator
got
deployed
so
did
mesh
sync.
So
do
the
two
controllers,
mesh
sync
and
nats
and
they're
both
running
in
kubernetes
they're
running
in
kubernetes,
because
mesh
reserver
that's
running
on
your
local
system.
It
was
able
to
successfully
connect
to
that
kubernetes
instance.
B
Think
I
think,
let
me
let
me
rephrase
my
question.
I
like
a
problem
early.
So
what
what's
the
problem
is
earlier
when
I
was
using
this
machinic
code
base,
which
was
using
a
normal
ml
file
based
of
approach?
I
just
upload
my
ml
file.
It
used
to
connect
and
bring
up
the
operator
and
sync
and
ads
in
my
managed
cluster.
Okay.
I
Yeah
there's
gas,
there's
no
doubt
so
proven.
Do
you
so
your
cube?
Config
is
also
in
dot
cube
directory
right
or
is
it
somewhere
else.
I
No
actually,
the
hem
client
that
we
use
internet
that
so
that
defaults
to
that
particular
directory.
So
it's
it's
a
internal
implementation.
That
is
yes,
it
does
require
it
by
default,
or
what
you
can
do
is
if,
if
it
is
in
different
directory,
you
can
actually
specify
cue
point
and
that
is
picked
by
that
helm,
client
and
that
that
is
the
reason
sort
of.
F
B
Are
you
suggesting
to
keep
this
cube
config
file
into
the
cube
directory
or
merge
merge
those
things
I'm
going
like.
I
Yeah,
so
what
you
can,
what
should
work
for
you
if
you're
running
make
run
local?
I
think
that's
how
you
spend
up
your
memory
server.
Can
you
try
adding
cube
config
equals
so
basically
add
the
environmental
variable
cube
point
v
equals
wherever.
I
B
I
already
did
that,
like
I
already
did
that
I
can
use
the
contacts
and
everything,
but
still
this
this
one
is
not
working.
Okay,
okay,
I
can.
I
B
G
Just
as
a
quick
housekeeping
item
to
prevent
point-
hey,
it's
been
a
great
measuring
dev,
call
very
nice
to
see
you
all.
We
have,
I
think
the
build
and
release
meeting.
Maybe
tomorrow
I
have
to
go.
Look
at
the
calendar.
We
have
the
community
call
on
friday
we're
going
to
send
out
an
email
as
a
request
for
comment
and
all
the
the
upcoming
road
map
and
so
thus
concludes
today's
dev
meeting.
There
are
a
number
of
us
that
are
interested
in
what
in
helping
prana
or
preven
jeez.
G
F
B
G
As
a
point
of
clarification,
just
from
for
me
a
question
for
foodkarsh
and
maybe
for
rudraksha
as
well
as
a
couple
of
individuals
that
have
kind
of
been
there's
a
number
of
people,
but
as
two
people
that
were
kind
of
focused
in
this
area
about
how
mesh
reserver
is
you're
using
internally
using
helm
to
deploy
the
mesh
free
operator
boot
cars.
You
were
noting
that
the
default
configuration
internally
is
that
mesh
reservoir
will
look
to
kind
of
the
standard.
G
Location
of
you
know
the
current
user,
their
hidden
folder,
their
hidden
dot,
cube
folder
and
look
for
the
config
file.
But
I
think
you
were
also
saying,
while
that's
the
default
like
like
the
fact
that
it's
the
default
also
implies.
I
It's
not
I
mean
so
the
mesh
kit
function.
I
think
we
can't
go
into
detail.
A
measured
function,
apply
him
chart
that
uses
him
client
and
him
client.
It's
the
requirement
of
him
that
cube
conflict
should
actually
point
to
the
cue
config,
so
that
can
use
it
internally.
It
doesn't
accept
so
it
doesn't
accept
clients
it's
from
outside,
so
it
creates
it
internally,
and
that
requires
coupon
for
to
point
to
the
right
direction.
I
If
it
does
not
so
in
adapters,
that's
the
reason
we
actually
set
up
cube,
config
manually
and
and
and
what
can
happen.
The
reason
I
was
sort
of
pointing
to
this
direction
is
because
this
ping
test
that
is
actually
a
server
instead
of
instead
of
being
dc
pro,
we
actually
send
a
we
actually
retrieve
version
of
of
a
kubernetes
api
server,
and
that
happens.
The
the
method
that
gets
used
internally
is
different.
While
this
apply
thing
is
actually
this
uses
him
client
and
then
client
use
uses,
cube
config.
I
So
the
reason
so
yeah
that
that
was
mostly
the
reason
I
was
sort
of
yeah
I
wanted.
I
wanted
to
know
if
cube
config
is
actually
pointed
to
the
right
direction,
because
that's
mostly
the
only
thing
other
than
networking.
Obviously
that
differs.
B
So
basically,
my
cube
config
is
in
some
other
directory,
not
in
the
that's
for
sure.
I
Okay,
so.
F
B
B
I
can
set
a
cube,
config
equal
to
environmental
variable
like
to
my
download
this
one
and
then
I
can
use
it.
I
can
do
a
helmet
like
whatever
it
is
right
so
set
configure
context,
use
context.
If
you
do
so,
then
I
can
do
it.
B
I
B
I
Yep,
but
probably
just
move
it
to
a
backup
and
copy
this
one
and
place
it
in
it,
basically
replace
it.
Okay,
I
understand.
F
F
I
This,
the
in
the
batch
and
yeah
try
reloading
the
page.
F
I
B
B
B
B
B
Anybody
knows
how
to
move
this
top
one
right.
B
B
I
B
I
I
B
Just
clean
up
my
things
and
then
start
it
again.
I
think
if
you
keep
my
cube
config
in
the
default,
it
works
fine.
So
when
we
are,
when
I'm
using
helm,
the
config
file
was,
I
can
keep
it
any
sorry.
When
I,
when
I
using
yaml
file,
I
can
keep
it
as
anywhere,
but
now
it's
a
mandatory
right
to
keep
it
in
a
local
directory.
Yeah.
I
Actually,
that's
that's
mostly
true,
but
only
for
the
local
environment,
if
you
would
have
deployed
it
in
up,
let's
say
in
a
container
or
inside
plus
inside
kubernetes
cluster.
This
is
not
a
like.
You
wouldn't
have
faced
this
issue.
The
reason
probably
we
haven't
stumbled
upon
it.
Although
this
this
is
unwants,
it
should
be
documented
yeah.
It's.
B
I
So
you
might
have
tried
uploading
your
cube
config
inside.
C
I
Yeah,
actually
so,
instead
of
uploading,
so
instead
of
uploading,
so
what
so?
There
are
two
things:
probably
you
can
do.
One
is
instead
of
provisioning,
it
in
kind.
You
can
directly
provision
it
into
your
this
particular,
which
you
want
to
point
to
that's
in
cluster
deployment.
The
second
is
actually,
then
you
will
have
to
actually
flatten
your
config
so
that
we
ensure
that
there
is
no
reference
to
other
files
or
something
so
there
are.
I
Although
this
I
mean
mystery
does
flawlessly
work,
some
kind,
because
that's
the
that's
my
development
environment,
one
once
there
also
is
that
you
need
to
have
a
load
balancer,
so
something
like
metal
will
be
present
there,
because
it's
it's
a
it's
a
it's
a
requirement
for
multiple
components
of
mystery,
that
load
balancer.
So
the
ability
to
provision
load
balances,
that's
sort
of
a
requirement.
B
I
think
I'm
not
understanding,
maybe
I'm
getting
confused
here.
So
basically,
what
I'm
saying
was
so
if
we
have
a
ui
to
provision
this
cubeconfig
file,
I
can
provision
from
anywhere
right.
So
that's
my
question:
why
do
I
need
to
copy
to
dot
coupon
if
help
support
it?
So
suppose
you
as
you're
saying
so
helm
wanted
you
to
keep
that
in
a
dot
home
directory
itself.
B
I
Yeah
actually
yeah
almost
yeah,
that's
almost
right.
So
what
right
now
happens
is
yeah
you,
you
are
actually
almost
right
so
right
now,
what
happens
is
that
the
helm
client
that
we
use
internally,
it's
a
good
sdk
that
we
use
that
requires
a
cube,
config
environment
available
to
present.
If
it
is
not
present,
it
will
go
ahead
and
look
into
dot
cube,
slash,
config
file.
It
will
try
to
pick
that
thing
up
right
now.
What
was
happening
was
so
in
adapters.
I
We
use
the
same
thing,
but
we
set
up
a
cube,
config
environmental
variable
in
internally,
not
not
to
the
parent
process,
which
we
cannot
do,
but
we
do
set
up
that
environmental
variable,
but
in
memory
server
we
are
not
because
usually
it
is
the
case
that
is
it.
Is
it
the
coupon
fit
does
point
to
that
thing?
I
The
reason
I
did
not
suggest
to
like
the
reason
I
did
not
say,
is
to
add
a
cube,
config
environmental
variable
and
then
do
make
run,
make
run
local
because
you
would
have
faced
different
issues
in
that
one
case.
There
is
a
bit
inconsistency
in
how
hem
client
it
wants
cube
config
to
be
set
in
the
way
mystery
server
wants
it
to
be
set
up.
I
Machine
server
wants
it
to
be
a
directory,
while
him
client
wants
to
be
a
file,
a
config
file,
so
yeah,
so
that
was
that
there
are
a
few
new
ones.
I
think.
B
I
think,
according
to
me
right
according
to
me,
according
to
what
you're,
explaining
I'm
I'm
getting
the
things
now.
So,
basically,
what
we
should
do
is
we
should
get
the
environmental
variable
before
we
install
the
helpline
helm
upload,
and
then
we
need
to
set
that
environmental
variable
in
the
code
itself
right.
I
Yeah,
actually,
that
that's
exactly
right,
that's
something
that
probably
we
missed
in
machine
server
while
we
do
it
in
adapters,
machine
server
does
need
to
point
it.
Basically
it
it
needs
to
synchronize
the
cube
config
that
is
local
to
machine
server
with
the
handling,
and
that
is
the
synchronization.
That
probably,
I
think
we
didn't
miss
out
in
machine
server.
G
All
right
who
who's
crack
pravindi,
would
you
mind
popping
open
just
a
a
quick
issue?
You
don't
have
to
describe
it
necessarily
very
well,
so
that
others
can
others
can
fill
in
kind
of
the
details,
but
it
sounds
like
it
sounds
like
this
is
something
we
should
track
right.
B
I'll
try
to
work
on
it.
Maybe
so
let
me
check
if
I
can
fix
it.
So
basically,
you
know
I
understood
the
problem,
so
I
got
the
gist
now.
So,
basically,
what
what
he's
telling
me
is
like
our
cube,
config
file
right?
We
need
to
set
the
environmental
for
help
to
work
so
operator
helm
installed
to
work
right.
We
need
to
have
an
environmental
variable,
but
by
default,
if
you
don't
set
anything,
then
it
takes
the
dot
cube.
Config
file
like
that
is
that's
where
the
helm
is
working.
That's
what
he's
doing?
B
G
That
sounds
yeah.
That
sounds
great
sounds
like.
We
could
also
just
confirm
that
this
is
in
the
docs
somewhere.
There's,
there's
kind
of
you
know,
there's
some
deployment
docs
like
there's
an
advanced
deployments.
I
don't
know
that
this
necessary
constitutes
advanced,
but
but
advanced,
meaning
kind
of
custom
deployments
is
probably
the
right.
A
good
area
out.
B
Of
cluster
deployment
right
out
of
cluster
deployment,
so
where
the
cube
config
file
is
not
in
the
like
home,
cube
cube
directory,
so
those
ammo
files
are
it's
not
working.
Those
are
not.
E
B
No,
no!
This
is
not
the
problem,
so
let
me
rephrase
again
right,
so
I
can
clean
up
my
measuring
server.
So
still
it
will
not
work.
Even
if
I
do
a
fresh,
install
right
and
even
if
I
do
in
the
first
install
it
will
not
work,
even
if
I
don't
have
an
operator
running
in
my
out
of
cluster
deployment
mode
right.
So
if
I
don't
have
anything
there,
also,
it
will
not
run
unless
I
copy
this
amal
cube
config
file,
what
I
have
into
the
home
directory
of
dot
q
file.
Okay.
B
B
Okay,
but
if
I
keep
it
in
a
cube
con
file,
the
same
ml
file,
I
copy
replaced
my
dot
cube
con
file
with
this
cube
config
file.
What
I
was
uploading
here
that
time
what
happens
measuring
picks
up
by
default
helm
also
uses
the
same
environmental
variable
and
it
starts
installing
the
operator
on
the
cluster.
B
But
if
I
upload
in
some
other
directory,
I
think
helm
is
unable
to
set
the
take
the
environmental
variable
configuration.
Maybe
we
need
to
set
the
cube
config
to
this
and
that
and
then,
like
you
need
to
apply
the
help.
G
Nice
very
good,
okay,
yeah,
very
good.
Well,
it
sounds
like
we'll
pop,
probably
there'll,
be
a
couple
of
issues
opened
and
preven.
It
sounds
that
sounds
great
man.
It
sounds
like
you
might.
You
might
try
your
hand
at
this
item.
So
that's
fantastic
yeah.
B
And
one
more
question
means
like
if
I
want
to
set
set
things
up
for
like
some
problems
I
have
in
setting
up
like
home
to
contact
like
I
can
just
discuss
only
clack
or
discuss.
That's
normally
confuses
me
like
a
lot.
G
It
is
really
where
we're
trying
to
push
all
of
us,
like,
I
wouldn't
say
people,
but
I
would
say
you
know
all
of
us
to
to
go
use
it
because,
like
utkarsh
can
tell
you
well
as
an
example,
I
just
since
he's
here
that,
like
there's
some
air,
there's
we're
going
on
account
of
me
having
repeated
myself
61
times
verbally
on
one
teacher
like
and,
and
so
my
the
point
is,
is
like
we
lose
the
history
of
slack
and
there's
a
ton
of
awesome
things
that
have
been
there's
one
I'll
point
it
out
in
slack,
but
if
we
ever
have
a
kind
of
a
long
conversation,
there's
this
cool
little
slash
command.
G
That
I'll
point
out
in
slack
that
we
could
use
to
just
so.
Basically,
we
want
to
like
persist
things
we
want
to
yeah.
Okay,
there's
actually
a
lot
of
maintainers
measuring
maintainers
and
you
know
some
of
them
spend
a
day
a
week
or
what
have
you
and,
and
so
they
missed
the
other
six
days
of,
like
all
the
content
that
went
by,
and
so
they
don't
even
like
so
yeah
but
but
yeah.
B
G
Yeah
they
will,
if
they
another
thing,
that's
been
a
real
challenge
for
us
is
like
do
your
be,
and
this
is
this-
we've
said
this
like
to
a
million
people.
So
don't
this
isn't
aimed
at
you
specifically,
but
just
like
yeah
some
conversations.
You
have
to
do
a
direct
message
because
it's
just
a
kind
of
a
private
you
know,
but
to
the
best
of
your
ability
to
do
it
publicly
publicly
publicly,
because
everybody
benefits.
B
Thanks
for
spending
more
time,
guys
like
thanks
yeah
thanks
for
helping
us.
G
This
is
great
last
note.
I'll
say
this
was.
I
think
this
was
mesher's
maiden
voyage
onto
oracle
cloud
infrastructure,
so
assuming
that
you're,
ultimately,
you
know
successful
here
and
kind
of
measures
doing
its
thing.
We
might
want
to
update
the
compatibility
matrix
for
the
kubernetes
platforms
that
meshrey
is
compatible
with.
So
this
is
this.
Is
this.
B
Basically,
let
me
give
you
a
history
right,
so
we
support
like
kubernetes,
managed
clusters
right
so
there,
where
we
have
a
bare
metal
like
a
oci
and
we
bring
up
the
kubernetes
into
that
platform
right
oci
instance.
Okay,
so
so
it's
just
not
it's
like
a
bare
metal
where
you
run
a
kubernetes
right.
So
that's
it!
The
oca
is
just
an
instance
type.
G
B
All
three
options
like
you
can
have:
we
can
we
can
deploy
a
bare
metal
and
convert
it
to
a
kubernetes
cluster
or,
like
you,
can
use
a
eks
or
gcp
or
whatever
it
is
so
that
is
the
next.
But
what
I'm
using
is
oci
instance,
just
like
a
bare
vm
so
where
we
bring
up
the
like
a
managed
graphic
kubernetes
cluster
there
and
I'm
just
using
your
your
measuring
to
connect
and
from
the
operator
there
and
experiment
with
that.
It's
nothing
to
do
with
the
game.
Oca
is
a
manager
cluster.
G
Fair
enough
all
right,
gents
brevin,
that
was
great
catch.
You
catch
you
guys
next
time
catch
you
guys
on
the
next
call,
yeah
sure,
okay,
bye-bye.