►
Description
Meshery Development Meeting (October 26th, 2022)
Agenda:
KubeCon 2022
[Gaurav] Project Office Hours Event
Get your Zoom backgrounds here
[Nithish] Help Wanted: Static list of components
Components are a way of exposing the capabilities of the underlying platform. In case of application development, Components are entities that are created/registered by the platform engineers and used by the operators.
Github workflow for generating and registering components
[Ashish] Migration update in postgres so far
[Harkirat] rjsf field error template #6439
[Harkirat] Fixing the behavior of json Schema with $id.
[Uzair] Fix search in deployment modal #6427
[Uzair] Detect in-cluster vs out cluster deployment #6431
[Sufyaan Khateeb] Changes in PR
[Pranav] WIP: Delete MeshSync data on deleting K8s Context #6435
[Ashish/Abhi] Subscription refactor [Draft-#6409]
[Ashish] Enforcing provider #6440
[Ashish] Adding MeshModel components into the database
[Ruturaj] Demo for workflow to update meshery.io/catalog
For publishing the applications publicly the user’s public design can be linked to their member profile.
[help wanted] UI for publish workflow
[Nishant] Deprecate meshery/homebrew-tap for publishing mesheryctl.
Join the community at https://layer5.io/community
Find Meshery on:
GitHub: https://github.com/meshery
Twitter: https://twitter.com/mesheryio
LinkedIn: https://www.linkedin.com/showcase/meshery/
Docker Hub: https://hub.docker.com/u/layer5/
A
A
Meeting
today
is
October
26th
and
before
we
begin,
I
I'll
share
the
meeting
minutes
in
chat.
So
you
all
can
write
your
name
in
the
attendance
list
and
we
do
follow
our
tradition
to
welcome
newcomers
on
every
call.
So
if
you
are
joining
this
call
for
the
first
time
feel
free
to
unmute
yourself
and
give
a
quick
introduction.
I
see
couple
of
newcomers
on
the
call.
A
A
Oh
okay,
so
should
I
get
my
introduction
yeah,
you
can
basically
I'm
a
student
pursue
my
engineering
in
Bangalore
and
I've
recently
joined
mashery
and
I'm
working
on
machidi
CTL
on
on
some
issue.
So
today,
Lee
Lee
had
asked
me
to
just
present
whatever
I
have
done
in
the
issue,
so
I'll
be
looking
forward
to
that
yeah.
That's
great!
Thank
you
sufian!
A
So,
let's
start
with
first
Urgent
items
so
today
we
also
have
coupon
2022
for
Project
R
mystery
project
officers.
So
for
that
you
all
can
get
RSVP
from
here.
Let
me
share
the
link
in
chat,
so
I
do
encourage
everyone
to
join
us
for
the
Mysteries
office
project
officers
and
if
you
do
join,
do
get
yourself
a
zoom
background
which
represents
layer
5
as
a
community,
so
you
can
also
get
the
layer,
5
Community
backgrounds,
from
the
layer,
5
Community,
Drive.
B
He
just
messaged
that
he's
that
he's
held
up
today
so
yeah.
Let
me
speak
to
that
real
real
fast
on
the
dishes,
behalf
just
being
familiar
with
his
work,
so
so
everybody
there
are
I'm,
not
sure
what
that
link
goes
to,
but
we
may
as
well
click
it
so
nitish
Karthik
has
been
working
on
a
collection
of
custom
resources,
a
collection
of
components.
So
this
is
you
know.
B
One
of
the
ways
that
messery
manages
infrastructure
is
is
by
well
digesting
and
comprehending
kubernetes
custom,
resource
definitions
and
creating
components
out
of
them,
and
so
the
project
has
been
doing
this
programmatically
for
10
different
service
meshes
for
a
while
now
and
in
the
this
new.
This
upcoming
release,
v0.7
man
I,
can't
wait
till
we
get
to
1.0
by
the
way,
but
v0.7
we're
taking
that
same
approach
and
doing
it,
which
is
generating
components
for
meshery
to
manage
we're
doing
it.
Programmatically
across
about
500
different
apps.
B
I
guess
is
the
way
of
saying
it:
500
different
kubernetes
workloads
and
then
for
each
of
those
500.
There
are
on
average,
maybe
five
individual
components
so
five
times
five
I'm
gonna
carry
the
one
anyway.
There's
gonna
be
a
lot
of
components,
and
so,
when
the
dish
Karthik
is
is
working
on
is
just
the
is
the
pre-processing
of
the
components
of
component
generation,
which
is
to
say
when
a
user
runs
meshery
or
it
should
be
cognizant
of
and
ready
to
manage
any
number
of
pre-existing
like
pre-released
software.
B
Why
so?
You
want
to
make
sure
that,
as
measuring
goes
to
interface
with
you
have
all
these
Integrations
and
interface
with
these
other
systems
that
it's
ready
to
do
so
as
soon
as
someone
downloads
measuring
that's
kind
of
one
goal.
Another
one
is
that
it
could
manage
not
just
the
current
version
of
Argo
as
the
example,
but
also
the
prior
version
of
Argo
CD
in
the
prior
version
to
that
in
the
private
version.
B
B
So
there's
a
lot
of
there's
some
sophistication
or
some
complexity,
rather
inside
of
that
type
of
maintaining
that
type
of
compatibility,
and
so
we're
trying
to
do
it
as
programmatically
as
possible,
which
is
what
this
effort
is
about
and
so
turns
out
that
there's
a
portion
of
the
engineering
that
happens
for
meshery
that
happens
in
spreadsheets,
so
we
may
have
to.
We
may
not
want
to
talk
about
that
publicly.
B
We
do
a
lot
of
this
type
of
processing,
pre-processing
in
GitHub,
workflows,
and
so
some
of
the
work
here
is
being
done
in
that
way.
Measuring
in
that
workflow
will
reach
out
to
artifact
Hub,
so
in
a
very
real
way,
it
grabs
all
10,
000,
Helm,
charts
and
digest
them
crunches
through
them
pulls
out
components,
generates
components,
pulls
them
out
and
then
makes
them
manageable
by
measuring
boy.
We
should
go
tell
that
story
somewhere.
That
sounds
impressive,
so
that
that's
in
flight
good.
C
D
Yeah,
just
a
second
I'm
having
cold.
So
if
at
any
point
my
voice
is
incomprehensible
or
inaudible,
just
let
me
know:
okay,
okay,
so
we
are
migrating
to
postgres.
We
were
using
sqlite
up
until
this
point,
so
there
are
a
bunch
of
considerations
that
have
that
that
are
to
be
made.
There
is
an
open,
PR
I'll
provide
the
link
to
the
pr
there,
and
so
recently
a
liveness
probe
is
added
to
the
communities
deployment.
So
we
have
communities,
deployment,
Docker
compose
deployments
and
you
can
deploy
it
locally.
D
So
for
kubernetes
there
was
a
live
mess
provided
for
the
database,
so
that
if
database
crashes,
then
it
can
be
restarted,
and
this
is
a
TCP
based
lightness
probe.
So
basically,
we
will
try
to
check
if
postgres
is
there
using
the
TCP
probe
and
we
don't
need
a
separate
Helm
chart
for
our
database.
The
database
is
currently
the
way
our
this
current
model
works
is
by
current
I
mean
the
one
in
the
open
pull
request.
Is
we
have
made
the
database
pluggable,
so
you
can
plug
your
own
database?
D
D
So
CLI
contributions
are
the
easiest,
so
we
would
need
a
message
ETL
system
database
as
well
as
this
check
to
be
there
in
the
message
CDL
system
check.
When
we
do
the
pre-flight
checks,
we
would
want
to
check
if
the
database
is
present
or
yeah,
and
this
would
need
to
be
added
in
the
config.am.
So
there
is
a
bunch
of
considerations
around
around
messy
CTL.
So
if
you're
a
CLI
contributor,
this
will
be
a
kind
of
a
legend
issue.
I'll
create
some
tasks
in
it.
D
So
if
you
want
to
get
get
assigned
some
of
these
things,
then
you
can
right
off
the
bat
you
I
can
start
using
it
and
I'm
already
have
a
running.
So
this
is
like
this
is
currently
I
am
connected
to
a
postless
instance
which
is
running
inside
my
container
here
and,
as
you
can
see,
I
have
all
the
tables
all
the
basically.
These
are
all
the
tables
that
Misty
uses
the
applications
filters
Etc,
so
I
have
all
my
activities
objects.
These
objects
are
also
the
objects
that
you're
seeing
here.
D
I
have
my
applications.
I
have
my
okay,
so
yeah,
then
I
I
deleted
the
designs,
while
I
was
testing
things
so
I'll
just
demo
with
the
application
so
yeah.
There
are
a
bunch
of
the
duplicates
here
because
you
know
I
was
connect
I'm
connected
to
the
same
database
instance
and
I'm
seeding
again
and
again,
when
the
message
instance
starts.
So
it
has
seen
in
these
components
in
the
postgres
instance.
D
Now,
if
you
want
to
know
the
status
of
or
what
instance
you're
connected
to,
if
you
go
here,
you
can
see
a
few
things
like
this.
Logger
field
can
be
removed
like
you.
Can
users
can
get
stuff
like
what's
the
in
like
what's
the
they're
using
postgres
or
some
other
engine
and
what's
the
address
of
the
database,
etc,
etc?
And
if
you
do
a
post
operation,
which
has
this
body,
you
can
change
stuff
like
username,
you
can
change
host
all
that
on
runtime.
D
You
can
change
port
in
this
case
I
to
save
time
and
not
do
curl
and
all
I
what
I
did
was
I
when
I
made
this
request,
this
actually
went
and
changed
the
port.
So
if
I'll
reload
I'll
see
that
the
port
is
5433
now
and
this
port
is
basically
now
pointing
to
this
database
and
the
other
one
and
if
I
look
at
here,
I'll
I'll
see
an
empty
database
here,
because
this
was
connected
later
on.
Okay,
this
is
yeah.
D
Actually
my
message
actually
updated
when
I
switched
the
database,
as
you
can
see,
the
new
objects
were
added
to
the
to
the
new
instance
yeah.
So
there
are
a
few
bugs
here
and
there
in
this
implementation.
There
are
few
error
logs.
You
might
see
here
these
like.
Actually
the
objects
are
being
added,
but
still
the
database
is
closed.
This
this
is,
this
element
is
coming,
I
know
where
this
coming
from.
D
So
basically,
if
you
cleanups
are
required
here
and
yeah
other
than
that
this
PR,
as
you
can
see
now,
in
my
new
instance,
I,
have
the
all
the
data.
My
message
has
pumped
the
data
into
my
new
instance.
So
all
both
my
postgres
instants
are
hydrated
with
the
with
the
with
the
with
the
data
and
if
I
come,
but
it
only
this
hydration
only
happens
for
the
year
for
you
for
your
communities.
D
So
these
objects
are
communities
objects.
So
when
you
connect
to
a
new
database
instance,
will
dynamically
get
all
the
data
from
your
cluster
and
also
hydrate
your
new
database,
although,
if
I'll
go
in
application,
I'll
see
no
applications
found,
even
though
I
had
like
four
here,
because
this
is
a
new
instance
right
now.
There
is
no
logic
that
when
you
connect
to
a
new
instance,
we
already
also
see
the
new
data
there.
So
maybe
that
can
be
added.
D
Maybe
that's
no
big
deal
but
anyway,
so
you
can
have
different
database
instances
with
different
content
and
and
all
and
yeah.
That's
that's
just
example
of
the
flexibility
that
we
can
achieve.
One
of
the
reasons
for
this
migration
to
postgres
was
to
have
like,
as
the
name
of
the
talk,
suggests,
workflow
engine.
So
as
we
move
towards
a
workflow
engine
like
temporal,
it
would
require
its
own.
It
would
require
connection
to
a
database
and
that
database
is
mandated
as
postgres.
D
So
we
will
make
use
of
the
same
database
and
in
metric
Cloud
that
we
have.
We
also
use
the
same
database,
so
in
cases
of
on-prem
deployments,
we
can
like
reuse
the
same
postgres
instance
for
all
the
messy
stuff
and
all
the
message
cloud
stuff,
and
so
there
are
a
bunch
of
utilities
for
this
approach
and
and
that's
kind
of
it
in
this
particular
task.
So
far,
the
other
tasks
are
down
below
so
I'll
come
after
other
guys.
D
D
B
A
quick,
quick
question:
she
you
had
I
didn't
catch
it
all,
but
you
had
switched
out.
Maybe
the
database
on
the
Fly
and
I
didn't
know
if
that
was
that
was
like
taking
the
same
type
of
database
like
postgres
and
dumping
it
and
kind
of
restarting,
or
was
that,
like
switching
actively
switching
from
one
type
to
the.
D
C
D
Actually,
actually
I
yeah
I
didn't
do
anything.
Everything
is
done
by
Gom.
So
all
the
Reddit
goes
to
Gom.
So
Gom
is
an
orm
library
that
we
use
here
and
it
lets
us
which
the
underlying
engine
that
we
use
so
either,
whether
it's
sqlite
engine
or
postgres
engine.
So
whenever
we
switch
the
database,
all
I
do
is
just
I
create
a
new
database
instance.
So
we
have
only
one
Global
instance
at
a
time,
so
I
acquire
a
lock
on
that
instance
and
I
create
this
new
instance
for
whatever
this
new
Json.
D
So
there
you
saw
the
field
type.
So
if
you
on
the
runtime,
if
you'll
say
not
postgres
but
sqlite,
I'll
create
a
new
database
instance
pointing
to
that
sqlite
file
and
all
and
assign
that
to
the
global
instance
so
and
how
the
how
it
basically
creates
like
in
the
whole
internals
is
managed
by
most
of
it
is
managed
by
Gom.
D
If
you're
interested,
you
can
look
at,
you
can
go
and
mesh
kit
and
there
is
the
code
there
which
interfaces
with
both
the
postgres
in
postgres
and
sqlite
stuff,
but
that
interface
is
like
very
the
retention.
Mostly
most
of
the
code
is
in
bomb,
so
we
don't
have
to
take
care
of
any
of
this.
That's
why
this
migration
was
like
very
straightforward.
A
E
E
Yeah
definitely
so
actually
we
have
moved
to
like
rjsf
V5,
currently
like
past
week,
so
some
of
the
changes
were
underlying
so
I
added
some
changes
to
the
template.
So
earlier
we
were
using
the
field
template
so
I
added
the
feed
error
template
to
control
the
errors,
that's
what
I
did,
and
the
next
item
is
like
the
behavior
of
Json
schema
with
the
dollar
ID.
So
actually
I
can
demonstrate
Ashish
what
it
was.
So
actually
so
you
see
this
like
this
is
the
validation
error,
so
the
replica.
E
So
whenever
I,
it's
like
the
retrieval,
your
retrieval
URI
and
the
base
URI,
the
is
the
layer,
Five
Dot,
maybe
definition,
workload
application.
So
whenever
I
add
this,
so
the
all
the
validation
are
suppressed.
It's
for
the
all
the
global
thing
adjacent
schema
that
are
using
the
that
are
all
like
everyone.
Using
this
Json
schema
with
the
dollar
ID
stuff.
C
C
E
E
E
E
A
E
It's
here,
I
guess
so,
let's
see
here
it's
here
like
you
can
see.
So
it's
the
base,
urine
and
I
was
like
thinking
around
the
reading.
The
Json
schema
dog.
So
what
they
have
is
like,
like
it's
like
the
current
retrieval
UI,
the
that
this
thing
is
like
being
validated
against
the
base,
your
right
stuff.
So
that's
why
it
suppressed
most
of
the
errors
like
I
can
send
you
the
link.
If
you
can.
C
D
E
A
A
A
The
Surge
went
out
of
scope
after
the
first
key
stroke,
so
the
fix
was
this.
Work
was
introduced
by
due
to
one
of
the
this
positive
regression
issue
introduced
in
this
PR.
Basically,
we
were
managing
the
state
incorrectly.
So
this
was
the
first
item
and
the
second
item
is
detection
of
type
of
measure
deployment.
A
So
when
measuring
is
deployed
using
a
machine,
serial
system
starts,
the
measure
is
measuring
container
and
all
the
things
are
deployed
inside
the
cluster,
and
in
that
we
don't
have
access
to
contact
to
any
of
the
users
context
like
User
Group
config.
So
if
we
generate
the
the
cube
config
and
we
gave
them
give
the
name
as
context
name
as
in
cluster
and
then
up
till
this
point,
we
know
that
the
deployment
is
in
cluster
but
later
on,
we
lose
the
information
that
whether
the
deployment
is
in
cluster
or
cluster.
A
So
basically
I
introduced
a
new
New
Field
in
the
k8s
context,
struct,
and
then
we
are
persisting
the
type
of
deployment
in
the
in
that
context
in
that
struct.
So
so,
yes,
this
is
how
we
are
maintaining
the
type
of
deployment
and
there
is
a
dependent
PR
in
missionary
Cloud
as
well,
which
updates
the
k8's
context
table
schema.
A
So,
yes,
this
was
my
two
items.
This
one
happens
yes
and
any
one
have
some
good
name
or
consideration
of
what
should
what
should
be
the
name
of
the
context
which,
which
is
deployed
in
first
in
inside
a
cluster
like
we
currently
give
name
in
as
in
cluster
and
in
the
settings
page.
If
you
will
see
you
will
have
in
cluster
and
type
of
deployment.
Also
in
cluster
Ashish,
you
were
mentioning
giving
some
unique
name
like
something.
D
Yeah,
a
mod
of
ideas
for
now
I'll
have
to
think
about
it
because
yeah
we
are
creating
the
context
here
ourselves,
so
yeah
I,
don't
think
it
should
be
a
random
name
or
something.
But
what
attribute
to
use
like?
Maybe
something
we
can
query
in
the
cube
system,
a
namespaced,
but
again
we
yeah,
there's
a
bunch
of
considerations
around
here,
but
for
now
I
think
we
should
get
this
PR
merged.
So,
oh.
B
Yeah
and
the
question
the
question
that
you're
asking
was
there
is
about
Aesthetics
writers
about
the
term,
the
term.
Sorry,
the
terms
in
cluster
or
out
of
cluster.
Is
that
right
or
are
you
asking
about
how
to
universally
and
confidently
identify,
whether
you're
in
or
out
of
cluster,
it's
the
former
right,
it's
just
Aesthetics,
yeah,
and
so
so
yeah.
So
there's
really
nothing
to
investigate.
Ashish
is
just.
D
This
deployment
name
as
well
as
deployment
type,
so
on
line
number
265,
you
can
say
that's
the
context
name
that
you're
providing
as
in
cluster
and
here
below
is
just
include
the
type
so
of
course,
type
has
to
be
in
cluster
out
of
cluster.
That's
that's.
That
was
the
aesthetic
part.
The
logic
part
was
here:
what
to
name
the
context,
because
what's
gonna
happen
is
we'll
have
two
columns,
both
of
which
will
say
in
cluster
and
in
cluster
side
by
side.
D
B
B
Yeah
I
got
it.
It's
sorry,
yeah
in
context
cluster
that
confuses
okay.
That
is
it.
Thank
you
so
much
for
saying
something
for
coming
up
with,
like
brainstorming,
an
idea,
keep
the
ideas
coming.
What
else?
What
else
those
two
terms
are
so
darn
close
to
each
other?
They
get
confused,
so
much
I
think
putting
them
together
is
so
you
have
an
in
context
cluster.
Your
context
is
in
cluster
I.
Don't
know
my
context
is
defined
out
a
cluster
It's,
a
combination
of
credentials
and
a
pointer
to
where
my
cluster
is
that's.
B
Hussein
any
terms
that
you've
heard
floating
around
okay,
I'm
gonna,
deploy
this
application
in
cluster
B,
so
like
in
cluster,
is
kind
of
part
of
the
sentence.
I'm
going
to
do
this
in
cluster
I'll.
Do
it
in
this
namespace
I'm
going
to
run
this
locally
on
my
machine
I'm
going
to
do
that
out
of
outside
of
the
cluster.
B
A
A
A
So
I'm
just
going
to
say
what
changes
I've
made
like
I
have
made
changes
to
five
files.
I
was
working
on
this
issue.
Six,
two
seven
three:
it
was
related
to
measuri
exp
filter.
Basically,
the
command
was
not
working
because
the
API
endpoint
had
changed
or
it
was
depreciated.
So
I
just
changed
the
end
point
and
and
I
added
some
error
handling.
That's
all
I'm,
sorry
to
interrupt
you.
So
friends,
so
like
your
screen
is
black.
A
I
have
shared
it,
but
I,
don't
know
why
yeah
you
can
reshare
your
screen
or
maybe
the
full
screen
I
had
done
full
screen.
Okay,
just
okay,
can
you
say
it
now:
yep,
it's
fine
yeah!
So,
basically,
I
have
made
changes
to
these
five
files
here
and
the
the
issue
that
I
was
working
was
measure
the
CDL
exp
filter.
A
B
B
Yeah,
it
seems
arbitrary
that
we've
limited
it
to
GitHub,
but
but
yeah
good.
Well
so
this
is,
this
is
It's
Great
To
I
mean
I
mean
this
is
much
needed.
Obviously,
we've
got
the
wrong
pointers.
We've
got
we've
since
promoted
support
for
awesome
filters
out
of
experimental,
but
we've
you
know
neglected
to
update
the
references
here.
B
So
that's
helpful
one
other
thing
that
is
encouraging
about
the
work
that
you're
doing
here
is
well
one
there's
much
more
of
it
needed
Hussein
can
talk
to
you
about
that,
but
two
there's
this
this
errors
new
this
function
that
you're
using
is
absolutely
the
right
function
and
it's
one
that
it
ends
up.
Accepting
a
few
parameters
ends
up.
Accepting
a
lot
of
information
about
is,
like
it'll,
have
a
short
description,
a
long
description
it
can
carry
out.
A
B
A
B
I,
don't
know
this
is
yeah.
I
know,
that's
great.
This
is
an
opportunity
to
improve
them.
So
since
you're
kind
of
spending
time
in
here,
looking
at
a
couple
of
things
anyway,
the
empty
see
those
empty
strings
on
the
end
as
parameters
on
the
end
of
your
function,
yeah,
they
those
things,
end
up
getting
published
out
on
measure.I
or
docstat
mystery.io,
so
I'm
just
sharing
this
screen
so
Anita
or
Antoinette
Caldwell
she's
sick
this
morning,
but
is
normally
on
this
call.
B
She's
been
going
around
trying
to
get
a
lot
of
these
filled
in
so
the
error
codes
that
you're
working
with
in
Mastery
CTL,
they
all
they
will
be
automatically
published
to
this
documentation
page
and
so
the
empty
strings
that
are
there
that
are
not
filled
in
they'll
they'll.
Show
like
this
it'll,
be
like
a
it'll,
be
a
short
description.
It
may
be
the
same
thing
for
the
long
description,
which
is
not
helpful
and
then,
if
nothing
in
the
other
strings,
it
won't
say
well
probably
what
caused
this
is.
B
The
fact
that
there
was
a
malformed
kubernetes
manifest
for
your
istio
and
whatever
the
probable
cause
was,
and
here's
how
you
should
probably
fix
that.
Some
of
that
you
won't
know-
and
that's
okay,
but
to
the
extent
that
you're
in
there
and
it
kind
of,
makes
sense
it's
like.
Oh,
you
know
invalid
GitHub,
URL,
wow,
probable
cause
could
be
you're
not
connected
to
these
or
You
misspelled,
something
or
you're
trying
to
connect
to
a
product,
not
a
public.
B
B
B
B
A
I
don't
know
much
about
it,
so
I
have
no
idea,
no
worries,
I'll
test
it
out
once
and
file
an
issue
if
I
found
another
and
the
deploys
are
not
working
like
this
filter
is
a
very
new
I.
Try
to
search
about
it,
but
I
didn't
find
much
information
over
it
and
I
I
was
just
using
the
URLs
that
were
provided
in
the
examples
for
checking.
If
the
thing
is
working
or
not,.
B
B
B
Is
it
great
nice
update
and
a
for
effort
on
Mastery
CTL
and
like
bonus
points
for
taking
on?
Maybe
some
error
codes,
I,
would
say
kind
of
a
c
minus
for
like
picking
up
the
requests
to
file
a
GitHub
issue,
I
encourage
you
to
get
really
comfortable
with
filing
issues
to
help
you
engage
in
the
project,
but
awesome
thanks
for
doing
this.
Thanks
for
presenting.
A
A
Once
again,
so
there
was
a
recent
ask
of
about
when
we
delete
a
message.
When
we
delete
a
kubernetes
context,
then
it
should
delete
the
messaging
data
related
to
it.
So
first,
let
me
tell
you
what
was
the
background
information
about
it
so
when
we
deploy
measuri,
whether
Insider
container
outside
the
container
machine
automatically
detects
your
organic
kubernetes
cluster
through
your
default
Cube
content
and
as
soon
as
that,
it
deploys
this
custom
controller
like
Mysterio
operator,
mesh,
sync
and
broker
machine
broker.
A
So
machine
is
actually
responsible
for
determining
different
workloads
in
your
running
kubernetes
environment
as
soon
as
it
does
that,
then
it
sends
that
the
missionary
server
and
machine
server
stores
that
stores
them
into
the
your
local
database,
which
ashes
were
just
discussing.
So
one
thing
to
note
that
as
soon
as
we
delete
a
kubernetes
Content,
then
measuring
wasn't
deleting
the
messaging
data
related
to
that.
So
this
will
be
asked
about
deleting
the
data
as
soon
as
the
context
is
deleted.
So
I
was
just
working
on
this
I've
almost
reached
to
the
end.
A
A
But
I
got
a
little
more
reviews
about
enhancing
the
messaging
framework
based
on
that,
so
I
I
did
that
as
well,
but
I
need
to
do
a
little
more
changes
so
that
I,
don't
repeat
myself
as
in
I,
need
to
add
few
more
gone
tags
in
the
machine.
Stuff
and
I
need
to
use
similar
messaging
framework
for
refreshing
machine
function,
which
we
have
so
I
just
need
to
do
a
little
more
work
over
there,
so
other
than
that.
It's
still
working
for
this.
So
any
questions
or
feedback.
A
C
B
Yeah
it's
when
they
choose
to
delete
a
kubernetes
context,
which
is
to
remove
a
connection
we
I
think
provide
them
a
modal
that
gives
them
a
little
bit
of
info
to.
Let
them
know
that
meshri
operator
will
be
removed
from
the
cluster
measurable,
disconnect
as
a
natural
Consequence,
the
well
the
mesh
sync
day.
What
you're
doing
is,
like
you
know,
cleaning
up
what
is
otherwise
the
orphaned
data
from
mesh
thinking,
so
it's
static,
it
becomes
it
stagnates
pretty
quickly
anyway,
so
it
certainly
seems
like
the
right
thing
to
be
doing
clearing
it
out.
B
B
B
B
Wow
hey,
while
I'm
making
wishes
you
know
while
you're
like
you
know,
let
me
just
pile
it
on
unless
other
people
on
the
call
have
a
different
suggestion.
You
know
I
might
like
the
the
add
cluster
button
is
always
just
kind
of
looked
strangely
out
of
place
for
me,
I,
don't
know
if
there's
a
better
positioning
or
not
I
figured
I'd
call
it
out
just
in
case
just
in
case.
B
I
think
I
added
an
R
to
sufyan's
name,
but
if
you,
if
you
look
on
these
Pages,
they
conveniently
have
a
table
that
or
a
grid
that
takes
up
the
vast
majority
of
the
page
and
conveniently
at
the
top
they'll
have
like
import.
You
know,
filter
on
catalog
views
or
other
action
buttons,
which
I
think
you
know
fits
and
there's
kind
of
a
consistent
ux.
B
When
you
go
into
settings,
though,
the
ad
cluster
only
makes
sense
in
context
of
the
environment,
and
consequently
you
only
see
that
button.
You
know
within
the
okay
good,
so
maybe
somebody
else
on
the
call
has
a
better
idea
like
it
just
feel
to
me
it
still
feels
misplaced
or
miscolored
or
you
know,
I,
don't
you
know,
I
don't
know,
okay,
maybe
not.
Does
anyone
else
feel
like
this
or
is
it
just
am
I?
Do
I
just
need
to
get
used
to
it?
A
E
B
C
D
So
so
I
mean
I
I.
We
already
talked
about
it
this
in
the
previous
web,
testicle
and
I.
Don't
think
there
is
any
more
update
on
it.
On
the
I
mean
there
is
no
more
progress
on
the
work
done
that
day,
so
you
can
just
go
to
the
draft
PL
and
see
the
current
progress,
but
yeah
it's
being
delayed
due
to
some
other
work.
That
was
more.
You
know,
prioritized
so
I
think
this.
B
D
A
B
A
So
that
if
there
are
any
problems
or
like
The,
Homebrew
code
is
incompatible
with
our
release
system,
so
we
can
still
use
our
custom
tab.
But
since
it's
been
a
lot
of
releases
and
a
lot
of
time
since
we
shipped
it
there
and
I
I
personally,
don't
think
there
are
any
problems
with
that
with
using
Homebrew
core
as
the
main
channel
to
publish
mystery
CTL,
we
can
like
stop
publishing
mystery
CTL
to
our
custom,
Tab
and
add,
and
also
like
anyone
who
is
still
who
has
installed
message.
C
A
Be
adding
a
migrations,
dot
Json
file,
which
will
tell
a
home
view
that
it
like
mystery
CTL
has
been
shifted
from
this
tab
to
The
Homebrew
code
Tab
and
whenever
home
view
will
be
updating
the
message
CTL,
which
was
installed
from
this
tab.
It
will
know
that
further
updates
are
being
published
and
the
core
tab,
so
yeah
I
think
we
can
replicate
what
do
other
guys
think
what
do
you
all
think
it
should
be.
B
B
The
nice
thing
is
that
in
this
case,
you're
you
I
mean
you're,
really
close
to
this,
and
and
you
okay
and
so
I
guess
to
summarize
for
everyone
else,
it
used
to
be
the
case
that
we
were
using
go
releaser
to
create
the
master
CTL
binary,
which
is
treated
separate
from
the
rest
of
the
measuri
containers,
the
10
or
so
measuring
containers.
Those
are
treated
through
a
different
process.
B
Binaries
and
releases
them
is
not
compatible
with
the
core
way
that
Homebrew
as
a
popular
packaging
package
manager
for
Linux
and
Mac
OS.
It's
not
compatible
with
the
way
that
they
do
their
core
releases.
So
when
you
go
use
a
tool
like
Brew,
you're,
very
similar
to
Yum
or
apt-get,
that
you
a
lot
of
times,
you
can
just
do
Brew
install
the
thing
you
know,
Cube,
CTL
or
Docker,
like
the
thing
that
you're
looking
for
and
thanks
to
Nishant.
That
is
true
for
Mastery
CTL.
B
It's
just
Brew
install
Mastery
CTL
and
that's
a
really
Pleasant
improvement
from
the
prior
instructions.
First,
you
had
to
add
our
repo
like
grab
our
tap
through
install
Brew,
tap,
install
something
something
anyway,
it's
much
more
concise,
it's
official
and
so
on
and
nishana
has
like
I'm
just
repeating
the
same
thing.
He
said,
which
is
we've
been
on
this
news
system
for
15
well,
actually
like
25
releases
or
something
for
a
while,
we
hit
a
snag
or
two
as
you
might
expect,
not
and
but
generally
don't.
B
D
So
my
the
other
thing
was
there
was
a
open
ask
about
enforcing
a
remote
provider
in
certain
deployment
was
in
certain
deployment
instances.
So
we
would
like
to
have
this
meshri
some.
We
would
like
to
have
mercy
on
play:
Dot
question.io
and
their
unauthenticated
endpoints.
We
don't
want
unauthenticated
endpoints
to
be
there,
so
this
was
a
two-part
problem.
D
I
have
the
two
changes
that
I've
made
I'll
quickly
summarize
so
before
right
now
we
only
have
like
one
authentication
mechanism,
which
is
provider
authentication
and
the
other
endpoints
which
are
like
not
under
the
authentication
mechanism.
They
are
completely
by
so
we
don't
use
the
auth
middleware
for
those,
so
this
PR
creates
a
like,
introduces
a
consistency
such
that
all
the
no
matter.
What
Handler
you
add
on
whatever
endpoint
you
have
to
use
the
auth
middleware
and
then
the
second
part
comes,
which
is
that
we
can
extend
this
idea
of
Provider
authentication.
D
Why
are
this
Authentication
mechanism
enum?
So
you
can
have
multiple
authentication
mechanism
enums
inside
of
the
middleware.
You
can
have
your
so.
The
older
logic
would
be
like
when
you
have
your
provider
auth
and
for
certain
endpoints.
You
can
enforce
a
north
and
these
authentication
mechanisms
will
not
be
respected,
respected
if
you
pass
on
the
enforce
provider
flag
with
a
particular
provider.
D
So
while
deploying
your
machine,
if
you
say
and
force
provider
remote
or
enforce
provider,
local
or
whatever
you
enforce
a
provider
enforcing
a
provider
would
mean
that
no
other
authentication
mechanism
would
be
respected
other
than
the
provided
authentication
mechanism,
and
then,
even
in
that
thing,
we'll
only
allow
the
request
for
that
provider
the
what
are
the
use
case.
So
let
me
first
demo
it
real
quick.
So
the
right
now
I'm
logged
in
is
a
remote
provider
and
I
ran
this.
D
D
If
I
try
to
go
on
an
unauthenticated
endpoint,
which
was
something
like
slash,
API,
slash,
component,
slash,
kubernetes,
I'll
just
get
a
401,
and
if
I
try
to
log
out
right
now
and
log
back
in
as
Lund
provider,
I'll
again
get
a
401.
So
basically
no
request
is
allowed
other
than
the
remote
provider.
So
similarly
you
can
even
enforce
an
unproiter,
but
I
wouldn't
see
a
use
case
for
that.
D
But
for
a
remote
provider
we
have
a
use
case
and
what
the
reason
for
just
not
bypassing
the
auth
completely
and
having
this
type
authentication
mechanisms
so
that
right
like
in
future,
we
can
have
so
adapters
register
their
components
to
match
the
server
when
endpoint,
whether
we
can't
whether
we
cannot
use
the
provider
authentication
mechanism,
because
you
know
that
those
are
not.
Those
are
systems
and
not
actual
users,
so
for
systems
tomorrow,
like
we
can
have
another
authentication
mechanism,
some
key
value
based
something
like
that.
D
Not
here
anyways
I
mean
that
you
we
create
some
secrets
and
all,
and
that
can
oh
and
that
change
will
be
additive,
or
these
are
like
a
switch
case
statement.
So
you'll
have
something
like
models.that
particular
authentication
mechanism
and
it'll.
Add
the
logic
here
and
then
use
whenever
you
basically
a
while
instantiating
in
the
I
in
the
server.go,
when
you
actually
so
yeah
the
file
is
Big.
So
it's
if
I
load
the
div
so
yeah.
D
They'll
have
to
know
that,
okay,
so
if
I'm
providing
provider
what
it
means
I'll
be
using
session
injector
so
to
have
the
all
the
provider
stuff
and,
if
I'm
using
like
so,
if
I'm,
using
something
like
in
here
North,
it
means
these
are
endpoints
which
require
no
authentication,
if
I'm
adding
some
so
basically
to
make
these
things
a
little
more
explicit,
so
authentication
mechanisms
and
overriding
flag
to
enforce
the
provider,
authentication
mechanism
and
also
a
specific
provider
type.
This
will
yeah.
So
that
was
my
quick
demo.
B
Yeah
I'd
love
to
see
something
in
the
Mastery
architecture,
deck
there's
a
few
different
scenarios
that
need
to
be
accounted
for.
What,
if
someone
wants
to
switch
providers
after
deployment
or
what
are
the
you
know,
how
do
they?
How
do
they
do
it?
What
happens
when
you
add
a
third
one?
Can
you
have
two
enabled
at
the
same
time
or
just
you
know,
yeah.
D
B
Yeah
I,
actually
don't
want
you
to
speak.
The
point
is
to
write
it
down.
It's
like
yeah,
okay,
okay,
well,
there's
a
glass
scene,
it's
great
same
good
and
you're
right,
it's
exactly
the
other
part
to
write
down
as
well.
You
know:
it'll
drive
some
clarity
like
unintentionally.
These
are
unauthenticated
perpetually
unauthenticated,
apis
and
and
here's.
Why
and
here's
how
that's
good.