►
Description
Argo Roadmap Updates For CD, Rollouts, Workflows, and Events - Jesse Suen & Alex Matyushentsev, Akuity; Saravanan Balasubramanian, Derek Wang, Intuit
Hear the latest updates and upcoming features from the Argo project maintainers. The roadmap will cover all the Argo subprojects including CD, Rollouts, Workflows, and Events. You’ll get the opportunity to have a discussion through Q&A and provide feedback in session.
A
Morning,
everyone,
so
there
are
four
leads
and
I,
don't
think
any
of
us
responsible
for
welcome
slide,
so
I
I
didn't
have
a
speech.
Welcome
so
I'll
just
jump
straight
to
part
I'm
going
to
talk
about,
and
thank
you,
okay,
let
me
try
to
fix
it.
So
first
I
will
introduce
myself
so
I'm,
Alex
and
I
work,
mostly
on
Argo
CD
I've
been
working
on
the
project
for
like
five
years
and
I.
Guess
that's
why
I'm
representing
the
project
today
and
I
will
cover
project
State,
as
of
today.
A
I
will
talk
about
what
was
what
we're
going
to
release
in,
like
hopefully
next
week
or
week
after
next
week
and
I
will
cover
a
few
roadmap
items,
and
so
the
first
slide
is
just
a
summary
of
current
project
state.
So,
as
you
can
see,
we've
basically
the
project
is,
you
know,
keep
keep
driving,
so
we
keep
getting
GitHub
stars
and
gig
Styles
itself.
A
They
don't
matter
much,
but
it's
a
good
indication
of
how
healthy
the
project
is
and
how
many
people
it's
attracting,
and
so
in
this
year
we've
got
3000
stars
and
it
make
it
basically
project.
Finally,
bypassed
the
10
000,
ten
thousand,
thousands
10K,
Stars,
Mark
and
I
think
it
was.
It
was
really
exciting
for
everyone
who
work
on
the
project
and
I.
A
Think
even
better
indication
is
how
many
people
better
education
of
project
popularity
is
how
many
people
are
contributing,
and
so
we've
got
around
3000
contributors
contributions
since
beginning
of
this
year,
and
that
includes
commits
and
Bug
reports
from
almost
400
people,
and
it's
awesome,
because
that
really
proves
that
project
is
doing
very
well
and
the
result
of
all
this
activity
features
and
actual
code
and
documentation
changes,
and
so
we've
observed
it
through
like
since
maybe
like
three
years
ago.
That
project
is
moving
forward
very
consistently.
A
We
are
literally
getting
around
100
commits
every
month
and
this
year
is
no
different.
So
to
this
point,
to
this
month,
we've
got
around
900
commits
as
usual.
I
mean
for,
like
that's
the
usual
number
for
like
nine
months
of
work
and
project,
also
releasing
publishing
new
releases
around
every
three
months,
and
this
is
still
happening.
So
we've
published
two
releases,
2.3
and
2.4
and
2.5
is
delayed,
but
it
also
a
very
common
thing.
A
So
we
delayed
release
by
like
two
three
weeks
usually
and
2.5
is
coming
I
think
it
was
supposed
to
be
released
maybe
a
week
ago,
but
we're
still
working
on
last
minute,
bug
fixes
and
finishing
some
feature
polishing,
but
basically
it
will
be
released
hopefully
soon
and
I
wanted.
To
give
you
a
little
like
just
describe
you
of
what's
coming
in
2.5
release
and
I
will
walk
quickly
through
features
and
I.
Think
because
it's
there
is
four
of
us.
A
A
What
is
that
all
right
and
so
2.5
features
so
the
first
one
I
think
the
most
anticipated
feature
for
the
last
year
is
multi-sources
support
and
I'm,
hoping
everyone
know
what
source
means
in
Argo
CD.
If
not
I
will
just
quickly
repeat
so.
Argos
CD
has
a
crg
called
application.
Application
has
a
source
and
destination
destination,
is
a
kubernetes
cluster
in
namespace
and
source?
Is
a
git
repository
or
help
repository,
and
so
it
worked
perfectly
fine
for
most
of
the
use
cases.
A
So,
as
of
2.5,
you
will
be
able
to
take
your
health
chart
home
chart,
don't
modify
it
and
then
connect
kind
of
your
private
repository
with
the
values
that
are
specific
for
your
environment
and
you
won't
have
to
create
a
git
repository
just
to
combine
the
home
chart
and
the
value
files.
So-
and
this
feature
it's
to
be
honest,
not
yet
merged,
but
the
pull
request
has
been
open
for
like
a
month.
A
A
Okay,
as
I
said,
like
no
questions,
I
guess
for
teachers
right
now,
so
I
can
just
move
on
to
the
next
feature.
Next
is
a
server
side
applied?
That's
maybe
not
the
most
visible
feature
like
you
won't
see
a
match
of
changes
in
the
user
interface,
but
it's
a
very
important
feature.
It's
a
I
would
say
it's
a
quality
improvement
feature
so
service
that
apply
refers
to
kubernetes
server
side
apply.
A
Basically,
the
object
that
you
modify
have
to
carry
a
heavy
metadata
information
and
what
it's
bad,
it's
not
great,
because
sometimes
you
want
to
modify
a
very
large
object,
such
as
custom
resource
definition
or
a
config
map
that
has
a
lot
of
data
and
the
metadata
basically
supposed
to
include
the
whole
object.
So
that
means
you
have
to
store
data
in
your
kubernetes
cluster
unnecessary,
and
sometimes
you
cannot
even
basically
the
limit
of
your
object.
A
The
total
size
of
the
object
is
one
megabyte
and
if
you
reserve
a
half
of
the
size
to
metadata,
that
leaves
you
only
500
kilobytes,
which
is
not
enough
and
basically
server
side
apply
resolve
those
limitations.
You
no
longer
need
to
carry
that
much
of
metadata
and
you
can
store
more
in
kubernetes
cluster.
So
an
argue
CG
finally
supports
this
way
of
modifying
objects
in
kubernetes
yeah.
So
that's
actually
it
will
really
improve
performance
of
the
algo
City
itself
and
the
cluster
that
it
manages.
A
A
You
must
trust
that
user.
You
must
basically
the
user
has
to
be
administrator.
So
this
support
of
multiple
namespaces,
your
end
users
can
create
applications
using
cubectl
and
they
don't
have
to
be
administrators
and
they
are
protected.
They
won't
be
able
to
shoot
themselves
in
a
foot
because,
because
of
how
this
feature
is
implemented-
and
we
can
talk
about
details
later,
I
guess
all
right,
I
didn't
tell
us
before
I
started
talking
about
the
slide,
how
many
major
features
we
have
in
2.5?
A
And
yes,
the
next
feature.
It's
another
long
time
anticipated
Improvement
that
ability
to
manage
application
sets
in
algocd,
using
ergocd,
IPI
and
CLI,
and
the
Greek
for
the
context
application
set
is
a
another
crg
in
Argo
CD
world
that
automate
creation
of
algo
CG
applications
and
because
it
was
meant
for
Argo
CG
administrators
first,
it
was
only
available
as
a
custom
resource
that
you
have
to
manage
in
a
cluster
itself,
so
you
must
be
an
admin
to
create.
A
You
know
to
be
able
to
create
application,
set
or
modify
it,
and
we
kind
of
we
intended
to
expose
it
to
end
users,
and
that
feature
is
a
first
step.
So
we've
now
created
API
and
CLI
that
uses
the
API
to
to
create
and
delete
and
modify
application
sets,
and
the
next
step
is
to
introduce
user
interface,
and
this
is
planned
for
2.6
series
and
2.5
only
got
API
and
CLA
and
then
finally
a
feature
that
you
going
to
see
for
sure.
A
So
it's
actually
a
set
of
enhancements
in
the
main
web
page
that
all
rgcg
users
frequently
use.
So
we
have
application
Details
page,
which
is
a
page
that
visualizes
application
State
and
provide
users
controls
to
control
the
application,
and
so
there
are
a
bunch
of
improvements
and
I'm
I.
Just
I
think
I
listed
like
top
three,
that
I
noticed,
but
there
are
more
and
so
basically
the
the
one
is.
A
Basically
you
get
a
button
called
group.
If
you
click
that
button
instead
of
a
tree,
you
will
get
a
Services
representation.
That
kind
of
wrap.
All
the
ports
they
send
traffic
to
and
what
it
makes.
Suddenly
you
get
the
same
amount
of
data
visualized
on
the
page,
but
it
takes
way
less
space
way
less
space
and
it's
much
easier
to
understand.
A
A
I
almost
forgot
I
had
one
more
slide,
so
the
roadmap
site-
and
this
is
again
it's
like
heavily
I-
chose
features
that
I
think
are
important
and
there
are
a
bunch
of
maintainers
here
in
that
room
and
I'm
pretty
sure
we
have
other
features
on
the
roadmap.
That's
maybe
can
even
make
it
quicker
than
the
features
that
on
that
slide,
but
feel
free
to
ask
questions
about
those
features
after
and
so
what
I
believe
is
kind
of
on
roadmap
and
will
be
most
likely.
A
A
We
will
finally
leverage
the
back
end
changes
in
argosity,
UI
and
finally
deliverable
feature
yeah
so
and
next
enhancement.
It's
a
emerge
of
Argo
CD
image
updater
into
our
gocity
itself,
and
so
probably
you
know
what
image
updater
is,
if
not
that
I'll
give
you
a
quick
summary.
Basically,
it's
a
tool
that
lets
you
connect
your
Docker
registry
and
teacher
CD
to
automatically
upgrade
images
every
time
when
a
new
image
is
pushed
into
Docker
registry
and
the
this
little
sentence,
I
wrote
basically
what
it
gives
you
is.
A
It's
like
a
similar
cicd
integration
without
scripting,
so
without
the
tool
you
usually
have
to
write
a
Jenkins
step
or
GitHub
action,
a
script
that
updates
a
deployment
repository
and
it's
such
a
common
use
case.
So
image
updater,
just
automates
it
and
it's
been
available.
The
project
is
still
available
for
like
a
couple
years
already.
It's
definitely
stable
people
use
it
in
production
and,
finally,
the
you
know,
maintainers
of
that
project
propose
to
just
merge
it
into
algocd,
make
it
available
to
everyone
and
everyone
agreed.
A
So
basically,
there
are
people
who
already
committed
to
work
on
it.
Hopefully
it
will
make
it
into
next
release
and
then
last
couple
features
that
I
found
in
our
roadmap.
I
know
that
it
has
a
lot
of
user
requests
and
basically
I
use
this
meeting
to
remind
all
the
maintainers
about
those
two
features.
So
one
is
ability
to
specify
secrets
in
our
CD
parameters,
and
this
is
the
I
think
it's
like.
Maybe
the
most
one
of
the
most
famous
features
from
among
M
users,
because
it
kind
of
brings
secret
management
into
rpcd.
A
At
the
same
time,
maintainers
didn't
want
to
commit
to
build
it,
because
it's
like
a
huge
chunk
of
work
to
introduce
secret
management
and
then
I
think
we
kind
of
in
the
contributors
meeting
discussions.
We
agreed
to
build
a
compromise,
so
there
is
a
we
want
to
let
users
manage
secrets
in
kubernetes
secrets
and
just
reference
them
in
our
good
CD
parameters.
So
it's
a
good
compromise,
basically
allows
end
users
to
connect
secret
management
solution
that
they
want
and
then
just
leverage
secrets
in
algocity
and
hopefully
it
will
solve
the
final
use
case.
A
You
know
end
users
will
be
able
to
yield
secrets
to
generate
kubernetes,
manifests
without
scripting
and
then
finally
another
feature
that
can
help
a
lot
of
crd,
basically
kubernetes
project
maintenance,
so
ability
to
specify
parent
child
relationship
between
objects,
and
the
best
example
is
I
guess
maybe
Argo
CD
itself.
So
in
kubernetes
there
is
a
native
way
to
specify
Which
object
produced
like
that.
You
can
specify
that
there
is
a
one
object
produced
in
other
kubernetes
object
and
argue
CD
leverages
this
native
way
to
visualize
the
three
resource
hierarchy.
A
B
So,
first
some
statistics
so
we're
at
1700
starts
and
I'm
happy
to
say
we
surpassed
Argo
events
recently.
So
it's
a
48
I
mean
it's
not
a
competition,
but
we're
we're
not
last
so
yeah,
it's
a
40
year-over-year
increase.
So
it's
I
think
it's
a
healthy.
You
know
increase
this
year,
we've
done
two
major
releases
of
1.2
and
the
1.3
release
which
I'll
be
going
over
and
that
included
36
new
features
and
66
bug
fixes
and
in
terms
of
contributors.
B
B
So
when
that
two
actually
released
a
few
months
ago,
but
I
thought
it
might
be
useful
to
kind
of
remind
people
what
that
was
about,
because
I'm
not
I
have
a
feeling.
Not
everyone
is
following
rollouts
as
much
so
I'll
quickly
go
through
one,
not
two
before
getting
the
one,
not
three
and
then
the
road
map.
B
B
So
let
me
practice
with
some
dry
runs,
so
you
do
have
the
ability
and
rollouts
to
mark
an
analysis
as
just
let
it
run,
but
don't
if
it
fails,
just
you
know
show
that
it
failed,
but
don't
actually
abort
the
release.
So
that's
and
the
analysis
dry
run
weighted
experiment.
Steps
is
a
feature
if
you're
running
an
experiment,
step
in
your
rollout
for
the
purposes
of
Baseline
versus
Canary
comparison.
B
And
that
way
all
your
metrics
are
are
equal
or
comparable.
So
that's
available
ping
pong
service
management-
if
you
are,
this
mainly
applies
to
ALB.
So
if
you
are
using
AWS
load,
balancer
controller,
one
of
the
challenges
with
that
their
implementation
is
that
they
don't
like
it
when
you
change
the
service
selectors
from
underneath
the
controller,
because
you
can't
take
advantage
of
pod
Readiness
gates
in
the
controller
so
and
we
we've
implemented
like
workarounds
around
that
problem
through
just
actually
talking
making
AWS
API
calls
to
kind
of
verify.
Like
hey.
B
Did
my
my
weight
actually
take
effect?
So
what
this
feature
is
about
ping
pong
service
management
is
that
we
actually
shift
the
reason
why
it's
called
ping
pong
is
because
we
shift
the
weight
back
and
forth
to
the
two
services
so
that
we're
never
changing
the
service
selectors
of
the
service
at
an
opportune
time,
and
so
this
is
compatible
with
the
way
AWS
load.
B
Bouncer
controller
works
is
with
something
the
AWS
team
recommended
in
the
issue
to
to
do
instead
of
like
changing
service
selectors,
and
that
just
applies
in
general,
with
the
database
load.
Balancing
controller
just
know
that
they
don't
like
it.
If
you
just
it's
change
the
target
of
a
service,
and
then
we
also
added
app
mesh
support
in
AWS
and
high
availability
is
and
and
some
scalability
performance
improvements
and
those
all
came
earlier
this
year.
B
Okay,
so
let's
talk
about
what
just
released
today,
one
not
three,
and
so
there
are
two
things
that
we've
been
talking
about
for
years:
Heather
base
routing
and
traffic
mirroring
and
then
so,
if
you
are
using
like
a
more
advanced
traffic
rather,
namely
istio
is
the
one
we
only
support
today,
but
you
can
actually
leverage
their
capabilities
for
doing
something
like
a
header
based
routing.
So
what
this
is
is
with
typical
Argo
worlds.
B
You
can
do
percentage
based
before
this,
but
now
there's
a
new
Canary
step
where
you
can
say,
based
on
this
HTTP
header
send
all
the
traffic
to
the
canary,
and
so
this
you
know,
maybe
if
you
need
to
have
some
stickiness,
while
you're
canoeing,
rather
than
just
some
random
percentage
basis
distribution.
This
will
let
you
do
that
I,
don't
know.
Maybe
you
want
all
your
Firefox
users
to
to
get
the
new
version,
or
something
word
like
that,
but
that's
now
possible,
with
the
header
base
routing.
B
Similarly,
a
traffic
mirroring
is
a
feature
available
in
the
SEO,
where
you
can
Shadow
your
traffic
to
another
service.
In
this
case
the
canary,
and
it
will
it's
mainly
used
for
read
requests,
so
it
won't
Shadow
your
your
quits
and
posts,
but
for
like
gets
and
stuff,
you
can
kind
of
see
how
your
Canary
is
behaving
by
mirroring
the
traffic
to
the
that
service.
B
We
added
support
for
traffic,
so
if
you're,
using
the
traffic
as
your
Ingress
for
Ross
you'll
have
native
first
class
support
enrolls
and
the
there's
some
improvements
to
the
UI
that
roloff's
dashboard,
where
you
can
expand
the
canary
steps.
So
previously
you
can
only
see
like
what
the
percentage
had
shifted,
but
there's
actually
some
more
details
that
that
you
can
show
in
there
and
the
other
Improvement
to
the
UI
was
analysis.
B
When
you
hover
over
some
of
the
analysis,
failures
or
successes,
you
can
see
some
tool
tips
about
like
what
actually
how
it
failed
and
oh
inflicts.
Db
was
added
as
a
first
class
support,
but
I
think
the
thing
that
excites
me
the
most
about
this
whole
list
is
actually
the
last
one.
It's
not
actually
a
rollouts
1.3
feature,
it's
actually
a
new
repo.
B
This
actually
started
out
as
a
just
a
local
post
interface
to
kind
of
view,
a
rollout
if
you
had
a
cube
context
to
your
namespace
and
that's
kind
of
how
it
started
and
that's
how
we
just
thought
it
would
exist.
But
then
people
asked
for
oh
I
would
like
to
run
this.
You
know
as
a
hosted
service
in
my
environment,
and
so
so
then
we
said:
okay,
that's
not
too
hard.
B
It's
all
you
need
to
do
is
like
add
package
in
an
image
and
and
and
create
some
manifest
for
people,
and
so
now
people
are
asking
okay
I'm
running
it,
but
then
now
anyone
who
can
get
to
the
UI
can,
like
you
know,
abort
a
roll
out,
promote
it
and
do
whatever
they
want.
So
I
think
there
might
be
some
ways
we
can
try
to
add
a
light
weight
off
without
going
full
blown
like
we
did
with
Argo,
CD
or
workflows,
but
I
think
it's
it's.
B
B
B
C
C
C
The
3.3
was
released
a
few
months
back
just
I
like
to
touch
base
like
what
are
the
major
feature
we
had
written
in
that
one.
The
one
is
like
a
life
cycle
hooks
which
will
help
the
user
to
integrate
with
the
notification
and
monitoring
system
for
the
thought
pro
life
cycle.
So
if
they
want
to
integrate
the
slack
or
page
of
Duty
for
the
workflow
lifecycle,
they
can
use
that
life
cycle.
C
The
second
one
is
like
a
plugin.
The
plugin
will
give
give
the
user
to
extend
that
Argo
workflow
to
Define
their
own
templates,
which
is
a
very
powerful
feature
which
was
released
in
the
3.3,
and
we
have
like
some
of
the
open
source
plugins
in
our
GitHub
lab
for
the
slack
integration,
python
integration
and
all
the
things
please
go
ahead
and
take
a
look.
C
Then
another
one
use
case
is
like:
if,
if
you
go,
if
your
organization
has
like
a
proprietary
functionality
which
cannot
be
open
source,
but
you
want
it
in
that
Argo
workflow,
you
can
develop
as
a
plugin.
The
third
one
is
like
the
multi-tenant
support
for
SSR
back
so
in
the
3.2.
Our
UI
will
support
like
a
centralized
cluster
level
or
backs
so
3.3
is
enable
you
can
configure
the
namespace
level
or
backs
for
the
Argo
UI.
C
It
comes
with
the
two
two
enhancement.
One
is
like
artifact
visualization.
Another
one
is
artifact
GC.
The
automatic
visualization
is
enabling
to
display
the
content
in
that
Argo
UI.
So
the
user
can
directly
see
that
artifact
contents
in
that
Argo
UI
I
just
put
a
snapshot
in
that,
so
they
it
can
show
that
the
image
which
is
the
artifact
generated
by
one
of
the
hour
plus
tip
they
can
when,
when
they
click
it,
they
can
immediately
see
it
in
that
UI.
They
no
need
to
go
to
that
any
cloud,
storage
or
anything.
C
The
second
one
is
like
a
artifact
garbage
collections,
so
another
problem
user
is
facing
like
to
cleaning
up
that
artifact,
which
is
generated
by
your
workflows.
So
we
implemented
artifact
garbage
collection,
so
that
user
can
Define
the
strategy
in
that
workflow
when
they
want
to
delete
the
artifacts
so
whether
in
the
workflow
completion
or
workload
deletion.
This
is
a
two
strategies
available
now.
C
So
then
Argo
controller
will
automatically
go
ahead
and
clean
up
that
artifacts.
Whatever
the
active
Factor
posterity,
they
configure
was
completed
the
security
audit
by
idea
logics
so
everything's
compiled
with
the
CNC
of
security
Norms,
the
third
one.
We
enabled
that
Azure
blood
storage
for
that
artifacts.
C
Let's
go
to
the
road
map,
so
the
these
are
the
road
map,
so
mainly
in
the
3.5.
We
like
to
step
back
clean
up
our
backlogs
of
bugs
and
establish
that
what
are
the
features
so
far
we
developed
so
giving
a
little
bit
time
to
the
community
to
start
using
the
new
features
find
a
few
bucks,
so
we
can
help
to
fix
those
bugs
and
we
have
like
a
few
Tech
tips.
So
we
are
like
to
complete
those
all
the
things
in
3.5
under
3.6.
C
D
D
We
have
two
major
releases,
106
and
107,
which
contain
42
new
features
and
70
bug
fixes,
and
there
were
197
commits
in
total,
and
all
of
this
were
down
by
56
contributors
and
our
GitHub
Stars
grew
from
1200
to
1600,
which
is
about
32
percent
increase
and
we're
now
like
48
stars
behind
Argo
world.
So
so,
please.
D
About
features,
one
of
the
most
important
features
we
have
with
the
past
releases
is
like:
we
had
a
new
event.
Bus
implementation,
which
is
based
on
gesturing
gesture,
is
a
product
from
another
cncf
project
net
and
the
reason
we
had
this
Json
event
bus
is
because
our
original
human
implementation,
which
is
also
based
on
another
nice
product,
not
streaming,
and
that
is
going
to
be
end
of
life
in
2003,
so
we
had
adjustment
as
a
replacement.
D
Another
reason
for
this
just
remember
us
is
gstream,
is
not
only
a
persistable
messaging
system,
but
also
provides
some
other
fancy
features
such
as
key
value
store.
So
we
leverage
the
schemadow
store
to
eliminate
some
of
the
restrictions
of
a
privileges
had
with
no
streaming
bus.
That
makes
other
events
even
more
powerful.
D
Another
nice
feature
we
had
is
the
Event
Source
filtering
and
that's
a
great
experiment
to
our
existing
sensor
building
features,
and
then,
with
this
feature,
you
have
the
ability
to
figure
out
those
messages.
You
are
not
interested
in
at
the
even
Source
level
so
that
you
can
reduce
the
messages
coming
to
the
system.
D
D
Our
supported,
Event
Source
family
has
always
improved
a
lot.
We're
now
supporting
big
bucket
cloud
and
server
event
sources
with
existing
GitHub
and
GitHub.
You
can
Source,
you
can
use
other
events
to
do
any
beat-ups
with
all
the
popular
Source
can
management
tools
we're
also
supporting
Reddit
risk
stream
even
source
event.
Transformation
feature
is
another
amazing
one
that
gives
you
a
chance
to
to
transform
the
the
event
you
received
before
it
passed
as
a
parameter
to
your
Downstream
and
Trigger
like
workflows
or
Lambda,
stuff
and
Trigger
condition.
D
D
In
terms
of
the
controller
Department,
we
combine
we
originally.
We
originally
have
had
like
three
different
departments
for
installation,
even
bus
controller
sensor,
controller
email,
Source
controller.
We
now
combine
all
of
those
three
into
one
controller
manager
Department,
and
we
also
enable
h2a
for
it,
so
that
will
simplify
your
installation
process.
D
D
D
D
So
if
you
only
have
one
single
events
in
in
the
image
source
object,
so
you
are
expecting.
You
know
there's
a
restart,
that's
okay
for
you,
but
if
you
do
multiple
Event
Source
in
one
crd
object
and,
for
example,
I
make
some
change
to
the
Copper
config.
So
it's
unfair
to
the
SQL
events
or
say
hey.
Why
do
you
restart
my
service
right?
D
The
other
thing
is
about
the
error
of
failure,
reporting
mechanism.
Let's
say
if
you
only
have
one
events
configured
in
one
investors
object
and
then
let's
say
the
Kafka
and
you
have
a
wrong
broker
URL
configured.
So
it's
expected
that
you
know
the
the
Kappa
events
of
watching
service
will
not
be
running
correctly
and
your
envelope
is
either
a
power
crash
or
some
you
know,
error
status
in
the
crd
options,
even
for
serial
object.
D
Is
we
just
you
know,
era
out
the
cup
tolerance
and
watching
service
and
prints
printing
out
some
logs,
but
for
sqs
what
events
watching
service
it
still
can
be
running
fine,
so
the
part
will
still
be
running
status
and
there's
you
know
besides
the
logs
you
see
in
the
in
part
and
then
there's
nothing
else
to
let
you
know
so
we're
trying
to
figure
out
a
way
to
have
a
better.
You
know
error
reporting
mechanism
so
that,
as
owner
you
can
quickly
detect
those
errors
but
also
think
about
it
too,
support
loading.
D
D
The
other
item
is
like
we
are
trying
to
support
figure
out
if
it's
a
good
way
to
support
like
Dynamic
Event
Source.
How
do
I
understand
that
I
will
give
you
a
you
know.
Example
use
case
right
so
today,
if
you
have
apps,
you
have
several
Upstream
Services.
There
are
publishing
events
to
a
single
topic
and
then
to
identify
which
Upstream
it
is.
D
So
that's
something
you're
not
gonna
happen.
So
why
don't?
We
just
you
know,
come
up
with
one
single,
even
Source,
watching
the
single
topic
and
then
dynamically
publish
events
to
different
categories
based
on
the
the
condition
is
specified,
for
example
based
on
the
type
you
know.
So
that's
you
know
the
dynamic
Event
Source
feature
I'm.
Thinking
of
the
last
item
in
on
the
on
the
list
is
about
the
tracing
open,
Telemetry
support,
that's
quite
straightforward,
and
so
even
driven
system.
You
know
we
want
to
have
some
accessibility.
D
There's
another
item.
I
didn't
list
in
the
in
the
in
a
real
map
is
I'm.
Thinking
of
you
know
not
sure
if
it's
a
good
idea
to
support
more,
even
buzz
like
Kafka,
you
know,
I
know
there
are
a
lot
of
companies.
They
have
management
Capital
Service
and
they
want
they
want
to
use
the
existing
Kafka
service
and
the
event
bus.
You
know,
so
you
don't
need
to
start
any
resource
in
in
your
customer
and
then
to
reduce
resource
or
get
better
support.
You
know
things
like
that.
D
E
For
the
first
time
in
Tech
conference
history
we're
actually
ahead
of
time,
we
were
supposed
to
have
a
break
at
10
30
to
10
45.
Before
we
break
the
workshops,
so
I'd
say
we
take
a
break
now
and
then
we'll
reconvene
at
10
30
instead
of
10
10
45.
is
everyone.
Okay,
with
that
they
actually
get
a
little
bit
longer
break,
but
then
we'll
get
a
little
bit
more
time
for
the
workshops
and
an
update
to
what
I
said
earlier
this
morning.
We're
actually
going
to
do
work,
flows
and
events
on
that
side.
B
E
From
from
here
and
there,
so
if
you
see
the
end
roll
up
that
side,
Brooklyn
events
on
this
side,
everyone
over
that,
so
there
should
be
coffee,
I,
think
there's
a
bit
of
fruit
outside
so
feel
free
to
God.
Refresh
yourself
come
back
in
here
at
10
30..
You
want
to
kick
off
the
workshops
thanks.
Everyone.