►
From YouTube: Argo Contributors Office Hours Mar 30th 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
Okay,
hi.
Everyone
welcome
to
the
contributors
meeting
today,
I'm
gonna
be
your
host
I'm,
Leo
and
starting
off
Justin
for
me.
Can
you
see
my
screen.
A
B
Yeah,
so
there
were
36
new
and
still
open
issues
managed
to
close
seven
of
them.
So
you
know
we're
still
trending
towards
infinite
issues.
B
So
this
one
somebody's
they're
saying
that
basically
they
get
in
a
situation
sometimes
where
our
OCD
repo
server
isn't
responsive,
because
it's
doing
too
many
things
and
that
it
responds
with
code,
400
and
they're,
suggesting
that
it
should
respond
with
code
503,
because
it's
expected
to
be
available.
B
It's
just
busy
right
now
and
503
would
have
a
default
retry
to
it.
I
don't
know
I
wanted
to
ask
about
it,
because
I
I
wasn't
sure.
If
that
was
a
good
idea
or
not.
I
was
thinking
that
if
it
was
code
503
and
your
server's
too
busy,
then
everything
keeps
retrying.
So
you
sort
of
DDOS
yourself,
but
also.
B
I
think
he's
getting
it
when
he's
trying
to
apply
a
resource
hang
on.
Let
me
look.
A
Interesting
because
if
it's
invalid
specular,
this
is
a
this
is
a
valid
bad
request,
right
so
for
under
maintenance.
He's.
C
A
B
Because
Argos
CD
server
is
getting
it's
not
able
to
access
repo
server
or
it's
not
yeah,
it's
not
able
to
access
repo
servers.
So
it's
because
it's
going
too
slow
right.
A
D
A
Because
that
would
kind
of
make
sense
to
me
if
that
is
really
the
case,
I
understand
400
as
appropriate
response.
But
if,
if
the
underlying
issue
is
repo
server,
not
not
responding
on
time
or
being
down
or
overloaded,
then
maybe
it's
a
bad
error
handling
on
the
API
server
side
of
things,
so
yeah
I
think
it
deserves
some
some
investigation.
Unless
someone
else
that
has
a
good
hint
for
this.
C
It's
bad
error,
handling,
invalidance.
That
care
is
only
there
because
this
is
the
section
of
code
where
the
API
server
is
trying
to
confirm
that
it's
a
valid
spec,
but
part
of
that
is
making
requests
to
the
repo
server
the
request
times
out
and
we're
not
treating
that
as
a
re-tribal
error.
Even
though
we
probably
should.
B
Think
that's
ultimately,
his
point
is
that
it
should
be
treated
as
a
retryable
error.
I
guess
there's
like
if
it
if
it
fails
to
generate
a
manifest
it's
unclear
if
it's
failing
to
generate
the
Manifest,
because
the
Manifest
is
invalid
and
taking
infinite
time
like
I,
don't
know
if
you
could
configure
a
manifest
to
be
infinitely
generating
you
probably
could,
in
which
case
it
would
just
time
out.
It
would
look
the
same
as
repo
server
not
being
able
to
respond,
because
it's
too
busy
with
other
things.
A
Okay,
so
yeah
it's,
it's
fixing
the
error
handling
most
likely.
A
I
guess
at
this
point
someone
needs
to
fix
the
error
handling
code.
I
think
you
did.
B
Should
we
treat
it
as
a
as
a
bug?
I
I
thought
it
kind
of
wasn't
a
bug.
I
moved
it
to
an
enhancement
request,
it's
not
exactly
a
bug,
but
it
is
maybe
better
error
handling
it's
an
enhancement
right,
yeah
enhancement,.
A
So
I
added
a
comment
just
to
yeah
just
to
have
the
the
yeah
the
the
outcome
of
this
discussion.
So
we
all
understand
this
is
a
improved
fixed
that
needs
to
be
done
in
the
error
handling.
Basically,
yeah
thanks
thanks
Dan.
B
All
right
and
I
think
that's
it
for
issue
triage.
A
Oh
and
panachatis
was
the
secondary,
not
sure
if
you've.
D
A
So
thanks
thanks
a
lot
Dan.
So
let's
elect
the
triage
for
for
next
week,
any
volunteers
for
primary
and
secondary.
E
F
Yes,
this
is
Jonathan.
West
I
can
do
primary
cool.
A
Thanks
so
you're
saying
you
can
be
primary
next
week:
correct
cool
thanks,
Jonathan
thanks
Regina,
all
right.
Let's
move
forward,
we
have
a
few
topics
today,
let's
start
with.
G
Okay,
so
before
I
share,
I
want
to
talk
about
this
feature,
so
in
Intuit
like
we
did,
some
follow
me
home
and-
and
there
was
some
concern
that
when
the
part
fails,
there's
not
enough
information
surface
on
the
UI,
which
user
can
understand
what
the
problem
is
happening.
G
So
this
feature
has
been
contributed
by
science
Hindu.
In
this
we
we
are
trying
to
surface
the
information
as
form
of
container
state.
So
if
the,
if
the
part
is
failed
due
to
Readiness
probe
or
a
livelessness
pro
or
any
other
Pro
failure,
then
would
show
up
the
status
from
the
container
status
field.
D
A
G
Okay,
so
I
try
to
set
up
some
like
failure
so
that
it
shows
up
on
the
screen.
So
when,
when
the
part
is
in
Crash
Pack
Loop
is
it
shows
up
this
information?
The
controller
is,
is
the
name
of
the
container
and
then
it
what's.
The
issue
is
so
basically
it
will
show
the
reason
and
then
it's
in
crashback
Loop.
It's
not
started
and
not
ready.
G
Okay
yeah,
so
this
is
where
I
I
think
is
live.
Liveliness
probe
is
failing,
so
the
container
comes
up
and
then
it
get
restarted
so
right
now
it
show
it's
restarted
where
it
shows
that
the
container
is
running,
started
and
ready
and
then,
when
it
gets
restarted,
I
guess
that's
the
message
you
were
just
seeing
before
so
I
got
some
of
the
screenshot
here.
G
So
the
first
one
is
I.
Guess
it's
the
liveliness
profiler
yeah
SEC
this
one,
the
one
which
I'm
highlighting
is,
if
you
read
it
says
container,
is
running.
It
started
and
not
ready.
So
that
means
the
container
came
up.
It
started,
but
not
ready
and
it's
due
to
the
Readiness
failure.
So.
G
Yeah
and
then,
if
there
is
an
issue
with
the
image
pull,
then
it
would
show
up
like
this
yeah,
so
I
idea
just
to
Bubble
Up
the
information
on
the
UI
so
so,
like
the
non-kubernet
user,
can
easily
understand
what's
happening
with
their
containers
and
and
for
some
of
the
application.
There
are
long
list
of
containers,
so
it's
very
hard
for
them
to
go
and
look
into
the
Container
status
and
figure
it
out.
G
So
in
this
it
they
will
at
least
see
all
the
list
of
container
and
what
this
current
status
of
the
container.
H
A
Yeah
I
think
and
ticket
that
was
linked.
There
is
a
screenshot
when
there
are
multiple
containers.
Maybe
I
can.
G
Yeah
I
think
right
now
we
are
not
showing
the
init
container.
It's
only
the
like
the
application
containers
are
showing
up.
A
So
that's
good
any
reason
why
not
adding
any
containers
to
the
list.
G
A
Yeah,
because
the
unique
container
needs
to
exit
right,
if
I'm,
not
mistaken,
I'm
not
sure
but
yeah
it'll
be
great
to
have
a
new
container
in
the
list
as
well.
A
Cool
thanks
all
right,
so
let
me.
A
G
A
Okay,
let's
get
back
to
the
next
topic:
improved
editor,
Michael.
C
Yeah
so
there's
a
PR
up
to
basically
add
kubernetes
Auto
completion
and
like
little
squiggly
lines
under
things
that
are
errors
in
our
manifest
editor.
I
think
it'd
be
a
really
slick
feature.
The
only
things
I
guess
I
need
now
are
for
folks
to
try
it
out
locally
I
need
to
try
it
out
locally
as
well
and
I
identify
if
there
are
any
pain
points,
front-end
performance
issues,
ETC
the
other
thing
I
want
to
do
is
read
through
the
yarn
Dot
Lock
make
sure
where
all
these
dependencies
are
coming
from.
C
So
we
just
have
a
solid
sense
of
what
we're
running
in
our
front
end
and
who
maintains
those
packages
yeah
mostly
just
wanted
to
draw
people's
attention
to
the
pr.
A
Cool
one
thing:
I
I
remember
when
I
was
talking
with
them
when
they
first
had.
This
idea
is
to
make
sure
that
this
does
not
make
any
outbound
request
outside
of
where
are
those
City
Y
is
running?
Do
I?
Are
you?
Do
you
know
Michael?
If
that
is
the
case,
if
everything
happens
contained
in
the
in
the
JavaScript
layer,
I.
C
Think
it's
all
internal
I
need
to
run
it
locally,
and
just
look
at
the
network
see
what
happens
right
right.
Okay,.
A
Yeah
I
think
that's
that's.
That
would
be
a
important
thing
to
the
boat
check,
but
the
future.
The
future
is
pretty
nice
for
sure
cool.
Any
comments,
suggestions
on
this
one
I
think
it's
I
saw
your
comment
here.
It's
planned
for
two
eight
right,
yeah.
A
Okay,
thanks
Michael
all
right
moving
forward
Regina
you.
F
Have
yeah
this
should
be
quick
so
for
multiple
sources:
it
still
needed
the
CLI
and
it
was
originally
kind
of
targeted
for
2.7,
although
I
don't
see
anything
about
it
in
the
roadmap.
F
But
I
I
wanted
to
work
on
this,
but
we
said
previously
that
we
wanted
to
have
a
proposal
for
it.
First
and
I
can
start
working
on
the
proposal
next
week,
but
I
don't
think
that
it
would
be
ready
for
2.7.
So
is
it
okay
to
assign
this
for
2.8.
A
Sorry,
original
I,
I,
don't
think
I
was
here
when
it
was
suggested
to
have
a
proposal
for
that.
What
was
the
re?
What
is
the
the
reason
for
the
proposal.
F
It
it
was
mentioned
in
the
actual
proposal
for
multiple
sources.
Okay,
it
stated
in
that
proposal
that
we
would
need
a
additional
proposal
for
CLI
and
a
separate
one
for
UI
as
well,
but
I'm
just
doing
the
CLI
part.
D
All
right,
Leo
I,
can
start
there
during
the
proposal
approval
for
the
multiple
sources
we
were
just
making
the
changes
there
and
for
CLI
and
UI
I
think
Alex
confirmed
that
we'll
require
a
different
proposal
and
we
would
work
on
that
later
on.
Okay,
once
things
just
get
a
little
steep
yeah.
Thank
you.
F
And
since
it's
not
in
rc1-
and
it
said
everything
has
to
be
in
by
April
10th
I
believe
for
Milestone
2.7,
so
I
just
want
to
make
sure
it
was
okay
and
acknowledge
that
it
would
be
moved
to
2.8.
F
Sorry
about
that
no
2.8
Milestone
is
a
is
it
I
just.
D
F
A
A
Oh
and
then
you'll
probably
bring
to
to
this
to
this
meeting
in
the
future.
Yeah
thanks
thanks,
Regina
anything
else
from
this
one
nope.
A
Thank
you,
okay.
So
moving
forward
Blake
scuffle
demo.
Do
you
want
to
share
the
screen?
You
want
me
to
Sure?
Yes,.
H
H
Right
so
I
promised
a
few
weeks
back.
That
I
would
do
a
demo
since
John
and
Michael
wanted
to
to
show
very
ad
hoc,
so
don't
be
for
one.
That
is
not
an
amazing.
It's
not
going
to
be
a
mind-blowing
demo,
but
at
least
to
show
what
it's
about
so
so
the
gist
of
it
is.
H
You
can
just
send
scaffold
here.
It's
basically,
you
can
think
of
it.
As
the
Gorman
workflow,
with
the
different
stats.
It
will
deploy
directly
into
a
kubernetes
cluster,
whether
it
can
be
remotely
or
it
can
be.
A
local
in
this
case,
I
have
a
Docker
for
desktop
cluster
and
basically,
when
incremental
changes
happens,
it
will
rebuild
the
docker
container
and
then
redeploy
into
the
given
into
the
local
cluster.
So
I
just
reuse,
the
existing
the
existing
production
Docker
file,
I
added
a
mount
type
here.
So
that
means
that
it
would
reuse
the
go
cache.
H
So
if
there
are
any
changes
happened,
it
will
rebuild
and
basically
what
happens
is
that
we
have.
We
can
just
say
this
with
this
flow,
so
we
have
a
scaffold
Dev
and
what
happens
then
is
that
it
will.
It
will
take
the
way
I've
configured
it
here
is
that
this
will
basically
take
the
so
I
added
a
customize,
a
customization
which
takes
the
existing
well
pretty
much
what
we
have
here.
So
we
can
just
see
here
that
this
will.
This
will
take.
You
know,
let's
add,
to
secrets
that
we
can
log
in.
H
H
It
will
tag
a
hash
and
then
it
will
just
deploy
and,
of
course,
if,
if
we
don't
have,
if
not,
if
it
always
tries
to
pull,
then
it
will
try
the
remote
it
will
try
to
get
it
remotely
and
it
won't
work.
So
this
is
why
we
need
to
do
customizations
to
actually
just
patch
that
we
manipulator,
if
not
present,
but
otherwise,
it's
just
the
same
old,
pretty
much.
The
same
old
deployments
that
we
have
already
and
what
I
find
to
be
kind
of
useful
is
that
you
can
really
make
scaffolds.
H
So
the
only
real
difference
is
just
we
just
check
that
it
creates
a
namespace
I.
Also
added
in
this
Branch
I
also
added
a
a
scaffolds
configuration
for
the
repo
server
or
another
for
the
e2e
server.
So
what
that
means
is
that
we
can
okay,
we'll
let
it
do
its
thing,
of
course,
with
this
wait
a
bit
now,
I
will
deploy
everything,
and
then
we
just
give
it
a
give
it
a
few.
H
The
main
appeal
for
me
is
that
this
makes
running
the
way
I've
configured.
It
is
that
the
end-to-end
test
becomes
quite
easy
to
run.
This
is
kind
of
where
my
frustration
was
built.
It
was
starting
with
fellow
scaffolds,
where
you
know
doing
the
whole
Docker
and
Docker
thing
with
the
interim
test,
made
it
very
slow
to
run
at
least
on
Mac
I
mean
I'm
guessing
the
Linux
guys
have
a
bit
of
an
easier
time,
but
we
can
just
take
any
oh
God,
damn
it
someone's
ringing
the
doorbell,
but
I
will.
H
D
H
Relevant
environment
variables
to
be
able
to
run
to
the
internal,
to
run
down
to
end
tests,
I
just
replaced
the
locals
with
ostocker
internal
so
that
the
repo
server
can
do
its
thing.
But
what
this
means
is
that's
from
the
comfort
of
the
idea.
I
can
just
do
run
test
grade
employment,
I,
don't
know
if
this
one
actually
worked,
I'm
just
taking
a
random
end-to-end
test,
but
I've
done
this
with
mother
and
twin
tests
and
it
works
another
sec
if
I
have
to
do
anything,
no
okay.
So
far,
so
good
anyways.
H
I'm
sorry,
the
gas
guy
is
coming,
but
okay,
let's
go
piling
split
slope,
but
I
can
show
in
the
meantime,
I
can
also
show
that
we
also
have
the
UI
we
can
log
on
admin
and
we
can
lose
their
typical
password.
H
Their
full
Argo
server
cabinet
here
is
that
we
don't
have
the
whole
incrementary
compilation
with
when
it
comes
to
JavaScript.
Just
for
the
reason
that
there
it
is
that's
going
on,
we
don't
have
that
I
mean
if
any
any
changes
to
any
typescript
will,
of
course,
lead
to
that
a
whole
Docker
image
get
the
docker
image
gets
rebuilt
and
whatnot,
but
at
least
when
it
comes
to
low-lying
it,
life
becomes
a
bit
simpler.
H
But
I
think
I
got
the
gist
of
it.
Oh
I
think
it
got
the
gist
of
it.
I
don't
know.
If
there's
anything
okay,
this
test
failed.
You
dismissed
I'm,
not
sure
why
I
don't
know
what
this
I
don't
know.
What
else
you
guys
would
like
to
say.
That's
I'm
happy
to
give
more
of
a
overview
if
there
any
questions
and
whatnot
I'll
stop
sharing.
Unless,
if
anyone
has
anything
to
add.
A
Cool
thanks,
Blake
I
guess
one
question
from
my
side
is
okay,
so
this
is
mainly
for
development
purposes.
Right.
Yes,
yes,
and
if
you're,
if
you're,
using
Docker
approach
for
deploying
the
Argo
City
components
locally,
and
you
want
that
to
get
automatically
automatically
reload
and
whenever
you
you
change
something
in
one
of
the
one
of
the
components
that
rocd
uses,
for
example,
repo
server.
Then
it
gets
automatically
refreshed.
You
don't
have
to
redeploy
the
whole
Docker
image.
A
It's
a
hot
deploy,
kind
of
thing,
right,
correct,
correct
cool
cool
yeah
I,
usually
when
I
run
into
tests
locally,
I,
don't
use
the
docker
approach.
I
just
deploy.
I
just
start
the
process
directly
using
the
the
Gorman
approach
that
we
have,
which
is
yeah
I
understand
when
you
say,
is
very
slow.
If
we
go
with
the
docker
approach,
that's
why
I
usually
go
with
Gorman
one
and
that's
much
much
faster
I.
Don't
think
we
have
a
super
great
documentation
on
how
to
do
that,
and
that's
I
guess
my
point.
A
I
guess
this
whole
this
whole
documentation
and
procedure
on
how
to
run
Argo
City
locally
is
a
little
bit
messed
up
right
now,
in
my
opinion,
I'm,
not
sure
if
everybody
agrees
but
yeah
I
think
documentation
is
not
great
I
like
the
approach,
but
I
also
think
we
should
do
some
Cleanup
in
the
dock.
A
So
if
we
want
to
go
that
route,
that
will
be
the
official
supported
way
of
dealing
with
deploying
locally
using
the
docker
containers
and
not
having
like
three
different
methods
of
doing
so,
because
I
know
there
are
in
in
the
make
make
file.
There
are
some
make
targets
that
attempt
to
do
the
same
so
I
guess
what
I'm
yeah?
What
I'm
trying
to
say
is.
It
will
be
great
to
have
one
official
way
to
to
to
to
run
tests
locally
properly
documented,
basically,.
H
Yeah
that
makes
sense
I
just
want
to
point
out
that
there's
some.
There
are
some
things
which
I
do
agree,
that
the
worm
approach
works
as
well,
but
there
are
some
things
which
I
think
aren't.
H
This
is
something
which
I
guess
we
need
to
discuss
as
well.
It's
like
because
there's
some
things
which
cannot
really
be
tested
without
having
like
the
full
kubernetes
environment,
because
I
mean
there
was
some
issue
that
we
had
with
with
the
certificate
reloader
for
example,
which
which
needed
a
full
like
you
know
how.
How
does
how
did
the?
How
how
Argo
CD
application
control?
How?
How
would
that
work
to
test
the
whole
entrance
so
that
that's
that's
related
but
topic
for
another
for
another
day
for
sure,
but
I.
A
I
never
work
with
scaffold
before
but,
for
example,
if
you
have
a
change,
a
coding
in
repo
server,
for
example,
what
is
the?
What
is
the
procedure?
What
does
the
workflow
look
like
you
save
the
file
but
and
then,
once
you
save,
the
scaffold
will
automatically
recompile
and
update
the
docker
image.
Is
that
that's
the
way?
That's
right?
That's.
H
I
I
Interesting
I
last
week,
I
heard
about
octeto
I
think
it
does
something
similar,
also
a
hard
reloading
in
a
container
did
you
did
you
consider
this.
H
I
do
not
but
I
haven't
I.
Could
you
link
that
I'm
not
sure
what
that
is,
I
mean
the
reason
why
I
went
myself
as
I
worked
in
the
past
long
long
time
ago,
and
then
I
remembered
it
while
working
with
this
but
I'm
happy
to
look
at
some
other.
I
A
A
What
would
be
the
the
the
benefit
of
using
octet
to
be
compared
to
what
Blake
is
is
showing.
I
I
am
not
sure,
because
I
didn't
get
far
with
playing
with
it.
It's
just
that
and
they
I
think
they.
They
are
both
about
heart,
reloading
and
stuff.
In
the
container.
A
I
guess
my
point
is
that
we
already
have
a
POC
really
executed
by
by
Blake,
and
it
looks
great
Blake
has
experience
in
it.
A
If
we
want
to
look
into
a
different
tool,
I
guess
the
starting
point
I
would
suggest
is
to
have
a
clear
understanding
of
what
is
a
killer
feature
that
would
justify
us
spending
time
on
doing
the
same
you'll
see
in
a
different
in
a
different
tool.
That's
the
the
way.
I
see
it
at
this
point.
H
Yeah,
that's
going
back!
You
can
see
that
just
going
back
to
say
that
indeed,
so
here,
basically
rebuilds
and
it
caches
all
the
layers.
Apart
from
the
part
where
yeah
pretty
much.
H
H
Not
that
I
can
think
of
at
the
top
of
my
head.
I
mean
the
pain
point
I
would
say
is
like
there's
still
not
a
great
solution
for
yeah.
If
we
work
on
front
and
stuff,
it
would
be
a
kind
of
a
soccer
experience,
because
then
that
implies
the
whole
rebuild
I
mean.
Maybe
we
can
do
some
split
and
you
know
have
a
that.
H
That's
something
that's
to
do
and
to
be
honest,
as
for
other
pain
points,
one
which
is
slightly
annoying,
which
I
seems
like
you
know,
when
you
do
a
redeploy,
then
the
the
tailoring
the
logs
sometimes
is
a
bit
if
they
now
it
seems
like
it's.
Okay,
but
I've
noticed
it
in
the
past.
H
Where
it's
you
know
you
rebuild,
and
then
it
redeploys
I
mean
it's
everything's
up
and
running,
but
then
the
tail
and
the
logs,
for
whatever
reason,
doesn't
really
they
work
as
one
expects,
but
on
top
of
my
head,
I
cannot
think
of
any
I
mean
I'm
sure
there
are
I
mean
I,
mean
I'm,
sure,
there's
using
a
lot
more
anger.
Now
I
would
I
mean
I've
been
I've,
been
doing
this
like
when
working
on
things
on
the
sides
like
the
latest
PRS
and
such
and
I've
at
least
been
using
it
anger.
H
It's
been
pretty
okay
in
that
sense,
but
I'm
sure
there's
room
for
improvement,
trying
to
think
if
there's
anything
else
that
I
can't
think
of
the
top
of
my
head.
No
I'm
kind
of.
D
A
I,
don't
know
about
the
others,
but
I
like
the
the
demo
I
think
it's.
It
would
be
to
provide
some
great
flexibility
and
speed
up
while
running
et
tests
locally.
A
This
could
be
even
faster
than
the
Gorman
approach,
because
Gorman
requires
us
to
rebuild
and
redeploy
is
just
that
the
process
fast,
so
you
don't
mind
doing
it
over
and
over
again,
but
if
this
is
the,
if
this
does
some
sort
of
a
hot
kind
of
deploy
that
will
be
great
to
have,
in
my
opinion,.
A
Regina
you
mentioned
about
octeto.
Would
you
be
willing
to
do
some
sort
of
comparison
between
the
two
tools
just
to
make
sure
if,
if
you're
not
missing
a
killer
feature
that
octeto
provides
and
scaffold
doesn't.
I
Yeah
I
would
love
to
do
that
and
I
might
need
help
with
are
trying
to
to
adopt
to
to
integrate
it
with
Argos.
It
is
build
and
deployment
process
because
I'm
still
very
new,
with
this.
A
I
A
Would
you
be,
and
would
you
be
willing
to
to
take
a
look
at
scaffold
to
see
how
it
looks
like
for
you.
I
Yeah
sure
sure,
actually
these
demos
look
yeah.
It
looks
great
I
think
it
will
speed
up
things
intensive,
like
significantly.
D
A
I
think
we
have
one
more.
What
time
is
it?
We
have
seven
minutes
and
we
have
one
more.
It's
from
you
right,
Regina,
re-implementing,
custom
actions.
D
D
A
A
Yeah,
if
you
can
just
provide
us
an
overview
of
why
you're
bringing
this
up
and
yeah,
if
you,
if
you
need
some
input
from
us
or
anything.
I
Yes,
so
I
used
I
started
working
on
the
feature
that
basically
enables
creating
like
I'm,
creating
ad
hoc
Resources
with
Argo
CD
by
using
custom
action.
I
This
there
was
a
requirement
from
users
to
be
able
to
create
jobs
like
in
an
ad
hocway
from
chrome
jobs,
and
so
this
led
to
basically
to
a
design
in
which
custom
actions
can
generally
enable
creating
new
resources
instead
of
just
patching
the
existing
ones
and
when
I
started
working
on
that
I,
there
was
some
things
that
I
realized.
First,
the
first
was
that,
basically,
if
we
are
going
to
enable
creating
new
resources,
then
this
would
result
in
longer
lower
code.
I
For
the
first
thing,
and
since
at
the
users,
this
code
resides
in
the
config
map
it,
it
can
make
the
config
map
long
and
maybe
more
error
prone,
and
this
is
the
central
config
map
of
Fargo
City.
I
So
my
first
thought
was
that
maybe
the
custom
actions
like
can
maybe
be
like
in
a
separate
config
Maps
or
something
so
that's
they're,
not
in
the
central
critical
config
map.
That
was
like
the
first
thing
that
I
thought
about
and
if
speaking
more
generally
then
it
looks
like
Argo
CD
is,
is
integrating
tightly
with
kubernetes
ecosystem
and
with
customize
and
help,
and
it
seems
that
Lua
is
kind
of
out
of
this
ecosystem
and
and
also
if
we
enable
creating
new
resources.
I
Then
this
might
lead
to
some
kind
of
abuse,
because
now
resources
can
be
created
without
actually
being
in
git,
and
so
I
started
thinking
about
ways
that
maybe
those
ad
hoc
resources
can
be
present
in
git
and
maybe
closer
to
the
kubernetes
approach
to
things.
And
so
one
of
the
options
is
maybe
having
them
as
a
custom
resource
of
kubernetes
those
custom
actions
and
instead
of
Fluor
code,
they
may
be.
Some
kind
of
templates
of
kubernetes
manifests
and
they
can
reside
in
git
and
then
invoking
a
custom.
I
Action
can
resemble
some
kind
of
think,
not
an
exact
sync
but
sort
of,
and
so
this
way
maybe
the
custom
actions
can
get
closer
to
get
UPS
approach,
while
still
enabling
to
be
invoked
in
an
ad
hoc
manner
and
the
other
approach
that
I
thought
about
was
for
the
custom
actions
to
be
some
kind
of
and
extension
of
the
existing
desired
state
that
the
user
already
has
either
in
his
git
repo
or
his
Helm
charts
or
or
whatever.
I
A
Yeah
we
have
two
minutes,
and
then
so
this
is
just
just
from
my
understanding.
This
is
a
proposal
that
you're
submitting
I
think
by
the
by
what
you're
describing
I
I
would
I
would
suggest
you
to
have
a
proposal
in
the
form
of
a
PR
proposal
where
we
can
comment
on
things
because
I,
don't
I
think
this
is
a
big
big.
D
A
Are
many
many
things
that
you
described
I'd
like
to
understand
more
how
you're
planning
to
solve
the
problem?
So
I,
don't
know
I,
don't
know
if
you're
familiar
Regina,
but
in
Ardo
City
we
have
a
folder
in
the
docs
called
proposals.
A
And
then
yeah
we
have,
we
have
a
few
of
them,
so
I
provided
a
few
proposals
in
the
past,
so
props
extension
is.
It
was
the
last
one
and
then
we
have
here
a
proposal,
formal
proposal
and
once
the
once
it
gets
merged.
He
means
the
proposal
was
accepted
and
it
was
properly
discussed
in
the
in
the
in
this
group
asynchronously.
So
that's
why
PR
it's
preferable
where
we
can
just
add
comments
and
follow
up
with
the
with
the
discussion
there.
A
D
A
Because
then
we
can
go
and
comment
straight
on
specifically
sentences
and
paragraphs.
A
And
that
would
allow
us
to
to
move
forward
asynchronous
in
more
efficient
way,
I
think
on
the
lower
part,
I
yeah
I
understand
more
or
less
what
you
said,
because
the
the
idea
of
having
Lua
adapt
at
that
specific
part
of
Arlo
city
is
because
Lua
is
is
a
very
easy
to
language
to
embed.
So
we
embedded
the
language
in
Argo
CD
and
then
it's
very
easy
to
write
Lewis
scripts
to
verify
if
a
specific
field
in
a
yaml
is
following
some
some
patterns.
A
So
it's
very
easy
to
to
implement
some
logic
and
Lua
is
a
is
a
great
feat
for
that
in
a
in
and
it
provides
a
very
low
footprint.
So
you
it
it's.
It's
not
a
heavy
language
to
run,
and
it's
very
very
easy
to
to
embed.
So
that's
the
main
reason
why
Lua
is
there
from
what
you're
describing
I
understand
more
or
less?
Why
you
want
to
have
something
like
customize
and
health,
but
yeah
I
think
it's
all
part
of
the
bigger
discussion.
A
Okay,
thanks
thanks
Regina
and
thanks
everybody
for
participating
and
see
you
next
week.