►
From YouTube: Argo Contributor Experience Office Hour 7th Jan 2020
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
B
B
D
That
the
decision
was
made
that
we
migrated
away
from
from
docker
hub
to
choi.
I
think
shubik
has
created
us
an
initial
organization.
Inquire.
Thank
you
for
that.
I
have
just
started
today
to
to
migrate
the
existing
content
from
docker
hub
to
quay,
and
now
the
question
is
how
to
move
forward.
So
I
was
thinking
that
we
can
use
the
the
1.9
release
of
agua
cd
to
officially
migrate
the
the
images
as
in
the
manifests
so
that
we
skip
1.8
the
next
patch
releases.
D
We
let
them
on
on
docker
hub
and
move
forward
with
the
1.9
release.
So
this
is
what
I
was
thinking.
Maybe
there
are
other
opinions
on
this,
so
I
I.
D
Yeah,
okay,
so
I
was
always
okay,
so
I
was
proposing
to
to
go
forward
with
v
1.9
so
that
we
use
the
v
1.9
release
to
also
have
our
installation
manifests
point
to
quay
and
mention
it
in
the
release.
Notes.
B
Mostly
I
I
I
was,
I
feel
like
it's
a
good
idea.
I
mean
I
don't
see
any
problem
with
changing
default,
manifests
and
pointing
it
to
play,
and
I
wasn't
sure
about.
Are
we
going
to
keep
publishing
to
docker
hub
as
a
kind
of
as
accuracy
or
we
just
abandoned
docker
hub
completely
and
we
move
to
quay
and
don't
ever
publish
to
doc.
C
C
It
is
because
of
rate
limiting
issues,
so
it
might
be
good
to
completely
move
away
just
to
avoid
somebody
thinking
that
you
know
things
would
just
continue
working
because
it
may
not
so
yeah,
and
I'm
also
wondering
that
are
there
I
mean
I
I
mean
I'm
sure
there
are
folks
who
might
be
complaining
right
now
that
hey
we're
trying
to
pull
the
image
from
upstream.
C
It's
not
working
because,
let's
say
they
have
a
ci
job
which
pulls
it
multiple
times
a
day
and
they're
running
out
of
you
know
the
number
of
times
they
can
hit
it
are,
there
is,
do
we
know
if
users
and
based
on
the
user
feedback
is?
Is
there
an
urgency
to
even
move
the
existing
images?
I
mean
of
the
versions
that
have
been
released.
Let's
say
1.8
to
quit.
C
D
C
D
B
E
Yeah,
I
I
think,
janae,
I
think
long
term.
We
only
want
one
yeah.
I
don't
think
we
should
keep
going
to
both.
I
I
think
transition
release
that
that
sounds
like
a
good
idea.
B
A
A
Of
course,
most
of
the
references
are
from
you
know
in
our
in
from
our
gopros,
but
like
alex
is
saying
there
may
be
some
people
who
are
like
manually
like
referring
to
it
from
other
scripts
or
configurations,
or
they
you
know
when
they
upgrade
they
do
it
by
bumping
up
the
version
number
or
something
like
that.
B
E
E
A
B
How
do
we
communicate?
How
do
we
are
we're
going
to
just
it's
not
channel
and.
A
E
Yeah
because
the
the
executor
image
is
pulled
with
every
workflow,
I.
A
Always
alex
so
we
need
to
coordinate
with
alex
collins
then
and
make
sure
they.
A
D
I
use
the
tool,
it's
called
scope
and
it's
it's.
It's
really
it's
to
copy
the
existing
images.
It's
a
one-liner,
so
you
can
just
you
can
just
tell
it
to
sync
from
github
repository
to
the
kwi
repository
and
it
just
goes
ahead
and
does
its
stuff
and
sings
all
the
text.
You
can
also
sync
single
tags
or
copy
single
text.
If
you
don't
want
the
old
craft
but
yeah,
it's
I
I've
done
a
write
up.
Just
let
me
yeah.
C
E
How
long
does
how
long
did
it
take
you
to
migrate
everything.
B
D
C
E
I'll,
let
I'll
and
I'll
let
alex
collins
know
the
that's.
We
should
do
that
for
the
next
release
for
workflows
and
I'll
make
sure
that
happens
for
rollout.
I
think
great.
B
All
right,
I
want
to
kind
of
insert
another
topic-
it's
not
listed
here,
but
it's
related
to
release,
and
I
so
we
noticed
it
only
yesterday.
So
we
have
release
branches,
as
you
know,
and
a
lot
of
users
use
release
branch
as
a
simple
way
to
self
upgrade:
argo
cd
to
patch
users.
B
B
But
yesterday
we
had
a
user
who
had
outage
so
basically,
we
just
cherry
picked
a
change
which
changes,
deployment
manifests
and
it
introduced
new
liveness
probe,
but
that
release
is
not
yet
published
and,
as
a
result,
users
self-upgraded
to
new
deployment
manifests.
But
deployment
manifests
still
were
running
the
previous
image
so
and
it
caused
the
problem.
Basically
it
just
didn't
so
liveness
probe
failed
and
there
was
an
outage
so
and
I
wanted
to
discuss
quickly
how
we
can
improve
it.
One
idea
was
to
introduce
floating
stable
tag
kind
of
so
we
basically,
we
have.
B
E
Yeah,
I
think
I
I
think
I
agree
with
the
use
case,
but
it's
nice
to
just
get
automatic
patch
releases
and
point
it
to
something.
On
the
other
hand,
I
don't
condone
the
use
of
a
branch
to
do
that,
because
we
should.
B
B
E
In
a
branch
and
break,
so
I
think
I
like
your
idea
of
a
floating
x,
dot
y
minor
tag
that
people,
so
they
have.
It's
almost
like
the
docker
image
tags
like
you,
have
the
patch
release
tags
and
you
also
have
the
minor
release
tags
so
that
people
can
choose
their
level
of
comfort.
Of
of,
I
guess,
risk.
B
And
that's
not
so
much
work
for
us.
We
can
just
add
one
more
line
and
release
script.
To
push
two
images,
because
there
are
two
tags:
one
is
patch
and
second,
actually
it
might
be.
No
it
it's
safe.
We
never
release,
you
know
previous,
yet
so
it's
safe
to
always
yeah
upgrade
minor,
realistic.
Most
recent
questions.
B
And
then
I
think
I
can
work
on
it.
At
least
I
can
create
an
issue
and
good
to
get
it
before.
D
D
Maybe
we
should.
We
should
also
document
that
somewhere,
so
we
so
if
you
use
customize,
you
have
the
following
options
or
if
you,
I
think
you
can
also
get
the
installation,
the
the
pre-rendered
installation
manifests
from
from
those
tanks
right.
B
B
B
E
Okay,
but
let's
let's
target
it
like
we'll
get
it
done
four
and
one
not
nine,
whether
or
not
we.
B
Okay,
awesome
so
next
topic
I
wanted
to
quickly
discuss
and
I'm
going
to
throw
books
here.
You
might
have
opinion
so
yesterday
we
presented
our
idea
to
extract
our
go
cd
notifications
core
into
a
library.
We
need
to
decide
where
that
library
goes
and
we
have
options.
B
Not
yet,
no
so,
and
just
to
kind
of
recap,
so
we
have.
Our
identifications
has
code
which
we
want
to
reuse
and
argo
allows
as
a
first
consumer
and
got
it.
We
really
like
ready
to
move
it
into
a
separate
repo
code
is
already
decoupled.
Logically,
it
kind
of
it
leaves
in
a
package
instead
of
argo
cd
notifications,
and
we
just
need
to
move
that
package
into
separate
depot,
and
then
next
question
is:
how
do
we
name
the
clipper?
D
C
C
Our
code
in
a
more
modular
way:
that's
it
right!
So
as
long
as
I
mean
this
is
not
a
fresh
new
project
that
we
are
starting,
we
are
basically
moving
onto
reusable
component
for
better
design,
so
it
doesn't
have
to
go
to
labs.
I
think
it
should
remain
in
argo
prada,
so
I
think
oh.
B
E
I
like
kit,
because
it's
it
differentiates
it
from
a
top-level
project
because
originally
I
thought
like
oh,
it
should
be
called
argo
notifications,
but
then
it
that
implies.
It's
almost
that
it's
a
it's
a
top-level
project
because
we
have
fireball
rollout.
So
I
go
cd
or
go
we're
watching.
Workflows
is
not
renamed
right
now
event
and
I
didn't
want
notifications
to
be
confused
as
a
top
level
project.
So
I
like
the
fact
that
it's
like
it's
like
github's
engine,
so
I
I
like
the
name.
C
Yeah,
I
think
I'm
I'm
good
with
that.
Could
you
could
you
point
me
to
look
or
could
you
drop
on
chat
the
exact
api
directory
that
you're
moving
out?
Yes,.
C
E
So
argo
cd
notifications,
which
is
now
splitting
into
two
repos,
that's
the
that's
the
reconciler
and
then
notification
engine
is,
is
simply
the
library
that
knows
how
to
talk
to
slack
and
pagerduty
and
email,
and
so
that
that
thing
that
understands
how
to
talk
to
everything
is
the
library
and
it
has
a
well-defined
api.
But
that
thing
it
also
has
a
standard
about
the
annotation
text
and
the
config
map.
Syntax.
B
Right
so
this
is
basically
it's
just
a
tool,
and
but
you
have
to
write
reconciler
that
uses
that
tool
and
the
reason
is
a
rollout.
Reconciler
is
going
to
be
very
different
from
argus
identifications,
because
argus
identifications
is
a
sidecar
and
maybe
a
rollout
is
going
to
be
just
you
know.
B
Okay,
thank
you,
yeah,
I'm
really
happy
with
the
name
and
location.
I
will
work
on
moving
the
course.
C
B
It's
a
I
mean
I
will
be
moving,
so
this
argus
identifications
now
consist
of
fairy
concealer
and
it
uses
package
in
the
same
repo,
I'm
going
to
move
the
package
just
the
package,
yes,
and
so
we're
going
to
have
one
reconciler
instead
of
obviously
unifications
another
one.
Instead
of
algorithms-
and
maybe
third
one
is
going
to
be
workflow
notifications.
E
In
case
of
argos
cd,
it's
actually
a
different,
a
whole
different
controller
in
binary
and
deployment
that
that
is
doing
in
the
fusion.
But
in
the
case
of
the
rollouts
I'm
I
don't
need
a
clear
separation.
I
think
it.
The
the
current
rollout
reconciler
can
do
the
work
too.
So
I
plan
to
just
import
the
code
and
use
it
directly.
E
B
Yeah
once
you
go
ahead:
okay
yeah,
that
was
yeah.
So
it's
basically
we've
got
a
lot
of
important
changes
in
one
in
1.8,
punch
release
branch,
and
so
there
was
a
security
fix
which
sneak
found
and
jan
fixed
it,
and
I
think
it's
the
most
important
one.
There
is
also
a
security
fix
for
bundled
decks.
B
We
upgraded
the
dex
version
and
they
were
a
very
kind
of
issue
which
caused
a
lot
of
noise.
Dex,
apparently
was
vulnerable
to
users
who
use
it
to
connect
connect
to
someone
of
providers
kind
of
in
danger.
So
we
needed
to
release
these
changes.
Plus
we
cherry
picked
one
kind
of
hotfix.
B
B
I
propose
to
just
roll
back
that
fix
from
the
release
branch
and
only
release
security
fixes
and
then
keep
working
on
on
a
better
fix
which
doesn't
cause
performance
problems
so
yeah
and
it's
I
feel
it's
like.
I
think
that's
the
only
thing
we
can
do
right
now.
I
don't
want
to
delay
a
release
of
security
fixes
for
another
week
before
we
get.
You
know
a
better
fix
for
buck,
which
I
understand.
B
Yeah,
I
think
in
this
case
here
I'm
I
already
started
at
work
and
I
feel
like
there's
just
no
other
option
we
have,
and
but
I
wanted
to
update
everyone
and
say
that
yeah
this
is
happening
and
just
to
give
a
little
bit
more
details.
So
we
just
we
use
a
library
called
message
pack
to
serialize
cache
and
store
it
in
radius,
and
this
guy
had
a
bug.
B
B
Yeah
he
tried
to
do
it
yesterday,
internally
and
small,
so
we
have
different
definition
of
big
or
small
installation
of
argo
cd.
So
if
rcd
manages
small
number
of
apps
but
many
clusters,
then
you
find,
but
if
you
have
a
lot
of
applications,
that
means
you
need
to
store.
So
many
you
know
small
caches
of
application,
state
and
radius
and
that
and
that
began
it
became
really
really
cpu
consuming
process.
B
B
No,
I
only
attempted
to
kind
of
test
it
internally
yesterday
and
the
testing
failed
yeah.
I
think
I
spent
like
three
hours
trying
to
figure.
A
B
B
B
So
it's
much
easier
to
operate
it
and
the
only
problem
with
the
dogs
was
that
we
wanted
to
keep
using
exact
same
like
looking
field
of
documentation
and
we
didn't
know
how
to
get
version
selector
in
the
ui,
so
people
can
switch
between
versions.
I
think
we
finally
found
how
to
do
that.
B
B
D
A
B
D
B
D
I
I
don't
know
about
the
financial
side
of
the
argo
project
yeah.
So,
but
it's
something
if
we
decide
to
go
with
read
the
docs
and
I
mean
they're
supporting
op
source
projects,
their
services
for
free,
so
maybe
in
the
future.
If
we
are
satisfied,
we
can
decide
to
sponsor
become
a
sponsor
of
read
the
dogs.
I
think
it
starts
at
ten
dollars
a
month
or
five
dollars
a
month
or
something
like
that.
B
And
it's
you
have
any
objections
about.
A
C
A
Yeah,
actually,
if
you
send
me
a
link
to
the
service
and
what
exactly
how
we
would
use
it,
I
can
ask
them
in
general:
I
don't,
I
don't
think,
there's
any
problem
in
things
or
some
service
that
would
be
useful
to
us
main
thing
to
watch
out
for
are
like
licensing
skills
with
a
number
of
users.
The
users
would
be
like
all
country,
videos
or
something.
B
D
Discuss
well,
if
not
I
I
am
made,
I
might
have
another
heads
up,
so
I
think
alex.
I
told
you
already
in
slack,
I'm
working
on
a
complete
rewrite
of
the
documentation
right
now,
so
I
started
somewhere
after
christmas
and
before
the
new
year
yeah,
just
just
as
a
heads
up
there's
something
coming.
D
D
B
D
Yeah
then
the
new
version
is
it's.
I
I
do
it
in
a
completely
separate
repository,
okay,
yeah
so
and
then
it's
it's
based
on
the
same
mk
docs
theme
and
the
look
is
the
same
and
if
it's,
if
it's
in
a
stable
state
someday,
we
can
just
replace
the
docs
directory
and
it
should
be
fine.
So
but
yeah
there's
there's
some
some
migration
issues
or
migration
path
that
we
need
to
consider.
But
we'll
we
work
that
out.
I'm
sure.
B
Okay,
thank
you
sounds
good,
so
I
think
if
there
is
nothing
else,
we
can
wrap
up
and
save
20
minutes.