►
From YouTube: Meshery Development Meeting (Oct 21st, 2020)
Description
Design considerations for mesheryctl.
A
B
B
A
B
You
know
a
later
latest
technology
step,
such
as
collection
of
data
from
a
mobile
device,
collecting
data
directly
from
the
medical
devices
and
transferring
data
to
the
back
end
right
to
the
cloud
and
doing
analytics
on
top
of
that
and
derive
some
insights
and
deliver
that
insights
to
downstream
enterprise
application
or
some
you
know
or
analytics
platform
or
existing
enterprise
applications.
So
eventually
the
data
will
be
submitted
back
to
fda
for
the
drug
approval.
A
B
Yeah,
it
is
so
what
I,
what
we
built
is
a
platform,
and
it
has
you
know
multiple
capabilities
like
hadoop
is
hadoop.
Is
there
and
tensorflow
spark?
So
in
this
case
the
customer
will
be
able
to
choose.
You
know
based
on
their
expertise.
You
know
in
the
technologies
that
they're
familiar
with
to
pick
the
platform
for
their
use
case.
Nice.
A
Nice,
okay,
yeah
you're,
not
holding
you're,
getting
you're,
getting
them,
you're
doing
some
etls
and
but
but
you're
you're,
getting
them
the
raw
stuff
if
they
want
it
or
in
the
form,
nice.
Nice.
All
right.
I
am
intrigued,
okay,
good!
Well!
You
you
got
suckered
in
man.
You
got
hooked.
Okay,
that's
nice!
Nice
to
have
you.
A
C
Oh
no
shoes
all
right,
I
like
to
start
off
meetings
with
a
basic
intro
to
the
meeting
for
anybody
who's
new,
so
the
mystery
def
call
is,
for
anything
related
around
or
based
on
the
mystery
projects
feel
free
to
add
any
topics
to
the
journalist,
even
before
the
meeting
or
during
the
meeting
you're,
not
that
strict.
Here
we
have
a
couple
of
issues
today
which
are
well
varied
and
yeah.
That's
it.
Let's
start.
D
C
D
Yeah,
so
basically,
I
just
needed
some
thoughts
on
this
on
this
pull
request.
So
one
of
the
major
aims
of
this
pull
request
was
to
add
an
adapter
flag
to
mesherie
so
that
it
is
able
to
choose
the
adapter
when,
when
it
spins
up
and
because
for
now
it
just
downloads
all
of
the
adapters
that
are
there
in
in
the
docker
compose
file.
E
D
Yeah
yeah
just
yeah,
just
a
few
comments
on
the
on
the
docker
compose
file.
So,
if
you
go
to
see,
we
have
all
of
the
adapters
in
them,
and
one
of
the
major
features
right
now
is
that
it
pulls
in
all
of
those
images
and
if
they
aren't
there,
it
updates
them.
So
what
we
plan
to
do
with
this
issue
is
probably
be
able
to
choose
an
adapter,
and
you
know
pick
it
up
right
up
from
the
docker
compose
file
and
just
load
that
adapter.
D
However,
I
was
since
I'm
new
to
the
community.
I
was
made
aware
that
this
needs
some,
of
course,
to
see
some
design
decisions
to
be
taken
because
it
it
needs
to
be
persistent
and
it
needs
to
be
in
accordance
with
you
know,
with
what
the
measuring
tool
in
general
provides
so
yeah.
That's
that's
the
overview
of
the
entire
request
that
it's
aiming
to
too
long.
D
Didn't
real
kind
of
a
description
is
just
we
want
to
choose
the
kind
of
adapter
that
we
are
planning
to
start
up
and
not
spin
up
and
pull
up
all
of
the
all
of
the
adapters.
So
I
would
just
like
community
feedback
on
this
yeah.
What
do
people
think
about
this?
Any
design
changes
any
sort
of
first
thoughts.
Right
of
the
top
of
your
mind,
can
go
ahead
and
you
know
share
them.
A
A
A
There
is
room
for
everyone
on
the
call
to
weigh
in
by
the
way
a
little
bit
about
what
we're
about
to
talk
about
is
well,
it's
it's
a
little
bit
about
user
experience
and
the
nice
thing
about
the
user
experience
or
discussions
around
user
experience
is
that
everybody's,
a
critic
everybody's
got
an
opinion
on
like
what
feels
good
and
what's
useful.
So
so
everyone's
opinion
is
valid
on
this
I'll
note
that
sladen,
if
you're,
able
to
navigate
back
out
of
the
thousand
tabs
that
you
have
open
back
to
the.
A
A
A
Why
the
why
the
syntax
is
structured
in
the
way
in
which
it
is
and-
and
so
anyway,
the
dock
is
pretty
decent.
It
needs
to
be
twice
to
three
times
as
long
because
there's
like
twice
to
three
times
as
many
features
that
are
in
mesherie
itself,
that
you
can
access
in
the
ui,
but
those
capabilities
have
yet
to
be
exposed
in
the
cli,
both
the
ui
and
the
cli.
They
both
access.
The
same
rest,
api
they're,
both
clients
of
the
rest
api.
A
So
the
only
thing
holding
us
back
from
making
messy
ctl
equally
as
powerful
at
or
more
or
than
the
ui
is,
is
just
some
thoughtful
design
as
to
the
syntax
structure
and
a
bit
of
the
interaction
so
I'll
before
we
even
start
I'll
say.
One
thing
that
would
be
really
helpful
is
to
take
lead
from
michael
and
for
and
to
begin
to
make
it
more.
We
were
trying
to
be
consistent
with
the
structure
of
how
we
look
at
each
of
the
commands.
A
That
that's
a
decent
one
like
this
here's
a
perf
command.
We
were
trying
to
be
intentional
with
like,
and
this
all
needs
to
be,
and
most
of
it
is
public-facing
documentation.
But
so
it
says:
hey
here's,
the
perf
command,
here's
the
name
flag
here,
it's
we
try
to
give
an
example
and
a
usage
and
and
if
you
go
up
a
little
bit,
there's
the
behav
like
yeah.
So
we
talk
about
behavior
usage,
etc.
A
We're
trying
to
be
consistent
with
how
we're
designing
these
and
one
of
the
things
that
we
didn't
really
that
until
who
had
pointed
it
out,
we
have
yet
to
really
include
a
scenario.
A
little
bit
of
like
a
user
story
would
be,
which
would
be
like
here's,
the
task
that
I'm
trying
to
accomplish,
and
so
if
we
apply
that
line
of
thought
to
the
pr
that
you
have
up,
which
is
a
good
pr
by
the
way.
A
A
If
these
other
four
didn't
start,
and
so
it's
a
good,
it's
a
use
case
that
you
know
that
we've
heard
multiple
times
and
and
should
be
there,
the
the
approach
that
has
been
taken
currently
in
the
in
the
pr
that
we
have
is
here's
the
use
case.
It's
or
or
the
way
in
which
the
approach
has
been
taken.
Is
I'm
a
user
I'd
like
to
spend
up
measuring
when
I
do
I
just
want
to
have
I
want
to.
I
want
to
remove
a
few
adapters,
or
maybe
the
more
common
use
case
would
be.
A
I
actually
just
want
to
use
measuring
for
one
adapter
or
one
type
of
surface
mesh
at
the
moment,
because
it's
easier
and
quicker
just
to
specify
here's
what
I
want
included
as
versus
like
here's,
the
seven
that
I
don't
want
included
and
by
the
way
next
week,
we'll
be
releasing
a
ninth
it'll,
be
the
nginx
service
mesh.
So
I
don't
I
don't
mean
we
behind
the
scenes.
I
mean
abhishek
who's
on
the
call,
the
all
of
you
guys
that
are
contributing
and
so
a
it
might
be.
A
We
should
give
some
thought.
It's
probably
more
likely
the
case
that
someone
is
interested
in
removing
the
waste
the
overhead
and
when
they
are
it's
probably
by
whittling
it
down
to
like
one
two,
maybe
three
adapters
and
the
other
six
or
what
have
you
can
be
set
aside
the
approach
here,
another
artifact
of
this.
Oh
man,
look
at
that
a
user
story
type,
that's
awesome!
A
The
other.
The
part
of
the
approach
here
is
that
this
is
transient
on
what
we're
doing
it's
a
one-time.
It
has
a
one-time
effect.
It's
mastery,
ctl
system
start,
you
know,
dash
dash
adapter.
You
know
here's
the
the
one
or
two
adapters
that
I
would
like
to
have
included
that
facilitates
this
use
case.
It
is
transient
in
so
much
as
that
only
has
an
effect
for
that
command,
which
is
a
valid
use
case.
A
It
might
be
that
we
would
find
more
often
than
not,
and-
and
we
should
take
a
poll
as
soon
as
I
say
this-
it
might
be
that
more
often
than
not
you
as
an
individual,
have
sort
of
permanently
crossed
out
the
need
to
interact
with
one
or
two
or
three
of
these
things
and
your
your
list
is
getting
smaller,
so
there
in
lie
the
question.
Maybe
our
first
question
to
everybody
which
is:
do
you
consider
it
to
be
the
case
that
what's
more
useful
to
you,
a
transient.
A
Start
the
message
system
start
with
only
a
specific
set
of
adapters
or
more
of
a
persistent
one,
where
you're
configuring
you're
configuring
like
a
yaml
file
or
there's
a
configuration
file
where
you're
that
has
identified
each
of
the
meshes,
and
you
are
persistent.
You
know
in
a
persistent
way,
you're
configuring.
E
A
A
E
Actually,
I
have
a
question:
there
exists
this
config
flag
or
option.
Doesn't
it
and
two
two
config
options?
One
is
called
config
and
the
other
one
config
file
or
something
remember
that's
right
to
admit.
So
how
would
that
be
relate
to
this
profile
id,
because
once
I
try
to
sort
of
do
exactly
the
same
just
having
the
console
adapter
and
then
I
thought
creating
a
new
yaml
file
in
the
measuring
folder.
A
E
Giving
that
a
startup
parameter
would
sort
of
lead
to
that.
I
only
get
the
the
measuring
adapter,
but
for
some
reason
I
couldn't
get
it
to
work.
That
was
a
couple
of
months
ago,
so
maybe
it's
working
now.
So
how
would
that
relate
to
this
idea
of
what
we're
talking
about
here,
which
I
think
is
really
good
and
useful
and
a
very
good
use
case
to
have
to
be
able
to
say
yeah.
I
just
want
to
work
with
the
concept.
A
Right,
it
directly
applies,
I'm
glad
that
you
brought
it
up.
That's
actually
why
I
wanted
to
have
this
conversation
was
really
sort
of
around
the
fact
that
I
don't
know
if
it's
hinted
in
what
the
questions
that
I'm
asking
I'm
not
trying
for
them
to
be
leading,
but
but
I'm
more
or
less
hinting
that
it's
my
belief
that
you'll
more
often
than
not
find
individuals
that
say
yeah.
You
know
these
other
five,
that's
great,
maybe
I'll
get
to
them.
But
for
now
you
know
I'd
like
to
look
at
these.
A
There's
two
yaml
files
today
that
deal
with
configuration
of
measuring
and
of
measuring,
ctl
and
and
the
notion
that
that
that
there's
two
individual,
that
there's
two
individual
files
for
controlling
mescheri's
configuration
is
perhaps
appropriate,
but
I
think
confused
at
the
moment
and
so
perhaps
best
to
describe
what
those
two
yaml
file
their
purpose
and
what
they
do,
such
that
something
like
a
profile
that
type
of
a
construct,
I
think,
is
quite
powerful,
pretty
interesting.
It
becomes
familiar
to
others.
A
We're
recording
this
so
hopefully
this
will
I'm
so
sensitive
now
that
I've
repeated
myself
some
like
10
times
on
so
many
subjects
over
the
last
and
I
it
doesn't
get
written
down
or
I
don't
go,
write
it
down,
and
it's
not
all
about
my
thoughts,
but
just
rather
somebody
needs
to
put
something
on
paper.
So
people
like
me
who
can
come
over
and
say
well,
why
didn't
you?
Why
isn't
this
a
profile
and
then.
A
A
If
for
all
of
you
who
have
installed
mescheri
ctl
locally,
you
should
find
that
in
your
users,
folder
there's
a
hidden
folder
called
dot
meshery,
it's
a
it's
an
application!
It's
a
hidden
application,
folder
that
stores
that
pers
persists;
information
about
mesher's
installation
on
your
local
machine
and
slayden.
Actually,
if
you
would
mind
doing
me
a
favor
just
if
you
did
an
ls
on
in
your
user's
home
folder
on
dot
messaging.
A
Is
one
of
the
two
files
meshri.yaml
is
nothing
more
than
a
docker
compose
file?
It's
literally
verbatim
a
docker
compose
version
three
compatible
file,
which
is
fine
for
starting
up
mesherie
locally.
It's
actually
to
to
this
use
case.
It's
actually
fairly
easy
to
manipulate.
You
can
go
in
and
just
comment
out
adapters
that
you
don't
want,
save
that
config
and
just
and
just
you
and
you're
good
to
go
like
you've
achieved
this
use
case.
Actually,
you've
achieved
a
persistent
use
case
like
that
you
can
make
a
duplicate
of
meshri.yaml
and
make
changes
to
it.
A
You
can
make
it.
You
can
make
a
change
there
and
when
you
start
measuring
you
can
there's
a
global
flag
for
what
measuring.yaml
you
want
to
use.
What
compose
file
you
want
to
use
that
makes
some
intuitive
sense.
I
think,
like
that,
that
meshri.yama
makes
some
sense
here.
It
is
right
here,
but
it's
to
the
extent
that
it's
literally
a
docker
compose
file.
A
It's
constrained
we're
constrained
measuring,
is
constrained
to
when,
when
you
do
mastery
ctl
system
start,
there
is
a
hard-coded
assumption
that
that
means
docker
compose
and
that's
not
good,
like
measure
really,
when
you
boil
it
down.
Mesh
will
run
on
a
bunch
of
platforms
on
windows
and
linux
and
a
bunch
of
different
kubernetes,
but
it
essentially
always
deploys
inside
of
a
container,
and
so
it
essentially
either
deploys
inside
of
docker
as
a
compose
app.
A
You
don't
have
to
have
it
as
a
compose
app,
but
it's
more
convenient
when
you
have
it
like
that
or
it
deploys
inside
of
one
of
like
10
different
kubernetes
environments.
It
either
deploys
on
docker
or
in
kubernetes.
It's
basically
either
deploying
inside
of
a
kubernetes
cluster
in
a
docker
container
or
it's
deploying
outside
of
a
kubernetes
environment
cluster.
A
But
we
have
hard-coded
the
fact
that
we
assume
that
when
you
do
a
mastery,
ctl
system
start
that
you're
choosing
docker.
So
I
mentioned
this
because
mastery.yaml
in
this
case,
we
need
to
have
a
degree
of
separation.
So
this
concept
of
inserting
a
profile
ooh,
that's
kind
of
nice,
like
we
can
start
to
bring
in
the
notion
of
in
cluster
out
of
cluster
or
like
in
docker
or
in
kubernetes.
A
Well,
it
was
meant.
I
think
that
we've
done
ourselves
wrong,
maybe
here
and
we
need
a
bit
of
a
redesign
that
and
profiles
probably
are
part
of
that
that
the
concept
here
was
that
all
right,
not
only
are
we
assuming
that
you
might
start
measuring
in
docker
we're
kind
of
also
assuming
that
you're
a
single
cluster
environment
that
you're
a
single
kubernetes
cluster
environment,
because
because
we
don't
really
have
the
concept
of
context
or
environment,
and
so
the
thought
was
well.
A
The
poor
man's
version
of
a
context
is
the
ability
to
set
that
inside
of
meshrictl
hyphen
config
like
have
a
yaml
file
there.
That
has
some
some
choices
of
you
know
some
a
way
of
configuring.
The
client
and
part
of
that
might
be
the
name
of
the
context
that
you
want
to
connect
to
the
name
of
the
kubernetes
cluster
and
so
sladen
you
kind
of
stepped
into.
A
A
D
So
so
maybe
maybe
we
could
yeah,
maybe
we
could
you
know
I
could
keep
it
in
a
draft
kind
of
a
state
where
we
keep
iterating
on.
You
know
on
whether
we
need
the
profile
or
you
know,
on
whether
we
can
have
it
as
a
persistent.
You
know,
mechanism
sort
of
and
we
can
keep
iterating
and
testing
until
we
find
something.
That's
you
know,
that's
really
concrete
and
we
want
to
go
ahead
with.
It
sounds
good
to
me.
A
Oh
yeah,
yeah,
slayton
and
actually
just
you're
beautiful,
that's
a
great
response
actually
just
before
we
close
tie
off
that
conversation,
yeah,
hey,
leaving
open
long,
live
drafts
or
long-lived
branches.
What
have
you
sounds
good?
There's
a
new
there's
this.
Oh
man,
boy!
This
is
a
bridge
to
another
topic,
but
I
don't
want
to
move
off
of
this
just
yet
and
that's
shady,
do
you
mind
pasting
in
the
link
to
the
design
spec
in
the
zoom
chat?
A
I'm
super
animated
about
the
fixing
this
because
I
think
actually
fixing
it
enables
other
things
that
people
ask
for
pretty
frequently
they
ask
for
they
say:
hey,
wouldn't
it
be
great
if
you
could
just
like
use
mastery
ctl
to
deploy
mesherie
into
a
kubernetes
cluster
like
yeah.
That's
actually
really
easy
to
do
you
just
go,
but
but
we
don't
have
the
right
structure.
We've
got
two
config
file:
two
global
flags
that
point
to
config
files
that.
A
So
I
guess
here's
my
or,
for
my
part,
here's
my
ask
and
others
may
have
a
different
approach
to
this,
but
it's
to
ask
for
like
if
anyone
would
like
to
wrangle
the
if
anyone
would
like
to
take
a
crack
at
the
design
of
what
they
think
that
this
looks
like
in
the
design
spec
and
you
guaranteed
you'll,
get
it
wrong
or
like
that's.
No,
I
mean
don't
don't
be
shy
about
that,
like
somebody
needs
to
write
something
down.
A
A
D
A
A
There
isn't
a
concern
with
the
use
case
for
allowing
someone
to
transiently
include
or
exclude
adapters.
It's
a
valid
use
case
that
the
issue
is.
I
think
that,
as
we
move
forward
near
term
that
syntax
structure
will
change,
knowing
that
we
should
put,
we
should
think
on
a
few
things.
That
one
is
that
there
there
are
two
global
flags.
A
A
Context
is
more
about
environment
and
like
which
cluster
you're
going
to
be
pointing
at
or
which
like
or
it's
simply,
the
fact
that
measuring
ctl
today
assumes
that
when
you
run
a
mastery
ctl
command
and
it
needs
to
talk
to
mesheri
the
server
that
meshrie,
the
server
is
located
at
localhost
like
well.
No,
because
you
might
have
a
client
installed
locally
and
mastery
server
could
be
in
kubernetes,
so
I'm
not
being
concise.
Frankly,
it's
like
a
lot
of
considerations.
D
Server
yeah
yeah,
I
definitely
understand
now.
I
could
get
a
much
much
more
clear
understanding
of
of
where
the
profile
and
where
the
context
is
exactly
coming
from
yeah
yeah.
With
respect
to
context,
you
made
a
great
point
that,
having
you
know
having
the
the
kind
of
kubernetes
or
the
docker
kind
of
construct
that
we
would
require,
gives
us
the
kind
of
flexibility
from
the
shell
to
be
able
to
make
that
happen,
but
yeah.
D
A
Yeah,
no,
that's
exactly
the
point.
Yeah,
I'm
excited
for
you,
honestly,
like
and
and
here's
the
thing
is,
there's
no.
I
haven't.
I
have
shame
in
many
areas,
although
I
try
to
avoid
it,
and
this
isn't
one
of
those
areas
I
would
say.
Not
only
is
this
not
a
shameful
thing
to
do.
This
is
the
most
appropriate
thing
to
do.
That
is
to
go
to
cubectl.
A
A
I
should
know
this
better
as
a
docker
captain,
but
they
had
recently
released
support
for
contexts,
and
that
was
more
in
like
context
of
cloud
deployments
like
which
cloud
do
you
want
to
deploy
your
containers
to
it's
not
to
go,
lift
I
other
people's
ideas
and
things
I
mean
sure,
that's
an
aspect
of
like
to
just
help
facilitate
the
design.
The
number
one
reason
why
to
go.
Do
that
is
like
joshua
was
saying:
hey,
I'm.
You
know
I
learned
about
meshri
like
a
week
ago.
A
I'm
definitely
new
to
meshary
as
I
go
to
touch
the
cli.
Wouldn't
it
be
great
if
all
of
the
time
that
joshua
is
spent
with
docker
cli
q,
ctl
cli
istio
ctf,
like
that,
he
just
it
just
mesri
ctl
intuitively
flows
for
him,
because
the
context
command
works
like
the
other
ones
and
the
profile
command
works.
A
A
A
Yeah,
you
know,
I
don't
think
he's
on.
I
I
toss
this
in
there
because
it's
a
good
point
of
discussion-
and
I
don't
mean
to
like-
maybe
we
should
come
back
to
it
because
otherwise
I'll
be
talking
the
whole
time
yeah.
This
is
highly
related.
I
would
say
this
is
probably
on
hold
pending
sladen's
designs.
Back
probably.
D
D
Actually,
because,
with
the
way
it
kind
of
it
lets
you,
you
know
generate
a
default
configuration
if
it
doesn't
exist
which,
which
was
which
is
kind
of
required
yeah
this
there
are
a
couple
of
after
we
get
the
design
spec
a
bit
more
elaborated.
Maybe
this
issue:
will
this
pull
request
will
make
much
more
sense,
so
yeah
look
out
for
this
one
nice.
A
And
required
all
right,
the
other
one
here
was
from
well.
If
I,
I
think
husana,
I
think
she
goes
by
hex
dump
and
she
goes
by
the
the
piscean.
I
think
she
goes
by
anything
but
her
name.
So
it's
hard
to
remember.
Oh
yeah,
who's,
who's
hannah,
quite
a
nice
lady
she's
recently
put
in.
I
think
this
might
be
her
first
pr
or
a
couple,
but
but
she's
working
on
mystery,
ctl
as
well,
and
I
thought
there's
a
collection
of
these
things.
We
should
kind
of
chat
through
them
slayden.
A
D
Yeah
yeah,
some
of
them,
some
of
them
actually
centered
around
having
multiple
permissions
regarding
you
know,
spinning
up
the
docker
demon.
If
it's
you
know,
if
it's
not
running
or
if
it's
not
existing,
so
that
that
was
my
major,
you
know
not
concern.
It
was
more
of
a
major
lookout,
kind
of
a
thing.
Look
out.
You
know
gotcha
kind
of
a
thing.
You
know
that
that
maybe
that
could
catch
us
off
guard
if
you're
running
on
different
operating
systems.
D
So
you
know
I
could
maybe
we
could
put
those
into
issues
somewhere
if
we
ever
run
into
them,
because
we're
dealing
with
multiple
operating
systems
on
this
one.
So
it's
a
good
idea
to
have
a
look
at
have
a
look
at
how
it's
how
it's
structured
yeah
with
the
docker
team,
because
we're
attempting
to
start
things
on
our
end
right
and
even
when
attempting
to
install
it.
If
I'm
not
wrong
so
right,
yeah.
Those
are
things
to
look
out
for.
A
A
A
Very
good,
okay,
so
so
literally,
it
would
have
taken
joshua
two
commands
to
install
mastery
ctl.
Actually,
if,
if
someone
would
anyone
on
the
call,
if
you
would
volunteer
to
create
an
issue
to
update
these
instructions,
it
look
it's
three
instructions,
but
it
can
be
reduced
to
two.
The
trick
here
is:
we
can
eliminate
this
line
and
instead,
what
needs
to
be
augmented
is
it
should
be
brew,
install
layer,
five
io,
slash,
mastery
ctl
and
those
two
lines
can
be
collapsed
into
one.
A
Oh
thank
you.
So
when
joshua
installed
with
brew,
he
ran
you
know
three
commands.
He
could
have
just
run
two
that
that's
not
the,
and
if,
for
anyone
who
uses
bash,
they
literally
run
one
command
that
will
do.
It
will
grab
the
mesh
free,
ctl,
binary,
that's
specific
to
their
to
either
mac
or
linux,
move
it
shuffle
it
around,
assign
the
right
permissions
to
execute
on
their
system.
It
will
then
also
take
liberty
of
invoking
measuring
ctl
system
start,
which
will
then
spin
up
docker
containers.
A
You
know,
could
the
compose
it'll
start
meshri
spin
up
the
containers
if
docker
is
present
and
available
and
it'll
also
when
the
containers
are
up
it'll,
auto
open
the
default,
the
user's
default
browser,
if
they're
in
a
if
they're,
not
in
a
headless
environment-
and
I
I
think-
or
I
think
most
of
us
here
thought
like
that's
really
that's
a
really
small
thing,
but
the
ux.
There
is
kind
of
neat
like
hey
you
just
you
did
one
thing
and
it
downloaded
self-installed
and
self-started
and
then
dropped
you
off
it.
A
You
know
you're
good
to
go
you're
good
to
start
managing
your
mesh.
Like
that's
kind
of
step.
Number
two
is
like
there.
There
ain't,
no
step
number
two
and
so,
but
but
if
docker
isn't
running,
then
you
you,
it
only
gets
so
far
and
hasana
was,
you
know,
saying:
well,
maybe
we
should
just
start
and
and
and
to
slayton's
point
like
what
do
you
guys?
What
do
you
all
think
is
starting
docker
on
someone's
machine?
A
D
D
D
Starting
and
stopping
and
starting
sick
if
they
are
testing
something
and
they
are
doing
some
doing
a
couple
of
iterations
with
mystery
ctl,
then
every
time,
if
the
prompt
comes
up,
it's
it's
a
bit,
annoying
yeah
yeah.
Just
just
to
add
to
that.
Maybe
it
could
could
it
could
ask
like
hey
docker
required.
Do
you
want
me
to
start
it,
and
then
you
just
hit
an
end?
Yes
or
no,
where
you
know
yeah
something
like
that.
D
D
A
Issue
is
that
I
think
she's
actually
addressing
no
no
she's
addressing
this
inside
of
mystery
ctl
itself.
So,
okay,
but
again
the
the
assumption
is
that
you're
deploying
to
docker
not
necessarily
deploying
to
kubernetes,
so
I'm
hoping
that
soon
we'll
be
have
a
design
for
and
we'll
be
able
to
use,
measuring
ctl
to
provision
measuring
inside
of
kubernetes
and
that
the
particular
enhancement
that
hasana
is
making
doesn't
apply
like
like
we'll
just.
A
A
A
D
This
is
really
annoying.
The
zoom
thing
is
blocking
me
to
change
stuff.
Okay,
so
basically
the
error
before
was:
we
are
not
able
to
purchase
the
link
to
the
local
office
of
grafana
while
we
reload
it
right,
and
we
couldn't
do
that
because
it
was
fetching
it
from
the
database
itself
and
it
takes
some
time
for
it
to
load
and
yeah.
I
pasted
it
in
a,
but
the
other
problem
over
here
is
in
the
database.
D
D
A
A
Yeah
no
thank
you
yeah
totally,
because
because
there's
another
istio
workshop
starting
in
an
hour-ish,
it's
not
advanced.
It's
the
intro
but
they're,
pretty
related.
We
use
nashville
to
teach
people
and
and
we're
going
to
show
some
grafana,
the
grafana
integration
and
so
yeah.
I'm
hoping
that
we
can
include
this.
This
is
a
relief
to
me
drew
it's.
Also,
it's
not
a
complete
relief
because
of
the
what
you
were
mentioning
about
this
being
somewhat
time-stamped
or
dated
of
a
solution
with
respect
to
dependency
on
react,
17.
D
Yeah
once
we
moved
to
the
act
17,
this
won't
work
and
yeah.
We
should
probably
change
the
structure
with
like
if
you,
if
I
want
to
use
the
redux
store
itself,
then
we
should
use
that
instead
of
having
both
prop
and
state,
save
the
same
things
like
I.
I
need
to
look
into
why
we
did
that
in
the
first
place,
to
understand
that
whether
changing
it
would
be
a
good
idea
or
not
so
yeah,
I
will
look
into
that
before.
Changing
anything,
don't
see
any
reason
yet
for
there
to
be
two
different
data.
A
A
In
the
cont
in
the
measuring
server
container,
the
instance
that's
running
mastery
server,
there's
a
little
little
bit
of
info
that
gets
stored
on
the
user's
local
file
system
that
we
were
just
looking
at
earlier
under
dot
mesh
re.
There's
two
small.
I
forget
the
name
of
the
the
golang
database
implementation
that
we're
using,
but
just
to
persist
some
data
to
the
file
system.
A
A
Bitcast
is
the
small
golang
implementation
of
datastore
that
we
use
to
persist
some
of
that
data
to
disk
to
file
system,
but
and
also,
I
think
it
provides
an
in-memory
cache
if
I
recall,
but
so
anyway,
I'm
mentioning
that
stuff.
I
don't
know,
I
think
it's
it's
related
to
the
redux,
and
you
know
things
that
we're
storing
inside
of
the
browser
but
yeah.
D
No,
the
thing
is
when
you
reload
all
the
data
in
redux
and
the
local
state
would
be
clear
and
the
initialized
data
would
go
there
and
we
are
fetching
the
data
from
the
server
itself.
The
service
cache
it
right
and
when
we
are
doing
it,
it
is
getting
saved
in
the
redux
state,
and
then
we
are
getting
that
same
data
and
using
it
in
the
local
set
itself
instead
of
directly
using
it,
which
is
causing
this
havoc,
where
the
local
state
is
only
updated.
D
A
A
Is
there
any
amount
of
concern
that
when
we
refresh
and
that
cache
is
wiped
that
yeah,
I
think,
as
you
guys
work
on
mesh
map
and
the
visual
topology,
there
might
be
a
different
concern
there
about
what
gets
cached
because
it
and
what
doesn't
how
much
of
it
gets
refreshed,
because
that
should
be
a
lot
of
data.
Or
I
expect
it
will
be
a
lot
of
data.
A
A
E
A
Yeah
or
at
the
the
community
meeting,
like
you
were
saying
like
actually
the
there's
two
items
on
the
agenda,
one
is
the
hashicorp
console
product
manager
will
come
present,
the
latest
and
greatest,
but
those
two
items
on
friday.
That
sounds
great.
I
mean
it's
we're
overdue.
On
the
common
library
discussion.