►
From YouTube: Argo Contributors Office Hours April 27th 2023
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).
A
Hello,
everyone
welcome
to
the
contributor
meeting.
I'm
gonna,
be
your
host
today,
I'm
Leo,
and
usually
we
start
always
with
three
action
discussion.
For
this
week
we
had
Ishita
in
Remington
for
doing
triage,
so
I
see
you
online
so.
B
B
There
was
only
one
enhancement
which
I
found
was
different
was
regarding
the
bot
summary
so
right
now
the
user
was
requesting
if
we
could
have
a
part
summary
in
the
UI
when
it
restarts
so
I'm,
not
sure
if
everyone
would
like
it,
but
I
found
it
would
be
very
cool
for
the
users
to
get
all
the
information
on
the
UI.
Regarding
this.
B
Is
okay,
yeah
I
just
wanted
to
see
if
we
are
thinking
to
add
this
in
our
roadmap
for
the
future
releases,
so
yeah
there's
not
much
I
think
even
for
this.
We
either
would
need
some
purpose
or
proposal
or
something
but
yeah.
B
D
A
A
D
C
A
A
Okay,
let
me
comment
this.
Thank
you,
so
we
can
yeah
keep
track
on
this
one
cool
thanks
thanks
such
thanks.
Everyone
all.
A
Okay,
all
right
thanks,
Ishita
I,
don't
see
Remington
here
so
I
guess
we
can
move
forward.
We
have
quite
a
few
topics
today.
I
am
not
sure
if
we're
gonna
have
time
to
cover
all
that.
So
if
there's
something
a
name
that
we
are
unable
to
cover
this
week,
so
we're
gonna
move
to
to
next
week.
Let's
see,
let's
see
how
that
goes.
So,
let's
try
to
do
it.
Sorry
I
was
almost
forgetting.
A
We
need
to
come
up
with
a
list
of
people
to
do
triage
and
discussion
next
week.
So
any
volunteers.
A
Okay,
so
Ash
stash
and
I
can
take
primary
as
it's
been
some
time.
I
didn't
do
it,
so
you
can
do
it
together.
Thanks
hush
Tosh.
You
are
you
into
this.
Yes,
sir
cool
all
right.
A
Okay,
first
topic:
we
have
for
the
day
is
discussion
brought
by
Lucas.
Are
you
in
the
call.
E
So
hi
cool
to
to
have
the
opportunity
to
to
present
the
question
here
so
I'm
working
for
a
devops
team
at
get
your
guide
and
there
we
are
using
the
Argo
CD
for
for
Wilder,
and
now
we
would
like
to
be
sure
that
our
Engineers
have
a
really
good
service
quality
and
we
would
like
to
also
be
informed
about
the
service
service
problems
before
it
will
cause
any
serious
issues.
E
So
because
it's
like
a
car,
a
core
part
of
our
infrastructure
right
I
mean
the
the
Argo
CD
service,
and
this
is
how
we
end
up
with
some
SLO
definition
and
we
are
using
data
docs.
So
we
just
based
on
the
Argo
CD
metrics.
So
for
all
the
services
there.
E
We
looked
on
the
on
those
which
are
important
to
us,
which
could
indicate
some
problems,
but
we
are
still
like
trying
to
find
at
the
threshold
and
try
to
find
the
right,
the
right
metrics,
which
we
should
monitor
and
what
values
they
should
have,
and
that's
why
I
was
also
searching
for
some
guidance
in
Argosy
and
I
didn't
found.
Once
now.
The
I
have
two
questions.
E
The
first
one
is,
if
you
have
any
SLO
definition
which
you
can
share,
I
would
really
appreciate
that,
and
the
second
is
how
I
can
contribute
to
the
project
with
my
solo
definition,
because
I
found
that
you
have
like
a
grafana
dashboard.
So
maybe
like
some
proposal
of
of
grafana
others,
I
I,
think
it's
called
like
that
in
in
grafana
will
be
okay
for
you,
because
yeah
then,
like
other
users,
can
can
also
use
that
and
we'll
just
have
it
right.
A
Yeah,
okay,
so
I
guess
my
my
my
take
on
this
I
I
believe
SLO
is
really
a
specific
Company
by
company,
so
I,
don't
I'm,
not
sure
we
can
kick.
We
can
come
up
with
something
that
works
for
all
the
companies
we
can
share.
Maybe
what
some
strategies
that
we
apply
for
a
company
at
the
size
of
Intuit
and
certain
things
that
we
do
but
again
yeah
coming
up
with
the
standard
that
fits
all
the
all
the
companies
that
use
this
Argo
I
think
it's
I'm,
not
sure.
A
If
we're
gonna
be
yeah.
D
I
think
so,
from
from
from
my
guts,
I
would
say
this:
maybe
a
topic
also
for
the
sick,
scalability.
D
D
We
we
definitely
can't
come
up
with
a
like.
You
said:
Leo
I
totally
agree
with
you
that
that
this
is
probably
a
very
specific
thing
to
set
up
for
each
and
every
company
according
to
their
requirements,
but
I
also
see
the
the
issue
here
that
it's
not
really
clear
what
you
know
what
people
can
look
for
in.
So
what
what?
What
exactly
the
metrics
are?
You
know
how
how
you
can
monitor
them
for
certain
things
to
happen
in
your
in
your
iocd
when
when
so,
what
what
can
you
look
for?
D
That
indicates
that
you
may
have
a
problem
or
that
you
may
run
into
a
problem
soon
right,
but
I.
I
also
think
that
that
this
is
something
a
task
that
six
scale
ability
could
take
on
right,
like
yeah.
D
F
I
I
agree
100,
so
I
posted
the
the
channel
in
slack
for
Argo
6
scalability.
There
was
a
presentation
in
there
like
two
weeks
ago
that
went
through
a
number
of
different
metrics
that
was
really
interesting
and
talked
about
how
they
changed
with
different
scalability
challenges.
So
I
would
join
that
channel.
F
You
can
check
out
the
presentation
there
and
then
we
have
six
scalability
meetings,
I
think
either
every
other
week,
so
we're
due
for
one
in
two
weeks
there
was
one
that
was
scheduled
for
earlier
this
week
that
got
canceled,
but
so
there'll
be
one
in
like
two
weeks
and
that'd
be
a
great
place
to
join
and
a
lot
of
the
people.
There
are,
you
know
big
heavy
Argo
users
and
be
a
good
place
to
brainstorm.
E
A
Cool
thanks
thanks
Lucas,
thanks
for
putting
this
discussion
up
all
right.
So,
let's
move
on
next
topic.
Alex
prune
outside
of
Argos
for
outside
of
our
auto
sync
Outlook
is
Alex
in
the
call.
A
I,
don't
see
him
in
the
call,
but
I
looked
in
the
in
the
issue
and
today
we
have
prune
as
part
of
the
automated
section
in
the
application
resource,
and
there
is
a
suggestion
to
move
prune
outside,
which
is
a
breaking
change,
so
I
think
I'll
I'll
just
reply
in
the
in
the
issue
itself.
Maybe
a
suggestion
to
avoid
the
breaking
change.
A
Okay,
so
Alex
is
not
here.
I.
D
D
It
is,
and
I
mean
for
each
of
the
manual
sync
you
just
specified
manually
right.
You
can
I
think
it's
enable
if
you
do
it
in
the
UI,
it's
enabled
by
default.
D
A
D
A
A
Small
usability
Improvement
is
I,
guess
my
biggest
problem
with
that
is
the
breaking
change.
That
would
that
would
it
would
yeah.
D
The
the
question
would
be
if
it
could
be
implemented
in
another
way,
because
I
think
the
it's
it's
good
to
have
that
you
know
I
mean
it's.
Ultimately,
this
is
an
auto
thing.
Configuration
toggle
and
maybe
there
could
be
another
set
instead
of
moving
that
setting
out
of.
G
D
Automated
sync,
maybe
we
could
implement
it
in
another
way,
like
you
know,
like
a
default
Behavior
setting
the
default
behavior
for
manual
sync
prune.
E
Yeah,
so
maybe
I
can
also
add
something
here,
because
I'm
actually
working
together
with
Alex
and
we
look
into
that
issue.
So
he
also
in
being
of
his
comment
hero.
That
is
particularly
the
problem
for
a
rollback.
So
if
we'll
roll
back
and
then
we'll
out
activate
the
auto
sync
we'll
lose
them
the
prune
options.
So
whenever
other,
like
solution,
will
also
cover
that
case,
which
is
like
fine,
but
it
will
be
nice
to
have
it,
because
if
somebody
now
will
rollback
needs
to
also
enable
the
prone.
D
So,
okay,
so
you're
saying
that
if
you,
if
you
roll
back
to
a
previous
revision
and
in
between
these
two
revisions,
you
have
introduced
the
new
resource.
D
So
you
don't
want
the
automated
prune
in
auto
sync
enabled
by
default,
but
in
case
of
a
rollback.
You
want
that
to
be
that
the
resources
are
being
pruned
that
were
introduced
in
between
these
two
revisions.
E
I
mean
it's
not
probably
about
about
the
revenues
itself,
but
normally,
if
you
all,
if
you
are
rolling
back
to
some
older
revision,
you
will
be
asked
if
you
should
disable
the
auto
sync
and
the
people
will
say
right,
and
in
that
moment
you
will
just
send
change
the
the
automate,
the
auto
sync
Behavior.
Now,
if
you
will
then
let's
say
fix
the
problem,
you
will
deploy
the
new
version.
E
A
Seems
to
be
like
a
different,
slightly
different
issue.
The
the
main
one
is
the
usability
issue
and
what
Alex
is
referring
yeah.
D
A
Okay,
so
maybe
what
I'm
gonna
do
I'm
gonna
reply
in
this
ticket?
Just
with
my
concerns
on
whether
or
not
it's
a
good
idea
to
break
this
API
yeah
maybe
have
an
additional
attribute
instead
of
moving
it
outside,
but
having
something
like
Auto
prune
outside
of
the
automated
yeah
I.
Don't
know,
maybe.
D
So
maybe
maybe
maybe
something
I
mean
to
to
address
this
specific
issue
right
this
one
that
that
the
prune
setting
is
lost
when
you
disable
auto
sync
I.
D
D
And
I
I
just
wanted
to
bring
up
the
idea.
So
when
you
you
know
in
the
rollback
scenario,
when
it
asks
you
do,
you
want
to
disable
auto
sync.
So
if
we
say
Yes,
we
could
store
the
previous
auto
sync
settings
in
the
status
right
of
the
application
yeah.
And
if
someone
turns
on
auto
sync
again,
then
we
will
look
in
the
status
field.
If
there
is
some
saved
settings
right
and
just
reapply
them
right
that
that
would
at
least
solve
this
particular
problem.
I
I
think
it
could
solve
it
like.
D
A
Okay,
so
yeah
I'm
gonna,
add
a
comment
here
with
what
we
discussed.
So
let's
move
on
all
right,
so
we
have
another
one
from
JD.
Are
you
in
the
college?
Id?
Yes,
you
are
application.
Controller
is
called
a
b
scalability
issue
open
here.
G
Yeah,
so
this
is
something
that
we're
observing
right
now,
even
with
openshift
githubs
we're
working
with
customers
and
running
into
some
scalability
issues
and
I
came
across
this
issue,
while
I
was
investigating,
which
looks
very
similar
in
terms
of
same
symptoms.
So
basically,
what
we're
observing
here
is
that
at
a
certain
load,
if
even
if,
like
the
application
control,
is
maintaining
a
certain
number
of
namespaces
and
namespace
scope
mode.
G
For
example,
at
some
point,
the
application
controller
gets
stuck
and
kind
of
crashes,
and
you
can
see
that,
because
in
the
logs
we've
noticed
a
lot
of
context,
canceled
error
messages
and
even
the
resource
usage,
the
CPU
usage
memory,
usage,
kind
of
drops
to
zero
and
then
basically
the
apps,
then
just
sit
there.
Unreconciled
and
react
control
has
to
be
restarted
to
get
back,
get
it
going
again.
G
So
these
are
the
same
symptoms
you're,
seeing
with
our
customer
as
well,
and
they
were
saying
that
it's
it's
happening
fairly,
frequently
at
least
once
a
day.
Almost
and
and
then
I
came
across
this
issue,
and
it
seems
like
a
lot
of
people
are
still
facing
this.
G
Even
with
recent
releases
and
I
kind
of
wanted
to
just
bring
this
up
with
the
community
and
see
if
if
this
is
something,
if
people
have
any
ideas
of
where
we
could
start
with
this,
because,
like
at
least
while
I
was
working,
trying
to
help
our
customer
out,
like
we've,
tried
suggesting
workarounds
like
increasing
the
status
processes,
operation
processes
to
a
certain
amount
or
set
repos
over
diamonds
Etc.
G
But
in
the
in
the
with
the
idea
of
trying
to
figure
out
what
the
underlying
issue
is
and
try
to
figure
out
a
long-term
solution
for
it.
I
thought
it'd
be
good
to
bring
it
up
here
and
see
if
people
in
the
community
have
thoughts
about
this.
D
I
I
do
have
some
thoughts
about
this.
Okay,
so
I
for
one
I
think
that
people
may,
in
this
issue
at
least
May
mix
up
things,
but
yeah
so
from
from
what
we've
researched
is
that
that
may
be
a
hidden
deadlock
somewhere
in
the
states,
processors,
so
at
least
two
or
three
of
of
our
customers.
D
They
they
ran
into
situations
where
the
reconciliation
completely
stops
right
and
from
the
locks.
I
could
see
that
that
the
status
processors
have
been
basically
eaten
up
right
they
and
then
there
is
another.
Then
there
is
another
issue
that
JDP
just
brought
up
like
with.
If
you,
if
you
manage
the
more
than
a
couple
of
namespaces,
then
we
we
run
into
a
cash
issue,
but
I
think
it's
it's
really.
It's
two
different
issues.
There.
D
So
and
this
this
issue
is
mixing
it
up,
because
someone
says:
hey:
I
I,
see
these
symptoms
and
someone
else
sees
similar
symptoms
and
says:
hey
I'm
affected
by
this
as
well
so
but
yeah
I.
Think
we
really
need
to
you
know
to
to
decipher
that,
but
yeah
it's
it's
becoming
a
more
and
more
pressuring
issue,
especially
like
the
the
deadlock
situation
and
I
I
think
it
happens.
D
There
is
somewhere
a
race
condition
when
a
cluster
is
being
removed
from
our
city
and
an
application
that
is
targeting
that
cluster
is
deleted
using
with
with
a
resource
finalizer.
So
that
that's
that's
my
the
trail
that
that
we
currently
follow.
D
That
would
trigger
that.
That
look.
A
D
A
To
it,
I
don't
remember,
having
had
this
controller
completely
stop
reconciling
issue
I,
don't
remember,
honestly
having
that.
D
At
an
increasing
Pace
at
large
customer
installations
so,
and
so
the
what
they
have
in
common
is
so
we
have.
We
have
something:
that's
called
Advanced
cluster
management
that
allows
you
to.
You
know.
D
D
Yeah,
probably
probably
because
so
what
what
happens
is
that
there
is
a
an
application
set
involved
in.
A
D
Scenario
with
the
new
cluster
decision
resource
generator,
which
basically
takes
those
ACM
things,
so
there
is
a
thing
called
placement
and
placement
decision
that
figures
out.
You
know
which
clusters
are
managed
by
which
obviously
and
I
I
have
seen
there
is
there.
There
is
a
bug
in
in
that
generator
or
in
in
the
constellation
that
this
generator
is
working
in
like
it.
D
It
removes
a
cluster
and
the
app
so
when
the
application,
so
the
application
set,
doesn't
generate
the
applications
on
a
per
cluster
base,
but
on
on
information
from
ACM,
and
when
this
information
is
removed,
the
application
that
removes
the
application
but
doesn't
remove
the
application,
finalizer
right,
the
resource,
finalizer
and
then
suddenly
thereafter,
the
cluster
is
removed
from
Iowa
City.
A
Yeah
that
makes
sense
I
had
into
it.
We
don't
have
that
removal
of
custers
happening
that
often
so
that
might
be
a
lead
on
yeah.
D
G
B
D
Mean
it
happens
often
that
people
say
hey
I
am
affected
by
this
as
well,
but
it's
a
different
thing
because
yeah,
so
just
it's
just
my
you
know
before
my
my
thoughts,
it's
easy
to
say:
hey
yeah,
I'm,
I'm
I'm
having
this
too,
but
I
have
not
seen
any
evidence
from
those
people
that
that
this
is
the
same
issue.
H
Our
binary
ship
with
race
detection
capability,
I,
don't
know
if
that's
a
thing
that
just
comes
with
the
go
binary
or
if
you
need
external
tooling,
but
if
this
is
a
race,
maybe
just
on
one
of
these
instances
where
it
happens,
every
day
run
the
race
detector.
A
E
D
I
D
Not
I'm
not
even
sure
if
it's
a
if
it's
a
race
condition
in
the
go
code
itself.
Like
a
you,
know,
the
classical
race
condition
that
the
race
detector
would
detect
like
concurrent
access
to
a
map
or
something
like
that.
D
A
D
No
I
I,
don't
know,
I,
think
it's
it's
it's
a
reference
to
Heisenberg
like
I,
don't
know
some.
H
A
Boom
yeah,
yeah
I
think.
The
main
challenge
here
is
really
reproduce
this
in
a
way
that
we
can
identify
the
root
cause
in
this
particular
case
here
I
this
error
message
bothers
me.
A
bit,
it
seems
like
go.
Client
is
being
throttle
right
so
which
might
explain
the
situation
where
even
having
one
single
cluster
synced
by
Ardo
CD
still
Cube
API
is
the
one
having
really
limiting
or
his
son
yeah
the
integration
that
way.
D
Is
something
we
also
currently
look
at
the
sixth
scalability,
because
there
is
the
client
side
limiting
in
the
go
plan
right
yeah,
which
can
be
tuned
using
environment
variables,
pretty
undocumented
feature
but
yeah,
so
the
I
think
they
mentioned
also
self-field
here
right,
maybe
I'm
confusing
things.
Maybe
this
is.
This
is
not
the
same
issue
that
we
see
in
with
our
customers.
G
Yeah
I
mean
specifically
with
our
customers.
The
other
thing
is
also
that
they
have
they're
using
an
app
of
apps
or
like
monolito
setup,
also
so
I
think
they
have
like
one
repo
which
have
like
150
apps
or
something
in
it
and
I
I.
G
Don't
know
if,
if
that's
a
use
case,
we
are
still
supporting
or
not
or
like,
if
the,
if
we,
if
there
is
something
we
we'd
consider
and
when
we
decide
to
pursue
an
optimization
to
try
and
support
it
in
that
case,
because
I'm
sure
that
even
that
kind
of
support
could
affect
the
efficiency
or
performance
of
of
an
instance
as
opposed
to
using
App,
set
or
or
you
know,
just
using
separate
apps
as
opposed
to
a
mono
repo.
G
F
A
A
Again,
adding
to
it
we're
not
facing
this
particular
issue,
we're
facing
different
issues
somehow,
at
the
same
part
of
the
code,
it's
related
to
sending
a
lot
of
high
volume
of
request
to
cube
API.
A
So
this
is
something
we
are
we
are
facing,
so
maybe
anything
we
some
improvements
that
we
do
in
that
front
could
potentially
help
addressing
this
issue
as
well.
Somehow
so,
yeah
I
have
different
mixed
feelings
about
this,
basically
want
to
ask
your
opinion
about
this
about
how
to
properly
address
this.
This
issue.
G
G
I
mean
I
could
also
bring
it
up
in
the
six
scalability
Channel
and
see
if
someone
has
something
to
say
about
this,
but
yeah
I
mean
I
kind
of
get
the
feeling
that
it
can
be
many
different
issues
underlying
all
kind
of
bubbling
to
the
surface.
Looking
the
same
died,
yeah
I,
don't
have
a
great
any
great
suggestions
for
immediate
actions
either,
especially
if,
if
have
if
race
condition,
detection
is
also
not
really
guaranteed
to
help
us
catch
these
things
at
compile
time.
A
I
can
do
that
cool.
Thank
you
thanks
thanks
everyone.
Thank
you
all
right.
So
next
one
Danny
Hawkins
hello.
I
I
Yeah,
basically,
if
you
look
at
the
linked
issue,
it's
been,
some
people
have
been
requesting
it
for
some
time.
It's
just
the
ability
to
substitute
and
vars
I'm
in
the
manifests
or
in
the
values
or
whatever,
and
then
I
needed
to
do
that
as
well
and
I
found
a
kind
of
hacky
way
to
do
it.
But
then
I
thought
you
know,
I'm
a
programmer,
I,
don't
know,
go
but
I'll
give
it
a
go
and
I
won't
go,
there's
a
lot
to
go
in
there
and
and
yeah.
I
So
I
thought
this
could
be
a
good
first
PR,
it's
quite
small.
Actually
it's
great
it's
a
very
small
PR
should
be
and
and
the
questions
I
had
I
had
two
questions
which
were
the
documentation
should
that
is
that
generally
included
as
part
of
the
the
pr
yes.
I
So
there's
a
thing
in
the
documentation
about
feature
States
being
Alpha
Beta
And
I
was
aware
that
this
changed
every
everything
would
run
through
it,
because
it's
the
final
generate
manifest
from
the
repo
server,
so
I
didn't
know
whether
it
should
go
into
alpha
or
a
beta
or
how
that
process
works
in
terms
of
release.
A
A
A
I
think
it
depends
on
whether
or
not
we
want
to
go
forward
with
this
approach.
I
quickly
looked
at
your
PR,
maybe
a
quick
suggestion
I
would
provide,
is
have
something
more.
A
A
Yeah,
this
is
one
thing
that
came
to
my
mind,
and
the
other
thing
is
that
is
the
fact
is
that
it's
a
more
conceptual
thing
and
I,
maybe
would
like
to
discuss
with
everyone
else
to
whether
or
not
maybe
to
discuss
if
this
is
a
good
idea
in
terms
of
because
this,
this
logic
is,
was
added
in
the
generate
manifest
function,
which
is
the
same
function
that
invokes
helm,
generator,
customized,
generator
and
standard
manifest
Manila
manifest
that.
A
A
A
Yeah
yeah,
so
one
thing
that
came
to
my
mind
is
is
mainly
the
question
to
whether
or
not
this
is
the
best
place
to
add,
because
we're
mainly
changing
how
so
this
is
not
a
feature
that
is
provided
by
customize
or
Helm
right.
So
we
are
adding
this
inside
Argo
City.
So
it's
really
specific
to
Ardo
City
itself,
I,
mainly
questioning
myself.
If
this
is
the
best
place
to
add
this
specific
logic
or
maybe
having
a
dedicated
CNP,
that
does
that.
I
We
wanted
to
use
CMP
and
we
did
CMP
for
quite
a
while.
The
biggest
problem
with
that
was
that
the
Fantastic
addition
of
multiple
sources,
it
doesn't
work
with
cmps,
it's
not
so.
You
can't
use
multiple
sources
like
for
the
helmet
for
the
values
inside
cmps,
and
so
that's
why
again,
I
put
it
at
the
very
final
stage
of
generating
manifest,
so
that,
basically,
everything
that
generates
a
manifest
can
benefit
from
it.
Yeah.
D
I
I
I
think
that
decision
is
logical.
So,
from
my
point
of
view
it
sounds
logical,
but
it
opens
up
a
broader
discussion
and,
in
my
point
of
view,
because
we
in
the
past,
we
have
talked
about
something
like
manifest
post
processor
right.
Basically,
this
is
a
manifest
post,
processor
so
and
a
I
think
the
general
consensus
back
then
was
no.
We
don't
want
to
have
manifest
post-processing,
but
I'm
I'm
not
opposed
at
all
at
re-evaluating
this
discussion
or
well
decision.
I.
D
Don't
think
we
made
a
decision
back
then,
but
I
think
it
yeah.
The
general
consensus
was
like
no.
We.
We
feel
that
this
is
not
our
space
of
doing
stuff
like
like
this
right,
but
to
be
fair
in
in
the
past.
That
discussion
was
also
around,
like
you
know,
having
a
post-processor
like
Cube
Linda,
or
something
like
that
that
would
affect
the
Manifest
generation
like
it
would
abort
if
the
linta
would
throw
some
warnings
or
errors
about
the
Manifest
so
but
yeah.
D
This
is
I
think
this
is
something
we
should
make
up
online
on
I
personally
I
like
this.
This
idea,
and
just
as
a
a
small
review
comment
on
the
Fly,
we
do
have
a
n
subst
utility
in
our
city,
like
a
library
right
that
that
is
a
little
bit
more
I.
D
Like
the
the
word
sophisticated,
then
the
symbols
brings
replays
because
the
the
strings
replace
no,
it's
it's
in
the
replays
and
Forest
and
manifest
like
you
know,
just
simply
replacing
a
string
is,
is
usually
not
sufficient.
So
that's,
maybe
maybe
you
could
take
a
look
at
the
I
think
it's
in
in
util
and
subs,
whether
you
could
leverage
that,
because
it
takes
in
a
lot
of
Education
considerations
which
goes
beyond
the
simple
strings
replaced.
So
that's.
H
Great
there's
a
there's,
an
open
PR
that
I
think
can
solve
this
in
two
different
ways:
it's
by
Greg
I'm,
forgetting
his
last
name,
but
basically
it
adds
the
ability
to
reference
arbitrary
files
from
other
sources
in
a
multi-source
app.
So
that
would
solve
your
first
problem
because
you
could
reference
an
external
values
file
or
whatever,
really
from
your
CMP
Source.
The
second
way
it
would
solve
it
is,
it
would
unblock
chained,
chained,
plugins
and
I.
H
I
I
H
Similar,
that's
that
plug-in
would
enable
this
I
mean
if
you're,
using
a
plug-in.
You
can
do
anything
you
want
you
can
reference
sources
from.
You
know,
chat
GPT!
If
you
want
this
plug,
or
this
PR
would
just
enable
like
pulling
information
from
another
source
and
taking
any
file.
That's
in
the
other
source.
A
One
of
one.
B
Pr
is
that
the
one
you're
talking
about
yeah
Michael?
This
was
the
same
here
which
might
do
is
addressing.
C
H
I
think
this,
this
is
a
far
more
generic
solution
which
could
solve
a
lot
more
use
cases.
One
thing
that
concerns
me
about
arbitrary
invar
substitution
is,
if
you
do
it
in
a
text
fashion,
it
means
that
an
invar
can
write
arbitrary,
yaml
I'm,
not
sure
we
want
to
build
a
tool
that
does
that.
That
may
be
a
little
bit
too
open-ended.
Unless
someone
does
it
in
a
plug-in,
in
which
case
they
can
do
what
they
want.
A
Yeah
I,
agree,
I
think
it's
a
really
good
use
case
for
plugins
and
what
you
described
as
having
chain
plugins
sounds
like
a
simple
uh-ish
way
to
address
this
Michael.
A
A
Cool
okay:
well,
thanks
a
lot
yeah
and
if
you
have
any
any
problems
issues
with
that,
let
us
know
we
can.
We
can
support
you
on
that
too.
Perfect.
B
B
B
Just
regarding
the
proposal,
which
I've
added
I
think
we
discussed
yeah,
we
discussed
briefly
last
time
too,
regarding
this
one
I
just
wanted
a
few
eyes
on
this.
This
is
just
this
proposal
just
gives
two
concerns:
that's
changing
the
changing
the
application
for
application
controller
from
stateful
set
to
deployments
and
dynamically
redistributing
the
Clusters
across
shards.
So
if
it's
possible,
it
would
be
great
to
get
some
reviews
on
this.
One.
H
A
Yeah,
unfortunately,
we
ran
out
of
time.
I'll
have
to
move
the
remaining
topics
for
next
week.
Discussion:
yeah,
sorry!
For
that,
any
last
minute
we
have
we're
already
over,
but
any
less
less
minute.
Super
important
topic,
someone
on
Spring,
if
not
yeah,
I'm
gonna,
move
those
things
to
to
the
next
to
next
meeting
and
see
you
next
Thursday
thanks
everyone,
foreign.