►
From YouTube: Meshery Development Meeting (Nov 17th, 2021)
Description
Meshery Development Meeting - November 17th, 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
B
I
just
joined,
can
you
give
me
five
minutes?
I
just
I'm
setting
up
my
system
and
also
okay,
sure.
C
Okay,
so
basically,
I
think
I
am
done
with
versioning
of
the
docs,
so
I
had
a
conversation
with
lee
yesterday
and
I've
done
it
till
only
v4
for
v
0.4.
C
I
was
just
working
on
some
recent
changes
that
were
suggested.
I
have
changed
it
in
the
latest
version.
I've
done
the
latest
over
here,
but
I
need
to
change
for
the
other
versions.
C
I
like
a
lot
needs
to
be
reviewed
because
I
have
checked
most
of
the
links
and
they
none
of
them
seem
to
be
broken,
except
for
this
one.
D
Now,
as
a
well
as
a
point
of
curiosity
like
as
good
going
back
into
the
annals
of
the
internet
and
building
out
like
the
dot,
one
version,
how
how?
How?
What
was
that,
like.
D
D
C
D
The
reality
is
that
overwhelmingly
people
will
be
wanting
the
latest
docs,
and
so,
while
it's
a
significant
capability,
it's
probably
not
not
necessarily
going
to
be
very
highly
used,
or
at
least
not
for
a
long
time.
Until
there
are
many
many
deployments
and
people
need
to
go
back
in
history.
D
Oh,
a
capital
home
sure
sure,
as
a
matter
of
fact
like
in
something
like
a
cookie,
so
deepak
was
just
pointing
out
the
cookie
crumb
trail
of
that's
on
the
docks
and
the
capitalization
of
the
words
yeah
in
general.
It
probably
works
really
well,
if
everything's
the
same
either
lowercase
everything
or
everything
is
uppercased,
and
that
way,
there's
just
uniformity
having
having
proper
capitalization
is
probably
something
of
a
distraction,
so
so
it
could
go
either
way.
It
looks
like
we're
using
lower
case
than
this.
In
this
instance,.
D
Nice
anybody
else
have
feedback
for
what
was
the
when
you
look
back
and
reflect
on
it.
So
we've
talked
about
versioning
of
docs
like
a
lot
of
times,
a
lot
like
I've.
I've
walked
a
lot
of
people
through
different
approaches
and
considerations
and
on
the
tail
end
of
it.
As
you
go
to
merge
this
today
and
you
look
back
in
retrospect,
is:
should
we
have
just
moved
to
docusaurus?
D
Should
we
have
gotten
a
new
theme
like?
Was
this
an
insert,
like
you
know,
a
very
difficult
to
surmount
effort?
Is
this
something
it's
that's
going
to
have
high
sustaining
costs
going
forward?
Are
there
manual
tasks
associated?
You
know?
Are
these
things
that
we
can
automate
or
what?
What?
D
C
Yeah
we
yeah
there
are.
There
is
some.
You
know
there
are
some
manual
tasks
that
need
to
be
done
like
if
we
go
to
a
different
version
yeah.
If
we
go
to
a
different
version.
All
these
the
links
over
here-
okay,
not
not
in
the
site
explicitly,
but
in
like
the
index
they've
had
to
be
like,
like
we
were
doing
it
last
night
like
they
would
have
to
be
changed
to
v
0.5
slash,
then,
whatever
the
link
is
after
that
and
that
would
prob
that
would
be
like
that.
C
C
C
Like
even
for
like
the
assets,
since
we
are
using
like
assets
from
a
specific
version,
they
would
have
to
be
changed
to
v0.5
slash
assets,
slash
whatever
the
name
of
the
file
is.
C
D
Oh
that
hurts
me
to
hear
you
say
that
I
don't
know
how
many
hours
I've
spent
yeah
great
okay.
So
since
there
isn't,
let
me
fill
in
some
blanks
here
for
others.
This
is
a
great
learning
thing
for
all
of
us.
I
wish
that
mesh
mate
aditya,
the
other
mesh
made
aditya
krishna
was
here.
This
is
one
of
the
things
that
he
had
one
of
the
first
things
he
had
wanted
to
do
when
he
joined
the
community
was
to
like
switch
over
from
jekyll
to
docusaurus
and
there's
no
question.
D
Docusaurus
is
a
better
there's,
a
more
feature-rich
framework
or
you
know,
designed
for
documentation
explicitly
while
jekyll
is
maybe
you
know
maybe
easier
to
pick
up.
Maybe
it
all
kind
of
comes
down
to
markdown
when
you
go
between
like
hugo
or
jekyll
or
what
have
you?
D
I
think
someone
on
the
call
was
recently
looking
at
developer
blog
tools,
so
I
won't
go
too
much
further
into
it
other
than
to
say
that,
so
we
got
started
on
jekyll
because
it
is
it's
been
around
a
long
time.
It
is
easy,
easy
to
pick
up.
There's
a
bunch
of
themes
out
there.
The
theme
that
we
chose
was
is
a
documentation
theme.
As
a
matter
of
fact,
this
isn't
the
first
theme
that
we
chose
for
the
project.
There
was
a
second
there's.
D
D
As
a
matter
of
fact,
I
mean
I
guess
to
be
forthright
like
I
was
trying
to
create
the
project
while
leading
technology
strategy
for
a
billion
dollar
company,
a
software
company,
solarwinds
and
and
so
didn't
didn't
want
it
to
conflict
with
my
day
job
so
kept
it
separate
because
because,
even
though
there
was
no
conflict,
the
the
two
words
open
and
source,
when
you
were
to
bring
them
together
in
front
of
a
solarwinds
lawyer,
they
would
have
a
physical
allergic
reacted
like
their
brain
would
explode
in
front
of
you.
D
So
consequently,
the
docs
look
a
lot
different.
We
did
choose
jekyll
so
about
in
the
dot
2
release.
We
moved
from
a
vmware,
I
mean
well
they're,
not
a
vmware
employee
moved
us
from
he's
not
at
vmware
anymore.
Is
that
stripe
moved
us
from
one
theme
to
the
next?
D
If
you
look
through
the
you
might
you
might
guess
who
he
is?
If
you
look
through
books
of
the
history,
not
important
as
we
move
forward,
we
recognize
not
having
versions
is
kind
of
painful.
We
should
probably
version
all
the
things
it's
kind
of
like
a
design
principle
discussion.
I
was
having
with
some
folks
about
databases.
Yesterday,
it's
like
we
should
id
all
the
things
every
record
should
have
its
own
id
every
table
should
have
an
id
column,
they're,
cheap,
they're,
free,
generate
them.
D
Liberally
the
same
thing
for
versions:
it's
like
version
all
the
things
same
thing
for
like
patterns
if
we
haven't
versioned
our
patterns,
yet
the
only
reason
we
wouldn't
have
versioned
surface
mesh
patterns
is
because
we
didn't
want
to
include
another
line
instead
of
our
yaml
and
ugly.
It
up
we'll
probably
have
to
give
way
and
at
some
point
kind
of
version
them.
D
So
back
to
this
discussion,
versioning
of
our
jekyll
docs,
you
can
do
it
in
a
bunch
of
different
ways.
The
way
that
aditya
has
chosen
like
he'd,
said
once
it's
really
it's
just
like
it's
really
hard
to
see
it
before
you
do
it.
I
said
this
when
alonso
was
trying
to
do
it
and
there's
another
guy
before
him,
another
guy
before
him
that,
like
it's
really
extraordinarily
simple,
it's
like
two
steps
when
you're
done
other
than
creating
the
drop
down,
leaving
the
drop
down
aside,
it's
like
two
steps,
those
two
steps
or
those
three
steps.
D
Two
steps
can
be
automated.
We
don't
and
they
can
be
included
in
the
workflow
and
even
what
you
know
you
were
talking
about,
inserting
the
the
v.0,
the
version
number
that
can
be
part
of
that
automation
that
were
to
do
that.
It's
not
a
high
need
to
go
automate
because
it's
because
it's
not
a
very
frequent
task,
it's
a
only
when
there
is
a
release
made
and
only
when
that
release
is
of
a
minor
significance
to
use
semantic,
versioning
semantics
so
of
a
dot
significance,
and
so
which
means
it's
like.
D
You
know,
once
every
three
three
months
or
so
three
four
months,
actually
at
the
pace
that
we
kind
of
know
at
the
pace
that
we
go,
it's
like
a
five
month
thing.
The
releases
have
just
gotten
bigger,
though,
if
you
look
at
what's
in
dot
five,
it's
like
it's
kind
of
energy
draining
just
to
think
about
writing
up
those
blog
posts
about
all
the
features,
but
that's
a
really
good
thing.
D
When
we
talk
about
dot
six,
like
we
did
last
week
on
the
community
call
last
week,
there's
a
lot
4.6,
that's
good
anyway.
Suffice
to
say
it
is
quite
simple
how
you
can
create
static,
copies
snapshots
and
time
of
the
of
the
documentation
which
is
nice.
It's
a
one-time
task.
You
never
have
to
build
it
again,
and
so,
as
a
dj
goes
to
merge
this
change.
D
D
D
Okay,
well,
what?
What
really
sorry?
I
can't
get
it
confused!
Okay!
So
right
now,
if
you
click
on
the
drop
down,
okay,
cool,
so
we're
like
slightly
off
because
right
now,
we're
still
like
the
latest
is
dot
five
and
so
and
so
like
in
concept
there.
This
would
be
a
conflict
like
we
wouldn't
have
these
two,
but
anyway,
that's
that's
fine,
we'll
we'll
figure
it
out
as
we
go
to
release.
D
That's
cool
yeah
people
should
check
it
out.
It's
it's
yeah
surprising
that
there
aren't
other
jekyll
deployments
that
have
wisened
to
what
eddie
has
done,
because
it's
bulletproof,
whereas
the
other
approaches
that
you
can
take
are
not
you'll
find
yourself
with
broken
legs.
A
B
Okay,
so
assuming
that
you
can
see
my
screen
and
I'm
also
audible,
I'm
audible,
okay,
so
this
is
about
seeding
content
in
meshi
for
local
providers.
So
currently
we
see
the
content
that
content
means
applications,
patterns
and
filters.
So
when
you
log
in
as
the
none
provider,
you
have
applications
filters
and
patterns
out
of
the
box,
so
the
issue
we
were
facing,
which
was
not
a
major
issue,
but
the
issue
was
that
we
did
not
have
similar
behavior
for
the
build
time
as
well
as
the
runtime.
B
So
these
content
that
you're,
seeing
as
of
now
these
this
content
is
seated
at
build
time,
so
you
can
see
inside
of
the
docker
file
we
actually.
So
this
content
is
present
inside
of
the
image
itself.
So
for
people
who
are
building
this
thing
in
the
dev
environment,
let's
say
they
they've,
they
clone
them
measuring
and
they
start
to
work
in
a
dev
environment
and
they're
thrown
with
a
bunch
of
errors.
Like
cannot
see
it.
Content
cannot
see
content
and
it
kind
of
acts
like
a
noise
for
them.
B
So
a
solution
for
that
was
to
have
similar
behavior,
both
in
the
dev
environment.
As
well
as
the
build
environment,
so
I
hope
this
is
the
last
time
I'm
going
over
the
cd
content
for
local
providers
I'll
go
over
all
the
use
cases
and
the
current
implementation
that
that
is
currently
in
this
pr
I'll
paste.
B
The
link
for
this
pr
after
I'm
done
sharing
my
screen,
and
I
have
reflected
all
of
the
changes
that
this
pr
makes
inside
this
design
spec
under
somewhere
around
here,
yeah
this
kind
of
briefs,
everything
that
I'm
going
to
say
so
just
to
show
it
a
little
show
short.
She
wanted
to
change.
D
D
This
is
one
of
those
things
that
people,
this
isn't
necessarily
direct
functionality
that
everyone
will
think
about
documenting,
like
oh
mastery
does
application
management,
and
we
should
document
that
when
someone
has
an
issue
when
someone
says
wow,
there's
a
bunch
of
sample
content
in
here
or
how
does
it
get
in
here?
How
do
I
add
more
or
why
is
it
duplicating
or
like
the
behavior
of
the
system
is
what
ashish
is
about
to
walk
through
and
that's
the
kind
of
juicy
stuff
that
needs
to
go
into
documentation.
D
That's
actually
helpful
because
the
documentation
that
says
click
on
the
the
application
looking
configure
icon
to
edit
the
application,
like
that's,
not
helpful
documentation.
What
is
helpful
is
here's,
why
you
have
sample
content
and
here's
where
it
comes
from
and
here's
how
that
works.
So
for
any
of
you
that
might
take
that
on
now.
Is
your
chance
and
if
you
don't,
then
I
will
on
friday
so
I'll
give
you
two
friday.
B
Yeah,
please
yeah,
okay,
so
the
thing
that
I
added
was
to
avoid
those
errors
and
the
way
we
do
it
is
if,
if
we
are
in
the
dev
environment
automatically
the
the
appropriate
folders
for
the
the
seed
content,
that
is
for
patents,
filters
and
applications
will
get
created.
B
And
if
the
skip
download
for
environmental
variable
is
not
said
which
would
be
default
behavior,
we
would
actually
download
the
content
each
and
every
time
before
that,
I
had
a
logic
that
to
check
whether
the
content
is
new,
but
that
that
kind
of
doesn't
make
sense
because
we're
going
to
hit
the
make
the
http
request
anyway.
So
why
not
just
update
the
content
to
the
latest
content?
B
So
currently,
I
have
seen
content
in
here
and
I
can't
see
it
because
there's
that
low
there's
that
that
zooms
thing
I
hope
you
can
see
it
so
currently
like
this-
is
the
seat
content
we
have,
and
if
I,
if
I,
if
I
go
ahead
and
delete
this,
which
would
be
the
default
which
would
be
the
basically
this
would.
This
would
come
out
of
the
box
for
most
people.
If
you're
working
in
the
dev
environment,
you
would
not
have
the
seat
content
directory
inside.
B
So,
as
the
messi
server
starts,
it
creates
the
appropriate
directories
for
this
and
that
extra
and
then
it
goes
and
makes
those
api
calls.
It
extracts
all
those
applications,
filters
and
patterns,
and
now
we
again
have
that
seat
content
in
here
with
appropriate
directories
and
yeah.
We
have
all
of
these,
so
this
is
the
one.
This
is
one
use
case.
B
So
if
we,
for
example
now,
if
I,
if
I
go
inside
of
like
patterns
and
I
delete
one
of
the
patterns-
and
I
make
I
basically
restart
my
server
let's
say-
then
we're
going
to
come
up
that
single
pattern
that
was
deleted.
It
will
will
have
that
pattern
again.
So
we
will
come
back
to
that
state
where
we
have
all
of
the
applications,
filters
and
patterns
that
we
are
that
we
are
supposed
to
have
so
yeah.
It
came
back
again
now.
B
Let's
say
that
now
coming
with
a
different
recipe,
so
so
this
environmental
variable
is
set
in
this
other
recipe
that
run
fast,
no
download
seeds.
So,
for
example,
if
someone
is
working
in
the
dev
environment
and
they're
going
to
restart
messy
server
a
number
of
times
10
times
20
times,
and
they
don't
want
to
go
ahead
and
hit
the
api
rate
limit
or
something
because
you
don't
want
to
do
that,
you
can
use
this
recipe.
This
will
completely
skip
that.
B
You
will
still
create
those
directories,
so
you
will
not
get
any
errors,
so
these
directories,
appropriate
directories-
directories
will
be
created,
except
the
only
change
would
be
that
you
would
not
download
the
content.
So
if
you
have
something
already
in
there,
that
would
be
that
would
go
there,
and
that
would
be
seated.
If
you
have
nothing
in
there,
nothing
would
be
seated.
So
the
only
part
that
would
be
skipped
is
download.
So
if
I
restart
the
server
again
and
this
time,
I
call
the
no
download
seed
wait.
B
B
Okay,
if
I
go
back
here
so
as
you
can
see,
we
didn't
get
the
patterns
back,
although
the
folders
would
be
created
and
we
will
get
no
errors.
As
you
can
see,
there
are
no
errors
right
now
here.
The
only
thing
that
that
happened
was
no
patterns
actually
got
actually
registered
because
there
was
nothing
in
there.
So
there
would
be
no
error
in
any
of
the
cases,
and
I
walked
through
all
of
the
possible
cases
that
I
could
think
of.
B
If,
if
anyone
has
any
suggestions,
then
they
can
put
down
the
suggestions-
and
I
will
I
would
reflect-
I
have
already
reflected
all
of
the
things
that
I
said
in
the
in
this
given
talk
and
all
of
the
things
that
I've
said
are
inside
of
this
pr,
and
I
think
this
this
might
be
ready
to
merge
if
there
are
no
other
issues
with
the
code
itself,
so
yeah,
that's
it
from
my
side.
D
D
I
don't
know,
if
that's
what
he's
thinking
but
to
aid
in
that
it
would
be
really
nice
if
someone
else
on
the
call
regurgitated
it
briefly,
because
it
would
be
confidence
inspiring
to
hear
other
people
grocking
what's
going
on
as
well,
that
will
serve
two
purposes,
one
to
help
me
feel
better
and
then
two
to
maybe
round
it
off
for
deepak.
D
F
D
In
that
case,
it's
probably
not
fair
to
actually
it
won't
work
because
I'll
tell
you
the
answers
and
I'll
ask
you
the
question,
and
I
will
just
told
you
the
answer.
So
it's
a
weird
it's
a
funny
time
so.
F
D
Yeah,
okay,
thank
you.
You
know,
let
me
ask
you
here
like
here's
a
different
challenge,
and
this
is
the
challenge-
is
to
explain
just
super
high
level
what
it
means
to
see
the
content.
D
Yeah:
here's
a
there's,
a
challenge,
an
open
challenge
on
the
call
for
anyone
to
other
than
ashish
to
explain
just
high
level
what
it
means
to
seed
content.
E
No
about
the
having
few
sample
patterns
and
filters
like
sample
filters
that
are
provided
by
us
is
that
the
context.
D
Yep
very
good:
that's
that
yep!
That's
it
and
then
are
you
at
all
familiar
with
how
that
those
sample
patterns
and
sample
filters
sample
apps
how
they
initially
get
ingested
into
measuring.
D
Yeah
yep
yep,
so
yeah
science
and
say
more
used
to
say
it
louder,
say
again.
G
D
Yep
totally
spot
on
there
there's
a
couple
of
ways
that
it's
done
and
and
scientin
that's
spot
on.
Meghan
is
description,
is
accurate
and
then
scientists
as
well
at
build
time
if,
if
you're,
a
user
of
measuring
and
you're
using
a
measuring
container
image,
then
at
when
that
image
was
built,
sample,
content
or
is
seeded
sample.
Content
is
in
incorporated
in
the
container
image
and
then
is
ready
for
meshri
to
pick
that
up
on
boot
to
to
read
the
local
file
system
and
ingest
it.
D
D
And
so
that's
what
he
actually
talked
about,
which
is
makes
me
no
longer
wonder
why
no
one
had
any
comments
for
him.
D
Good
okay,
well
she's!
So
there's
a
second
bit
to
this
that
that
anyone
who's
interested
can
watch
the
video
again
and
deepaks
can
help
document,
and
so.
B
So
that
that
doesn't
that
doesn't
become
an
issue
later
on.
So
for
okay,
I'm
assuming
that
my
screen
is
visible
so
for
patterns
and
filters,
we
have
a
central
repository
from
where
we
fetch
or
fetch
all
of
the
all
of
these
things.
So
there
is
no
hardcoding
there
whatsoever,
but
the
thing
is
for
applications.
We
don't
have
any
such
mechanism.
We
we
basically
either
go
to
different
repositories
to
fetch
one
yaml
file
or
we
fetch
multiple
yaml
files,
combine
them
into
one
yaml
file.
So
there
is
no.
B
B
So
if,
in
future
we
think
about
bundling
more
applications
and
we
don't
have
a
central
way
to
bundle
them
all,
you'll
have
to
do
is
to
first
figure
out
that
is
that
one
yaml
file
or
multiple
yml
files-
if
that
is
just
one
yaml
file,
just
add
one
entry
into
this
array
or
the
with
the
name
of
that
tml
file
and
the
url,
and
if
we're
going
to
script,
multiple
yaml
files
and
compile
combine
them
into
one
ml
file,
then
add
one
entry
here,
so
basically
there
would
be
minimal
code
change,
but
I
cannot
think
of
any
other
way
to
avoid
this
code
change,
because
there
is
no
central
repository.
B
D
D
External
file:
well,
no,
I'm
sorry.
It
have
environment
variables
that
would
list
off
the
sample
content
to
be
imported
or,
alternatively,
in
the
local
provider,
go
code
to
not
necessarily
hard
code,
those
locations,
understanding
that
that
can
be
fragile,
but
rather
to
have
those
hard-coded
locations
externalized
in
a
local
seedcontent.yaml
file.
D
B
And
where
exactly
would
that
one
yaml
file,
consisting
of
that
doesn't
need
to
be
yaml
file,
but
whatever?
Maybe
it's
a
json
yaml
whatever?
Has
that
this
metadata
about
what
are
the
urls
and
where
would
that
be
stored?
According
to
you,.
D
Yeah
probably
like,
like
in
terms
of
source
code,
it
might
go
under
the
inst
in
the
measuring
repo,
like
under
installation
deployments
or
installation
instead
of
deployment.
It
might
be
like
content
or
sam,
you
know
and
then,
as
it
gets
deposited.
You
know
during
build
time
as
it
gets
deposited
on
the
file
system,
whether
you're
building
the
container
image
or
you're
running
meshrey
locally,
that,
if
you're
running
measuring
locally,
you
would
look
to
slash,
install
and
you'd
find
the
json
file
there.
If
you
built
it,
then
we
would
still
just
deposit
it.
B
D
Yeah
right
yeah,
wherever
wherever
it
is,
and
in
whatever
form,
if
it's
in
go,
if
it's
externalized,
if
it's
yeah,
it
will
have
to
like
establish
the
thing
that
you're
doing
is
kind
of
establishing
precedence
for
where
we
all
and
anyone
who
contributes
to
measuring
like
has
to
understand
that
there's
sort
of
essentially
configurable,
you
know
configuration
items
and
that
yeah
it
is
something
we'll
have
to
like
include
in
the
contributing
guide.
Hey,
here's
the.
D
And
so,
by
the
way
that
was
just
off
the
hip
like
it,
you
know
understanding
that
you're
you're
trying
to
fend
off
against
any
fragility.
That's
there
externalizing
that
reference
might
be
helpful.
It
might,
I
think
you
know
a
lot
of
times
in
the
days
of
old
and
maybe
for
some
of
us
it's
current,
but
if
you're
writing
java,
what
do
you
call
those
in
java
when
you
externalize
config?
What's
it
called.
D
I
don't
know,
no
one
can
remember
yeah
anyway,
when
you
yeah
anyway,
externalize
config
is
good.
I
forget
what
it's
called
in
java.
D
D
Add
change
that
and
then
run
measure
supply
that
as
an
environment
variable
like
the
environment,
variable,
though
that
would
get
supplied
to
the
container,
to
the
docker
to
the
docker
command
to
the
docker
file.
It's
like,
should
we
itemize
every
single
one
of
them
in
there?
No,
we
really
only
want
to
specify
that
list
once
so.
The
environment
variable
would
probably
be
a
file
path
or
a
url
uri
to
the
json
file,
whatever
custom
json
file
that
they
want
to
provide
that
supplies.
Those
names
that
json
file
has
to
adhere
to
your
spec.
D
D
D
D
Now,
as
he
you
know,
the
hairs
on
the
back
of
his
neck
are
telling
him
geez
somebody's
going
to
forget
about
this.
You
know
these
links
are
going
to
move.
How
do
we
make
sure
that
it
isn't
fragile?
How
do
we
make
sure
that
people
are
aware
of
this?
Where
should
I
document
that?
How
do
we
call
it
out?
What's
the
central
point
for
like?
D
Where
should
be
the
central
point
like
well,
if
you
took
it
a
step
further
than
what
I
said,
it's
like
hey
in
a
measuring
docks,
there's
docks
that
say
here
is
the
hipster
app
all
the
sample.
Apps,
here's
the
hipster
sample
app
the
emoji
photo.
The
book
store
the
book
info
actually
like
not
repeating
ourselves
there
as
well,
it's
like
if
those
were
one
in
the
two,
if
those
were
referencing
the
same
two
lists,
if
that
json
was
also
being
interpreted
by
jekyll,
to
produce
the
page
that
describes
those
apps.
D
A
Thank
you.
So
we
have
an
ongoing
initiative
in
within
measuring
machinery
ui,
which
is
related
to
state
management
with
redux.
So
there
is
a
whole
refactoring
process
going
on
in
that
area
and
scientin
had
an
update
on
some
of
the
work
he
has
been
doing
in
that
area.
So,
scientist.
A
Probably
aditya
can
fill
in
on
the
rest
of
the
details
regarding
the
this
initiative.
G
Yeah
sure
I
can
talk
on
on
my
update,
so
I'm
sharing
my
screen
once.
G
So
basically,
this
is
the
state
management
state
management
tracker
where
we
are
actually
tracking
the
the
components
which
are
being
done
by
whom
so-
and
we
have
some
contributors
over
here.
So
from
my
last
update
I
have
given
in
the
previous
call
that
I
have
introduced
three
new
updates.
So
basically
I
have
changed
this
button.
Colors
to
that
of
this
app
part,
then
the
next
update
is
that
so
you
can
see.
G
Now
previously,
this
link
was
from
3000,
but
right
now
it's
changed
to
4001,
and
actually
this
will
check
if,
like
4000.1
is
already
bind
by
some
other
app,
so
it
will
go
to
4002
and
the
provider.
Ui
will
also
go
to
4003
and
like
that
it
will
increment
till
it
get
a
blank
like
empty
port.
So
in
that
port
this
will
get
deployed
like
the
ui
server,
and
this
was
basically
done
because
in
localhost
3000
we
have
graphene.
G
So
we
have
to
change
this
to.
I
have
changed
this
to
4001
and
the
third
one
was
that
previously.
If
we
click
on
these
things
and
nothing
was
working
but
right
now,
if
you
click,
it
shows
that
the
kubernetes
ping
successfully,
but-
and
this
is
real
time.
So
if
I
close
my
kubernetes,
you
can
see
that
it's,
it's
like
an
error,
and
in
now
this
time
we
have
done
it
through
our
notification.js.
So
you
can
see
in
the
tracker.
G
G
G
G
And
we
also
have
some
responsive
issues
over
here,
I'm
not
so
good
in
fixing
the
responsive
issues,
but
you
can
find
these
things
are
getting
overlapped
so
yeah.
Basically,
that's
also
needed,
like
the
responsive
things,
yeah.
G
D
I
guess
we
lost
a
usherman
anybody,
so
I
know
aditya.
You've
done
certainly
done
work
in
this
area
and
we're
finished
anywhere.
You
know
working
on
some
other
items
this
last
week,
but
is
anyone
else
so
megan
I
just
jumped
off.
I
ushman
just
jumped
off
santan.
Will
you
follow
up
with
the
two
of
them
to
just
confirm
whether
or
not
they're
still
tracking
on
these,
and
if
not,
let's
swap
out
their
names.
G
D
Yeah,
sorry,
if
they
are
working
on
them,
fantastic,
if
they
could
hint
as
to
when
they
think
they'll
have
a
review
ready.
G
A
Oh,
thank
you
and
so
moving
on
to
the
next
topic.
Rudraksha,
could
you
give
an
overview
of
your
spec
on
night
talk
as
a
measuring
component
yep,
so
I
will
quickly
share
my
screen.
A
A
The
current
scenario
is
that
the
nighthawk
service
sites
in
mystery
container
and
when
you
request
a
test
that
services
executed
ad
hoc
and
the
whole
stick,
is
about
splitting
it
out
of
the
container
so
that
we
can
build
further
things.
On
top
of
my
talk,
that
night
doc
supports
and
also
to
shrink
mysteries
container
size-
and
there
has
been
a
history
through
to
this-
I
don't
have
the
right
now,
but
there's
an
open
pr
and
also
other
than
that.
The
spec
addresses
some
of
things
around
a
new
component.
A
The
design
goals
here
are
externalizing,
nighthawk
and
lifecycle
management
and
basically
we'll
be
needing
the
nighthawk
service
in
the
mystery
puff
container
and
other
load
generators
I'll
keep
them
out
of
the
context
for
now
to
keep
this
short.
So
the
nighthawk
service
is
the
one
which
runs
tests
and
receives
configuration
and
other
than
that,
I
don't
think
that
we
should
need
another
binary.
I
was
doing
some
comparisons
of
the
output
performed
by
the
the
transformers,
and
this
is
the
go
transformer
that
we
have
like
the
purgo
that
was
added
recently.
A
A
So
I
guess
we
should
be
good
to
go
with
the
co-transformer
alone
and
we
don't
need
a
nighthawk
output
transform,
which
is
the
traditional
way
of
converting
the
output
produced.
When
I
talk
to
this
other
than
that,
it
highlights
some
things
around
night
of
life
cycle
management,
and
I
propose
that
in
the
container
in
the
mystery
perf
container,
we
have
the
nighthawk
service
running
and
ready
to
receive
this
and
the
mystery
perf
container.
It
has
a
grp
service
message.
D
Well,
question
question
on
the
output
transformer.
A
A
Some
of
them
were
because
the
result
returned
by
output,
transform
some
of
the
data
types
were
not
accurate,
so
we
converted
some
of
them
to
strings
and
some
of
them
to
pins
and
similar
stuff-
and
this
was
the
previous
one.
This
would
be
the
new
one,
so
yeah
I'll
also
do
it
again
like
for
everyone's
satisfaction,
put
some
more
screenshots,
and
this
is
actually
I'm
using
for
your
ui
for
showing
this
up
and
not
measuring
ui,
because
that
would
need
I
mean
integrating
it
into
mystery,
and
that
stuff
would
need
some
stuff.
A
I
mean
yeah,
it
should
be
so.
The
thing
is
that
inside
mystery
we
also
convert
the
incoming
result
to
a
struct
that
forty
actually
provides,
which
is
which
is
officially
golan
and
nighthawk
has
protos,
which
are
used,
for
I
mean
which
are
with
align
with
the
fourth
yo
api,
but
not
exactly
the
same
as
that.
There
are
some
discrepancies
and
that's
exactly
what
we
had
hacks
for
in
the
earlier
transformer,
but
anyways.
D
Okay,
nice,
okay,
so,
okay
to
regurgitate,
then
what
you're
concluding
is
that
the
issue
that
we
were
facing
before
probably
isn't
there
in
the
nighthawk,
the
c
plus
version
of
this
transform,
which
is
from
the
nighthawk
project
we
rather
than
using
that
separate
c
plus
output,
transform
that
that
binary
that
we
would
go
ahead
and
use
our
our
own.
Instead
of
using
the
current
model
that
we
have,
we
would
probably
just
rewrite
that
again
quickly
and
go.
A
D
Okay,
okay,
gotcha,
okay
yep,
so
the
decision
is
really
like:
hey
look.
These
two
things
are
essentially
equivalent,
the
you
can
choose
between
c
plus
or
go
and-
and
that
has
a
major
effect
in
terms
of
dealing
with
the
binary
or
dealing
with
the
library.
And
so
we
already
put
forth
the
effort
to
write
the
library
and
go
and
to
fit
into
the
project,
and
that
was
why
we
created
well.
It
was
formerly
called
go
nighthawk
now
it's
called
get
nighthawk
and
so
great,
okay,
okay,
so.
D
Make
sense
to
me,
but
so
for
other
people
that
are
on
the
call,
like
you
know,
what's
going
on
here
is
we're
looking
at
a
new
measuring
component,
so
we
have
measuring
adapters.
We
have
measuring
ctl.
D
D
D
We're
going
to
take
one
of
the
load,
generators,
nighthawk
and
externalize
it
from
the
measuring
container
for
purposes
of
eventually
doing
distributed
load
generation.
So
having
multiple
instances
of
this
meshery
perf
container,
that
has
a
copy
of
nighthawk
inside
generating
look
any
num.
You
know
spinning
up
any
number
of
those
generating
lots
and
lots
of
load,
doing
more
in-depth
analysis,
and
so
this
measuring
so
measuring
perf
as
a
new
container
becomes
a
new
measuring
component,
and
so
that's
the.
D
What
root
rocks
is
walking
through
the
considerations
around
you
know,
doing
that
and
c
plus
plus,
and
go
and
container
and
then
using
mesh
kit
and
how
to
communicate,
and
you
know
how
to
orchestrate
this,
how
to
make
sure
that
we're
communic-
you
know
I
just
said,
communicate
but
part
of
that
communication
is
like
you
having
a
data
sync
where
the
test
results
are
coming
back,
they're
being
coalesced
into
a
single
larger
test,
result
part
of
what
we're
is
being
part
of
this
focus.
D
If
any
of
you
come
to
the
service
mesh
performance
calls,
because
this
is
the
this
is
the
measuring
one
and
they
overlap
intentionally,
but
on
the
service
mesh
performance
calls.
We
talk
even
more
about
adaptive,
load,
control
and
writing
custom
adaptive
load
controllers
to
help
users
answer
questions
like
what
is
my
optimal
service
mesh
configuration
having
an
adaptive
load
controller
will
help
people
run
recursive,
optimization
routines
using
via
perform
via
characterizing
their
performance
using
nighthawk
using
this
external
flow
generator.
D
D
D
A
A
So
this
is
just
very
brief,
and
I've
been
calling
off
this
for
some
time.
So
this
is
about
biscuit
errors
and
basically,
we
move
to
musketeers
messages,
are
detailed
errors
which
take
you
through
all
the
description,
long
description
cause
into
this
remediation
of
any
relation,
and
we
have
some
pretty.
We
had
some
efforts
on
switching
all
the
errors
to
this
now.
Basically,
this
is
about
the
logger,
so
miskit
has
a
specified
logger.
A
You
can't
like
it
is
wrapped
it
traps
the
common
loggers
and
it
does
some
stuff
which
is
very
specific
to
miskit
logger.
A
I
mean
mesquite
errors
like
obtaining
the
error
code,
the
short
description
and
other
fields
that
are
in
a
meshkit
error
and
when
a
generic
error
doesn't
contain
these,
it
panics,
so
there's
a
way
to
fix
this.
However,
we
haven't
implemented
that
way,
because
we
want
to
enforce
testing,
and
you
can
see
this
pull
request
for
more
context,
and
here
are
the
talks
on
that.
A
A
When
an
error
actually
occurs-
and
this
is
a
panic,
it
would
shut
down
your
process
and
kill
your
container,
which
is
not
desired.
We
have
been
facing
these
kind
of
errors
too
much
and
nearly
every
week
I
see
at
least
one
or
two
of
these
getting
patched
or
getting
recorded,
which
was
pretty
annoying.
So
I
added
these
here
also,
I
added
this
vr,
which
is
showing
how
you
can
act
okay,
this
one
on
when
you
would
actually
use
a
biscuit
error
and
how
you
would
return
it.
So
that's
it
and
comments.
A
Yeah,
thank
you
for
noting
this
and
documenting
this.
That's
really
great,
like
comments
on
this.
Like
is
everyone
clear
what
this
is
and
like
what
mesh
guitars
are
and
what
it
does.
B
Before
coming
up,
all
of
this
documentation
on
all
was
because
of
frequent
panics
that
we
were
facing
because
of
these
errors.
So
my
proposal
was
to
use
the
ast
library
that
go
provides
and
to
do
some
static,
checking
at
the
build
time
at
the
compile
time.
Basically,
we
can
go
node
by
node
and
catch
these
errors
that
whenever
we
are
using
the
mastery
logger
are
we
passing
the
the
error
that
we
are
passing?
Is
it
actually
knows
at
the
build
time
that
whether
that
is
the
measuring
error?
B
Basically,
that
is
the
mesh
get
error
or
not.
We
can
catch
these,
and
if
we
see
that
that
particular
error
is
just
a
normal,
normal
error,
then
we
can,
you
know,
fail
at
the
compile
time,
because
that
is
easier
than
to
fail
at
the
runtime,
because
things
go
overlooked,
but
it's
impossible
to
figure
out
to
just
you
know
for
a
for
if
you
were
to
go
down
and
keep
checking
that,
whether
somehow
somehow
somewhere,
we
missed
something,
so
that
would
be
that
would
make
these
errors
disappear
forever.
B
So
yeah
that
that
was
kind
of
I
actually
have
created
the
issue
and
assigned
michael
to
it
so
yeah.
That
was
just
a
little
something
I
had
to
say
about
this.
A
So
just
a
quick
note
on
this,
we
could
have
put
this
into
utility
itself,
which
actually
does
the
error
checking
and
I
was
like
we
can
check
the
error
type.
The
utility
itself
can
check
the
error
type
and
michael
said
that
it
was
a
little
unfeasible
to
actually
check
the
error
type
using
go
ast.
So
I
I
don't
know
if
you're
doing
something
different
than
that.
But-
and
this
was
his
comment.
H
Yeah,
I
think
I
think
they
proposed
ashish
proposes
in
basically
more
towards
that.
Actually,
we
can't
get
the
type
of
error
just
reflect
off
and
I
think
it's
his
proposal
was
more
towards
having
a
having
a
having
a
tool
which
gets
linked
on
the
linker
stage
of
co-compilation,
rather
than
a
other
another
utility
that
we
run
manually.
So
what
would
happen
is
that
when
go
go,
compiler
receives
the
code
on
the
linker
stage.
It
performs
the
checks
automatically,
or
else
the
build
fails.
H
Compilation
fails
so
automatically
our
developer
would
probably
get
to
know.
I
think
that's
was
ashish
a
suggestion.
D
Like
I
guess
that
is
a
that
is
a
breaking
change
right,
at
least
for
the
next
time
you
try
to
compile
it.
It
would
be
a
breaking
change.
H
So
if
I
think
I
think
yeah
I
mean
it
would
not
compile
if
there
is
a
non-mushkid
error
there,
which
we
are
trying
to
log.
D
So,
but
assuming
all
right,
so
assuming
that
that's
the
case
and
we
go
forward
with
this
proposal,
yep
we'll
need
to
run
a.
We
need
to
run.
A
check
like
like
we'll
need
to
make
sure
that
we're
diligent
with
going
through
each
of
mesh
kits
depend
each
of
the
components.
B
And
I
think
it
would
it'll
it'll
have
to
go
and
mess
it
if
I'm
not
mistaken,
so
that
we
can
use
it
everywhere.
D
D
The
one
that
michael
was
commenting
on
it's
a
mess.
It's
a
pull
request.
I
guess
it's
a
closed,
pull
request
or
maybe
just
pointing
putting
a
pointer
from
that
closed,
pull
request
to
that
new
issue,
yeah
that
one.
D
A
I
know
we
have
two
other
topics
from
utkar
shanley
like
do.
We
need
to
discuss
this
on
the
call
now,
or
would
it
be
better
to
catch
up
on
this
later.
D
I'm
not
not
mine,
because
that
would
just
be
a
regurgitation
of
what
we
talked
about
on
friday,
I'll
put
it
into
a
slime.
Neither
mine.
A
Thank
you,
so
that's
it
for
today's
call.
We
are
10
minutes
after
so
that's
that's!
Okay.
I
guess
like
in
terms
of
overshooting
in
our
meetings
and
yep,
we
have
we
reached
a
thousand
stars
on
the
measuring
repo,
which
is
really
great.
So
congratulations
to
all
the
200
contributors
and
all
the
people
in
the
call
or
278
contributors.
D
E
A
So
lee
mentioned
about
the
smp
call
it's
tomorrow:
it
will
be
with
the
cncf
tag
network
meeting,
so
join
us
there
for
to
learn
more
about
night
talk
and
what
we
are
doing
around
network
and
adaptive
load
controllers.
And
if
you
are
interested
to
work
in
c,
plus
plus
jump
in
as
well
so
I'll
see
you
guys
bye.