►
From YouTube: Config WG Meeting - 10/18/2018
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
B
The
previous
week
was
e,
hey,
so
welcome
everyone,
I
I,
created
a
bunch
of
items
in
the
agenda
related
to
status,
check,
feel
free
to
add
anything
that
you
guys
might
be
interested
in
discussing
about
so
the
main
items.
Hacking
of
importance
is
the
galley
MCP
work
and
making
sure
for
landing
one
one.
So
that's
in
that
list,
and
then
we
can
discuss
about
you
nobody's
interested
in
maybe
the
new
API
governor
small.
What
we
are
trying
to
start
is
like
there's
some
difference.
B
Links
from
like
the
previous
week,
I
created
this
spreadsheet
for
tracking
some
of
the
change
check-ins
to
the
meal
to
the
API,
deco
and
I'm
planning.
To
do
the
same
thing
for
the
see
IDs
as
well.
So
today,
I'm
gonna
go
ahead
and
send
this
to
a
review
discussion
list
so
that
we
can
get
started
on
looking
at
things.
B
B
B
C
B
C
B
B
B
C
C
C
C
C
B
C
B
C
C
Sponsors
he
should
be
you
suppose,
all
right,
I
have
looked
into
it.
Ok
or
Daniel
hasn't
looked
into
it.
Yet.
Okay
and
I
told
him
we
want
to
get
the
metrics
and
we
want
to
get
the
luck
scopes
and
but
it
was
fairly
straightforward
and
isolated
span
tracing.
C
C
B
essentially
need
to
sew
all
of
the
API
group
or
CID
handling
is
going
to
move
together
into
the
release.
Time
like
in
within
two
releases,
I
think
that's
the
right
time
to
start
moving
things
out
of
the
namespaces
and
the
different
API
groups
and
moved
out
to
the
new
ones,
mainly
because
generally
for
Yeley,
this
is
just
a
config
change
right.
It's
a
it
is
almost
trivial
for
other
parts
like
this
is
this
is
a
lot
of
work.
B
C
B
C
B
C
We
had
only
one
dude,
no
you
pump
a
bit
and
yet
to
tell
pilot
to
you
make
sure
that
in
a
first
one,
always
good
yeah.
If
you
had
an
existing
deployment
but
I,
imagine
you
don't
want
it
to
have
a
break
chocolate
layer.
You
want
to
minimize
downtime.
There
is
quarter
okay,
Calais!
Then
you
start
rolling
each
other
pilot.
Okay,
that's
fair
enough!.
C
B
C
Been
looking
at
past
types
of
it
so
far,
nothing
major
just
because
of
how
many
clients
we
practice
have
been
working
on
tooling,
mostly
so
that
we
can
characterize
it
something
more
useful
in
larger
deployments
in
one
to
when
that
big
time
is
required.
I
think
the
important
thing
here
is.
We
can
answer
questions
that
people
have
so
for
my
demo.
Installs
in
that
I
think
typically,
we're
gonna
see
four
one
one,
it's
fine,
but
then,
if
the
question
was,
how
far
can
it
scale?
How
many
pilots
or
mixtures
can
I
connect
affecting
the
morning?
C
C
We
don't
have
the
kubernetes
service
registry
integrated
with
gallery
at
this
point
once
that
Bland's
that's!
So
that's
a
one
two
work
item
one
sometimes,
then
performance
will
be
more
critical
because
those
are
not
either
authored
and
they're,
potentially
higher
children,
ok,
enlarger
and
number
so
here
yeah.
So
this
is
a
kind
of
a
well-known
try.
Not
this
one.
There's
there's
another
one.
That's
four
one,
two
there's
there's
a
handful
of
issues
and
there
was
a
few
emails
about
our
scoping
of
resources.
C
Yeah,
that's
not
going
to
make
it
for
one
one:
yeah,
it's
related
to
the
the
cluster
versus
namespace
car
back
also
they
can
fit
groups
said
we're
gonna
make
these
changes
they're,
potentially
breaking.
Unless
we
and
they're
invasive
to
the
runtime
components.
We
could
hide
that
complexity
in
galley.
C
C
B
D
C
B
I
think
the
regular
deprecation
kills
each.
They
apply
to
the
current
set
of
configurations,
regardless
of
whether
they're
getting
replace,
with
the
almost
exact
same
thing
right
in
a
cluster
scope.
So
what
I
my
mental
model
is.
Okay,
we
introduced
a
new
series
that
we
think
are
the
right
think
based
live
side
by
side
with
the
old
thing,
and
the
system
runs
that
way
like
in
dual
mode
way
until
the
deprecation
of
the
old
stuff
is
complete,
and
then
you
get
on
the
on
the
new
plan
that
gives
time
for
people
to
migrate
in
this.
B
So
that
means
I
think
that
I
heal
timeline
for
this
is
look.
We
began
starting
introducing
this
in
one
to
write
the
news
here.
He
comes
in
online
in
one
to
be
deprecated
old
one
in
one
to
as
well
gallery
becomes
the
default
model
in
one
to
default,
ingestion
mode
for
one
to
write
and
in
one
three
gathering
is
the
only
mode
for
injecting
Copic
and
hearing
monkey
or
maybe
in
a
subsequent
release,
depending
on
the
cadence,
the
replicates
that
goes
away.
B
B
E
C
E
Couple
questions
I'm
part
of
a
group
at
Salesforce:
that's
it's
adopting
SEO
and
we
kind
of
have
all
kind
of
miscellaneous
things.
The
first
one
really
is,
you
know,
what's
kind
of
the
best
mode
of
communication,
it
seems
like
rocket
chat,
is
essentially
a
black
hole
for
questions
decided
up
there
this.
If
the
mailing
list,
I
guess,
is
the
best
Avenue
or
so.
B
I
think,
as
a
user,
the
mailing,
like
yeah
rocket,
said
I'm,
not
sure
how
well
it's
staffed
at
this
point,
but
I
think
the
mailing
lists
are
generally
where
people
ask
questions
and
like
get
more,
all
right,
I
see
more
activity
there
like
as
I
use
it
I
think
you
may
get
more
traction
there
like.
If
you
want
to
contribute
to
the
codes
we
can
discuss
about
it
separately.
Okay,.
E
B
The
four
one
one
the
target
is
Flavia
following
the
standards
released
cadence
and
in
the
release
key.
In
that
parlance,
the
eonni
is
gonna,
be
an
alpha
in
one
one,
which
is,
which
means
it's
going
to
be
off
by
default
right,
there's
gonna
be
a
flag
that
allows
you
to
enable
that,
but
functionally
you
shouldn't
see
any
changes
right.
So
if
you
are
seeing
a
change
as
a
problem,
so
in
one
so
it's
gonna
go
to
the
next
release.
B
E
D
E
Okay,
that's
just
so,
and
then
one
of
the
other
questions
we
had
was
around.
You
know
sitting
global
defaults
versus
you
know,
cluster
levels.
They
are
service,
specific
things,
so
you
know
ideally
there's
a
lot
of
things
that
we
would
like
to
set.
You
know
at
a
global
level
and
not
have
to
redefine
everywhere.
Is
there
any?
You
know,
move
towards
that?
It
seems.
Like
you
know,
kind
of
the
CRT
is
in
general.
It
seemed
like
they
were
kind
of
designed
at
different
times
and
involves
so
some
of
them
have
support
for
that.
E
B
Yes,
we
really
want
to
fix
this
problem.
Like
so
there's
a
there.
A
couple
problems
in
the
system
right,
one
of
them-
is
our
configuration.
Is
it
true
well-drained
like
too
great?
So
it's
like
we
are
talking.
We
are
using
very
small
pieces
of
conflict
that
this
is
like
fit
well
together
right
and
that's
more
of
a
semantic
structural
problem
with
the
context
themselves.
You
know
one
idea
to
fix
that
one
was
to
okay.
B
Why
don't
we
create
a
higher
level
configuration
like
you're,
calling
this
service
comping
and
you
create
one
of
those
things
it
maps
to
a
bunch
of
like
smaller
configurations
underneath
the
covers
it's
like
these,
then
there's
terminating
is
operator
pattern.
You
know
like
a
little
concrete,
you
know,
and
apart
from
that,
we
also
have
this
API
governance
pert
that
tries
to
align
the
APS
closer
together,
trying
to
make
sure
that
we
use
the
same
nomenclature.
You
know
the
features
fits
well
together
so
but
it's
gonna
be
an
ongoing
process.
E
B
Is
really
a
special
at
this
point?
We
really
want
to
get
this
done,
but
it's
always
gets
kind
of
the
prioritized.
We
always
have
a
bigger
fish
to
fry,
but
yeah
I
think
fundamental
at
some
point
behind
me.
We
need
to
get
to
us.
It
is
really
something
that
we
want
to
do
and
I
think
it's
going
to
improve
the
usability
of
the
system
quite
a
bit.
Okay,.
E
B
C
I'd
say
if
you
have
specific
examples
of
something
that's
not
easily
doable
today,
that
would
be
really
great
to
capture
in
other
Tamilian
Minister
getup
issue,
the
tennis,
the
pattern
we
tend
to
follow
is
we
let
things
accumulate,
so
we
under
really
understand
the
problem
before
we
try
to
create
a
solution
for
it.
Okay,.
E
C
Know
that
there's
a
Miss
question
our
Academy
container
a
resource,
so
you
can
match
things
together,
that's
kind
of
like
what
I
was
talking
about,
but
we
don't
know.
We
don't
know
what
that
looks
like.
So
there's
a
lot
of
discussion
on
that,
but
I
think
we're
still
in
the
requirements.
Gathering
phase
okay,
gotcha.
E
Yeah
I
think
that
one
of
the
kind
of
underlying
things
that
you're
coming
up
and
config
stuff
we're
looking
at
is
you
know,
that's
there's,
there's
a
lot
of
detailed
config
stuff
exposed
by
envoy
and
then
you
know,
SEO
is
I.
Guess
it
has
opinions
on
that
and
expose
the
some
of
it,
and
so
where?
Where
do
you
draw
the
line
for
what's
exposed?
And
what's
not,
is
the
intent
to
exposed
every
dial
and
knob
that
that
envoy
provides
or
because
either
there's?
E
Basically
the
things
like
you
know
the
retry
on
policies
right
now,
that's
hard-coded
and
you
know
being
able
to
configure
that
would
be
nice.
You
know
just
basically
like
TCP
keep
a
lot
of
things
like
that.
So
I,
don't
know.
You
know
what
the
strategy
is
long
term,
I
guess
for
yeah,
because
every
new
thing
gets
an
envoy,
there's
yeah
extra
work
that
that
gets
added
in
on
this.
You
aside,
I,
guess
to
to
duplicate
that
yeah.
B
So
I
think
I
may
be
the
very
of
trying
to
put
too
much
emphasis
on
fixing
the
particular
groups
configurations,
mainly
because
they
are
the
experts
in
that
I
know.
Networking
teams
the
expert
on
this
and
they
actually
had
the
they're
talking
to
customers
in
a
trade
might
recite
a
feedback
loop,
so
it's
kind
of
I'm
going
to
defer
to
them
for
these
kinds
of
decisions
right
now.
That
is
the
general
comping
story.
Yes,
I
think
the
desire
would
be
to
have
a
comp
like
common
coherence,
configuration
model,
that's
easy
to
use
right.
So,
okay,.