►
From YouTube: Keptn Community Meeting - February 17th, 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).
A
All
right
welcome
everyone
to
today's
captain
community
meeting,
February
17th
I
know
most
of
you
that
are
probably
watching
this
from
the
states
you're
watching
this
later
on,
because
today
is
a
holiday
in
the
United
States
Presidents
Day,
so
enjoy
the
day
off
for
those
to
join
us
live
thanks
for
joining
us
live.
We
have
a
packed
agenda,
but
we
still
try
to
keep
it
Swift
I
will.
Most
of
the
presentation
today
will
be
done
by
Jurgen
and
also
Ivar.
B
A
A
If
we
cannot
ski
anymore
and
then
we
are
losing
against
the
people
that
actually
get
this
no
I,
don't
even
know
where
the
snow
comes
down
these
days,
the
west
coast,
there's
a
lot
of
snow
yeah
all
right
enough
enough
with
enough
with
the
skiing
and
snow
and
everything
that
is
weather,
related
I
want
to
hand
it
over
to
you
guys,
folks,
there
online,
please,
you
know
if
you
have
questions,
put
them
into
the
the
Q&A
feature.
You
can
also
use
this
leg
channel.
A
C
You
Andy
I'll,
leave
it
to
you.
You
have
been
also
good
morning
from
my
side.
I
want
you
to
walk
you
through
this
community
meeting
and
we
have
a
very
packed
agenda
and
I'm
also
I'm
really
happy
to
have
llevarse
tus
here
in
this
meeting,
and
we
will
talk
about
feature
flags
just
in
a
minute,
but
before
I
want
to
give
a
big
shout
out
to
the
to
the
captain
team
that
just
came
up
with
a
new
captain
bridge,
so
we
have
an
early
adopter
program.
C
C
Have
captain
dot
six
installed
and
you
want
to
upgrade
to
the
new
bridge?
It's
basically
just
those
two
commands.
You
just
want
to
copy
them
here
and
execute
it
from
your
local
machine.
That's
connected
to
your
captain
installation
and
then
you
get
all
the
new
shiny
and
bright
captain
bridge,
and
this
is
also
the
bridge
we
are
going
to
use
later
on
in
the
demo.
So,
just
in
case
that
you
don't
get
like
confused,
we
will
go
in
to
use
the
new
bridge.
It
looks
quite
different
from
the
previous
version.
C
It
has
a
couple
of
more
features
and
yeah.
That's
the
two
commands.
You
can
also
upgrade
your
installation
to
the
new
bridge
and,
of
course,
we're
always
looking
for
feedback.
So
let
us
know
in
the
slack
workspace
in
the
channel
that
if
you
have
to
upgrade
it
and
how
you
like
the
new
version
and
if
there's
anything
you
would
like
to
see
in
the
new
version
as
well.
C
So
that's
pretty
cool
one
part
of
captain
where
we
are
continuously
improving
and
another
part
is
actually
improving
capabilities
around
self-healing
and
we
did
already
a
great
implementation
with
unleash,
which
is
a
feature
flagging
framework
and
I
will
invite
Eva
to
talk
a
little
bit
about
this.
So
we
have.
We
are
happy
to
have
him
here
and
he
prepared
actually
some
some
slides
for
us.
So
I
will
give
you
the
stage.
I
will
switch
here
in
Crescenta
mode,
and
you
just
let
me
know
when
I
should
continue
on
to
the
next
slide.
B
B
Most
of
you
are
probably
read
the
book
or
read
this
this
study
every
year
and
they
have
looked
into
what
or
they
have
kind
of
done,
a
survey
on
a
lot
of
the
software
companies
all
over
the
world
and
what
I
try
to
figure
out
is
what
do
the
elite
performers
do
different
from
the
low
performers
and
when
they
talk
about
elite
performance,
they
talk
about
successful
companies
that
earns
a
lot
of
money
is
like
Google
style
companies.
What
do
they
do?
B
The
lead
time
from
commit
to
deploy
is
much
much
more
frequent
or
is
much
lower
and
if
they
discover
error,
they're
really
much
faster
on
fixing
those
errors,
probably
because
they
did
prior
to
production
in
smaller
batches
or
much
much
easier
to
identify
the
errors,
but
also
that
I
automated
a
lot
of
the
process
around
putting
code
into
production
in
the
first
place,
but
I
also
find
that
I
actually
tend
to
have
less
errors
when
deploying.
So
this
is
interesting.
B
So
if
you
go
to
the
next
slide
so
that
they
identified
kind
of
which
techniques
and
approaches
to
software
development
do
these
elite
performers
use.
Of
course
it's
it's
about
DevOps,
of
course,
under
the
diverse
practices
that
we
know
like
deployment,
automation
and
continuous
delivery,
automated
testing,
but
they
also
have
this
trunk
base
development
as
a
technique
for
for
for
these
elite
performers.
So
if
you
go
to
the
next
slides,
yes,
this
is
trunk
base
development,
so
I
actually
assumed
that
most
of
the
attendees
are
on
this.
B
You
know
that
most
of
the
code
is
already
on
the
master
branch
and
working
in
this
way
also
requires
it
to
kind
of
think
up
the
features
you
are
working
on
and
actually
shut
them
down
into
smaller
pieces
that
you
can
integrate
with
the
master
branch
at
all
times.
But
then,
if
you
got
next
slide
there
again,
this
is
actually
where
you
need
a
technique
like
future
target,
because
even
though
it's
nice
to
split
up
your
new
feature
into
smaller
chunks,
it
doesn't
always
make
sense
to
release
those
small
parts
to
real
users.
B
B
So
it's
actually
about
being
able
to
to
the
isolated
the
the
new
feature
may
have
more
control
on
how
you
expose
that
feature
to
your
users.
So
if
you
go
to
the
next
slide,
so
so
what
is
a
feature?
Toggle
down
is
actually
just
on
this
statement.
In
your
code
right
and
it
protects
the
new
functionality
from
the
old
functionality
and
that's
about
it's
actually
a
very
simple
technique,
but
of
course
there
is
more
to
it
it's
more
about
how
do
you
know
which
user
to
enable
it
for
in
which
context
and
which
servers?
B
How
many
percent
of
your
users,
of
course
there
is
more
complexity,
but
the
principles
are
just
an
if
statement.
So
if
you
go
to
the
next
slide,
it's
about
so
what
you
you
happen.
What
happens
when
you
use
a
technique
like
future
toggles
is
that
you
are
actually
decoupling
the
process
of
putting
code
into
an
environment
from
releasing
features
to
your
users,
and
this
actually
is
a
way
to
reduce
the
risk.
There
is
always
risk
associated
with
with
putting
new
code
into
production.
B
You
can
never
take
that
away,
but
at
least
you
reduces
that
risk,
because
you're
not
necessarily
exposing
that
new
functionality
to
users
before
you
are
tested
on
yourself.
Yet
so,
if
you
go
to
the
next
slide,
yes,
there
is
downsides.
Yeah
future
toggles
by
Nature
is
technical
depth
in
your
code.
You
can
never
get
away
from
that
because
you
have
two
versions
of
a
functionality.
Probably
you
have
the
old
version.
B
You
have
the
new
version
you're
working
on,
and
this
increase
is
also
the
complexity
when
testing
your
codes
and
if
you
have
a
lot
of
feature
tunnels
going
on
in
the
same
code
base,
there
is
a
lot
of
different
variations
that
you
probably
haven't
talked
about
testing
together.
So
we
should
be
aware
of
the
downsides
of
using
future
terms.
B
B
So
you
don't
have
to
have
big
changes
all
the
time,
but
you
should
also
look
at
how
you
can
reduce
the
duration
of
a
toggle
or
how
long
you
keep
the
toggle
around
and
try
to
clean
off
the
toggle
when
it
has
served
its
purpose
because
having
the
table
around-
and
we
see
this
a
lot
of
times
with
users
of
English
or
is
that
they
keep
an
around
just
because
it
was
so
nice.
They
use
invested
a
lot
of
time
in
this
new
toggle
and
they
tend
to
not
remove
it
and
I.
B
B
It's
been
something
I've
been
working
on
for
for
quite
a
few
years
now
and
last
year
in
April,
we
decided
to
start
this
hosted
offering
and
the
host
of
this
it's
a
hosted.
Version
of
them
is
open
source,
but
it
also
adds
some
extra
additional
enterprise
functionality
to
it,
and
we
know
that
the
alicia's
is
been
tested
in
the
wild.
We
have
more
account
no
for
more
than
1
million
dollars
a
year
on
the
clients,
the
client
SDKs,
and
we
have
more
than
just
the
last
week
it's
almost
$16,000
of
the
of
the
node
SDK.
B
So
if
you
go
to
the
next
slide,
so
I'm
not
going
to
do
all
the
details
of
unleash,
but
I
think
and
conceptual
overview
on
how
it
works
is
important,
unleashes
compromised
of
three
parts.
You
have
the
unleash
admin
UI
very
actually
going
in
the
final
toggles,
your
rollout
strategy
and
stuff,
like
that
yeah
the
I
of
course,
integrate,
and
then
you
have
the
client
SDK
and
this
the
client
SDK.
That
does
all
the
work.
B
So
you
never
share
your
kind
of
user
data
with
an
leash.
It's
all
happens
locally
in
your
application
and
if
you
go
to
the
next
slide,
that
would
be
the
last
slide
for
now.
These
are
the
core
concept
that
you
should
be
aware
of
when
integrating
with
a
mesh-
and
you
have
the
context.
This
is
the
information
that
you
provide
to
the
SDK
that
is
kind
of
the
context
of
the
current
request.
Typically,
what
you
put
in
there
is
like
who
is
the
current
user?
B
What
is
the
user
session
ID
or,
if
you
do
some
regional
based
rollout
to
be
the
country
or
something
like
that,
and
the
activation
strategies
are
kind
of
the
rules
that
is
executed
inside
SDK
to
decide.
Given
this
context,
should
this
feature
that
will
be
enabled
or
disabled
and
that's
about
it,
cool
thank.
B
A
Not
kind
of
site
wreck
your
the
rest
of
the
presentation,
so
I
will
keep
it
to
the
end,
but
also
they
remind
the
folks
that
are
online.
If
you
have
questions,
please
put
them
in
either
in
zoom,
or
you
know
slack
you
canyou
can
use
all
the
different
channels
how
you
can
reach
us,
but
there's
a
lot
of
interest
and
yeah.
B
C
C
Things
to
happen,
awesome
thanks,
Andy,
so
and
thanks
you
vote
for
for
showing
us
all
these
details
on
unleash.
What
I
want
to
take
now
is
kind
of
to
see
unleash
in
action
and
also
how
it
works
together
with
with
Captain.
So
first,
if
you
are
interested
in
more
details
of
unleash,
the
presentation
has
more
content
in
it.
So
you
will
find
all
the
nodes
here
again
in
our
agenda,
which
is
also
shared
on
our
it's.
It's
you
know
it's
like
channel.
You
will
find
it.
C
You
will
also
find
it
in
our
community
github
repository,
so
you
will
also
find
you
will
always
find
the
links
to
the
capital
community
meeting
notes
and
you
will
find
the
links
to
the
slides
also
here.
So
let
us
not
take
a
look.
How
does
it
actually
look
unleash
and
what
I
have
here
configured
and
set
up
is
basically
very
simple.
Maybe
it's
like.
C
Actually,
we
could
kind
of
start
here
by
enabling
this.
So
we
just
click
start
here
and
maybe
heard
in
the
background
on
that
my
slack
notification
was
what's
coming
up.
So
let
me
show
you
there
should
be
in
my
slack
actually
because
captain
and
unleash
they
are
integrated.
So
I
can
see
now
in
my
configuration,
that
my
configuration
has
changed
for
the
cards
micro-service
in
my
stock
shop
and
I
can
also
see
what
I've
just
did
I
enable
the
promotion
and
assets
the
value
to
true.
So
that's
correct,
I,
just
clicked
on
the
feature
table.
C
I
just
enabled
one
feature
talking.
So
that's
one
great
part.
You
can
directly
see
this
in
your
slack
channels.
If
you
have
this
integration
set
up,
but
we
will
also
see
this
in
our
new
version
of
the
captain's
bridge.
So
just
right
now,
I'm
just
gonna
reload
this
and
we
should
be
able
to
see
the
change
also
in
the
captain's
bridge.
So
that
means
that
if
you
have
the
integration
setup
between
unleash
and
captain,
then
you
will
get
all
this
information
directly
in
captain
each
time
to
change
a
configuration.
C
So
you
you
don't
have
to
go
back
to
unleash
to
see
which
confirmations
right
now
actually
in
place,
but
you
can
just
see
this
in
your
captain's
bridge.
So
if
we
take
a
look
at
our
stock
shop,
so
we
can
see
this
right
now.
I
just
see
the
configuration,
change
and
I
will
see
all
the
like
all
the
details
on
the
right
hand
side.
C
So
that's
a
little
bit
of
the
the
captain's
bridge
where,
on
the
left
hand,
side
we
have
like
all
the
changes
that
happened
like
configuration
change
when
we
do
a
new
deployment,
but
also
like
configuration
changes
only
on
one
part
of
the
configuration
and
we
can
directly
see
which
stage
it
affected.
So
I
did
a
configuration,
change,
introduction
actually
and
here
on
the
right
hand,
side
I
can
see
all
the
all
the
details
on
this
before
I
go
back
to
unleash
to
take
a
look
and
what
what
was
there.
C
In
addition,
I
will
just
go
a
little
bit
into
dynaTrace
because
we
have
ours
it's
monitored
here
and
we
should
should
be
able
to
see
here
on
the
icons
controller.
We
will
also
see
there
was
a
configuration
change.
So
actually
captain
was
forwarding
this
configuration
change
from
unleash
to
dynaTrace,
so
we
cannot
see
that
there
was
now
a
configuration
change
done
in
unleash
and
we
set
the
value
to
true.
C
So
we
enabled
this
promotion
and
we
can
see
this
here
so
first
of
all,
we
can
see
kind
of
a
very
tight
integration
by
means
of
having
all
these
notifications,
like
the
information
in
all
the
tools
we
are
connecting
like
in
dynaTrace,
in
slack,
of
course,
in
the
captain's
bridge
and
everything
is
done
via
captain
with
the
same
mechanism.
We
always
do
like
it.
Everything
is
event
based
captain's,
sending
out
events
to
the
notification
service
or
to
dynaTrace,
and
it's
capturing
these
events
from
from
unleash.
B
This
is
actually
data
sharing
from
the
client
rusticate
back
to
on.
This
is,
of
course,
optional
because
of
privacy,
but
the
only
thing
that
you
see
is
how
often
does
it
evaluated
through
your
force,
and
this
is
really
nice
to
debug-
that
stuff
is
actually
working
as
you
expected,
if
you
do
gradual,
based
robots,
it's
really
nice
to
see
that
they
actually
are
evaluating
to
true,
sometimes
or
that
it
matches
the
percentage
you
are
expecting.
So
it's
a
simple
way
to
debug
that
stuff
words
that
you
actually
have
enabled
this
new
future
for
someone.
C
Sure
I
just
want
to
to
highlight
here,
like
the
numbers,
so
in
the
background
I'm,
just
adding
items
to
my
shopping
cart,
so
I
have
a
load
generation
up
in
the
background.
So
this
is
why
the
numbers
are
quite
high
here.
They
were
just
evaluated
like
in
the
last
hour
or
when,
when
I,
we
started
this
ok
strategies
and.
B
So
you
are
not
used
at
the
evil
strategy,
and
that
means
it's
basically
on
or
off
so
if
the
toggle
is
enabled
it
will
be
enabled
for
everyone,
but
there
is
other
strategies
as
well.
We
have
gradual
rollout
strategies,
we
have
user
with
ID,
where
you
can
target
specific
users,
and
you
can
even
define
your
own
custom
strategies.
B
Then
you
also
have
to
provide
the
implementation
on
the
client-side
so
that
you
can
do
more
fancy
thing
that
fits
more
with
what
you're
doing
as
a
business
could
be
that
you
want
specific
rules
for
specific
regions
or
stuff
like
that.
So
that's
the
extension
points.
Basically,
we
also
do
all
the
trail,
of
course.
So
we
have
a
history
tab,
which
is
basically
all
the
changes
that
happens
to
a
logo
so
that
you
can
see
huge
change
both
well
then,
this
is
quite
useful.
B
If
something
happens
and
you
don't
have
the
integration
together
with
captain,
it
would
be
really
nice
way
to
just
see
what
has
been
changed
on
this
territory,
of
course,
yourself
or
the
trail
across
all
Carlos.
Just
to
see,
if
something
happens
at
the
specific
time,
you
can
go
back
and
correlate
that
with
it
logs
for
a
foundation
that
does
it.
C
I
guess
that's
pretty
cool,
yeah
and
I
said
this
kind
of
integration
like
the
the
history
or
like
the
audit
trail,
is
also
available
in
captain.
Just
one
reminder:
if
we
want
to
use
feature
flex,
you
also
have
to
prepare
your
application
for
the
feature
flags.
So
in
our
case
we
have
with
a
very
simple
shopping,
cart
and
we
have
this
kind
of
feature
flag,
and
this
that's
called
a
sophisticated.
C
If
statement
where
actually
we
check
for
the
unleash
server
if
this
feature
flag
is
enabled,
so
we
have
our
application,
our
demo
application,
we
are
using
right
now
we
have
it
prepared
to
be
able
to
to
take
the
configuration
from
unleash.
So
if
it's
enabled
or
not-
and
basically
you
will
just
you
want
to
connect
to
your
unleash
server
instance-
and
you
would
just
then
fetch
the
configuration
and
mine
is
just
living
here
on
my
captain
McCluster.
Basically,
so
that's
that's
pretty
cool.
C
Let
us
now
see
what
really
happens
when
we,
when
we
updated
this
feature
flag.
So
in
the
metrics
we
can
see
like
a
lot
of
people
in
the
last
hour,
17%
of
the
people
they
are
now
receiving.
This
feature
flag
were
actually
for
them,
it's
active,
but
in
the
last
minute,
or
actually,
since
we
enabled
it
it's
100
percent.
C
So
also
we
should
see
in
dynaTrace
that
a
lot
of
that,
we
that
we
should
see
what's
happening
here
and
I-
was
just
looking
at
the
right-hand
side
corner
right
top
corner,
because
a
dinosaur
is
just
detected
a
province.
So
let
us
investigate
what
is
what
is
going
on
in
our
environment
and
actually
on
the
items
cards
controller.
There
seems
to
be
now
a
problem
and
dynaTrace
actually
gives
us
all
the
information
about
this
problem.
C
What
really
is
going
wrong
and
our
failure
rate
was
increasing
up
to
30%
I
said
it
here
for
a
custom
threshold
of
10%,
so
whenever
we
exceeded
I
want
to
get
alerted
or
actually
I
want
to
to
kick
off.
The
diary's
risk
root
cause
analysis
and
give
me
an
x
rays
problem
ticket,
and
what
I
can
see
here
is
actually
that
the
configuration
change
was
part
of
the
is
part
of
this
problem.
C
If
statement,
if
there
is,
if
we
enable
this
feature
flag,
if
we
want
to
enable
the
promotional
campaign,
then
there
is
some
other
things
we
do
in
the
code,
but
actually
the
only
thing
I
did
is
putting
there
an
exception
that
this
functionality
is
not
being
it's
not
yet
implemented.
So,
of
course,
we
are
causing
a
lot
of
errors
here
and
dynaTrace
was
able
to
catch
this
and
give
us
back
this
information.
This
configuration
change
is
now
the
root
cause
of
the
problem,
because
you
should
just
not
do
this
configuration
change.
C
You
should
not
enable
this,
but
what
we
can
also
see
is
that
captain
was
triggered
right
now
when
dynaTrace
was
opening
the
problem
ticket
captain
was
triggered
and
the
action
captain
should
do
is
actually
to
change
the
feature
toggle
back
to
off.
So
we
are
asking
captain
here
to
change
the
feature
toggle
and
send
it
back
to
off,
and
captain
responds
back
with
that.
It's
finished
this
feature
toggle
and
has
said
to
to
its
new
state
to
its
new
value,
and
it's
it's
off,
so
it's
disabled.
So
how
can
we
verify?
C
First
of
all,
we
can
take
a
look
in
the
captain's
bridge,
so
what
we
can
see
here?
Actually
it's
auto
reloading.
The
new
version
is
also
Auto
reloading.
So
we
can
see
a
problem
was
detected,
the
failure
rate
increase
and
we
can
see
the
problem
is
open.
We
can
see
all
the
details
of
this
problem
here
and
we
can
see
right
afterwards.
There
was
a
new
configuration
change
in
captain
captured
by
captain,
and
it
says
it's
now
set
to
false.
C
So
actually
this
configuration
changed
the
second
one
here
this
should
actually
come
from
from
unleash
because
that's
our
integration.
So,
let's
take
a
look
at
unleash
and
oh
it's
already
going
back.
Let
me
also
reload,
because
then
we
would
see
the
features
like
now:
I
switch
back
to
disabled
and
in
the
last
minute,
while
I
was
talking
and
explaining,
we
can
see
in
the
last
minute.
No
none
of
our
users
were
now
affected
by
this
feature,
flag
or
not
affected
by
the
feedback,
but
they
did
not
get
the
enabled
version
of
the
feature
flag.
So.
B
So
super
cool
because,
even
though
its
future
flights
enables
you
to
turn
back
off
features
that
proves
to
be
flaky
or
it
doesn't
work
as
expected,
really
fast,
often
much
much
faster
than
rolling
back
to
the
previous
version.
But
but
again
you
still
have
to
detect
the
problem
and
respond
to
it.
So
there
is
still
this
human
factor
that
you
have
to
have
proper,
alerting
in
players.
You
have
to
be
sitting
right
there
looking
at
Yeller
or
getting
the
alert
and
figuring
out
what
those
alert
mean
and
do
the
proper
action.
B
C
Exactly
because
it's
not
easy
to
to
kind
of
stay
ahead
of
all
this
compliment,
complexity,
I
would
say
because
now
you
have.
The
good
thing
is
that
you
have
deployments
and
release
is
decoupled,
but
now
it's
also
might
get
difficult
to
define.
Okay,
if
it's
not
the
deployment,
it
has
to
be
the
release.
So
where
is
the
where's?
The
error
we
can
see
a
lot
of
our
users
are
getting
an
error.
We
have
to
figure
this
out,
so
the
power
of
having
monitoring
and
captain
and
like
configuration
tools
or
configurations
at
runtime.
C
C
Demo
repository
because
for
the
user
itself
there
is
nothing
there
is
no
coding
involved,
so
the
user
doesn't
have
to
be
a
pro
in
in
Java
coding.
But
actually
what
I
did
is
to
my
I
added
a
revelation
file
to
my
config
repository,
that's
managed
by
captain
and
I,
just
edit
the
instructions.
What
captain
should
do
and
actually
I
added
the
instructions
that,
if
there
is
a
failure
rate
increase
for
discards,
microservice
in
production.
So
this
is
how
it's
configured.
C
Then
there
is
actually
an
action
that
should
be
executed,
and
it
should
be
that
the
enable
promotion
this
feature
toggle
should
be
set
to
offer
disabled.
So
there
is
not
a
lot
of
coding
involved,
actually
there's
no
coding
involved
at
all,
but
it's
just
a
matter
of
configuration
and
it
could
be
also.
You
want
to
have
scaling
or
rollback
or
other
things,
but
in
my
case,
I
just
have
like
a
couple
of
feature.
Toggles
and
I
just
want
to
provide
remediation
actions
in
risk
bombs.
C
If
there's
anything
going
wrong,
like
with
my
failure
rate
increase
I
just
want
to
turn
this
feature,
toggle
off
and
I.
Think
that's
pretty
cool
and
pretty
powerful.
We
can
also
take
a
look
now
in
Ghana
trees
if
it
actually
resolved
our
problem.
So
let's
take
a
look
back
to
our
to
our
cards
micro-service.
Actually
it's
green
already.
So
that
means
the
problem
is
closed.
We,
but
let
me
to
first
show
you
here
the
events
so
the
first
event,
the
word
the
first
one.
We
already
investigated
that
she'll
do
this.
C
So
captain
can
then
automatically
call
out
to
unleash
and
set
it
back
to
its
previous
state
or
just
turn
it
off.
So
that's
I
think
that's
a
pretty
cool
thing,
and
actually
we
talk
a
lot
now.
So
maybe
maybe
now
it's
a
it's
a
good
time
to
take
some
questions
and
if
you
have
some
questions
around
featured
feature
flags
for
a
bar
or
for
the
demo,
we
are
happy
to
take
yeah
sure.
A
And
just
as
a
reminder,
if
people
have
more
questions,
just
just
put
them
into
the
queue
a
feature
were
slick
or,
however,
you
want
to
reach
me
so
there's
a
couple
of
things
that
actually
go
when
they
came
in
the
beginning,
when,
even
when
you
explain
future
flagging
a
couple,
questions
on
how
should
be
organized
feature
flags.
Also.
B
A
B
B
Also,
to
add
to
that
is
that
you
have
the
synchronization
issue,
because
if
you
are
doing
a
macro
service
architecture,
you
will
never
be
able
to
guarantee
that
the
state
changes
exactly
the
same
time
in
all
services.
So
if
you
use
the
same
fee
across
a
lot
of
services,
that
would
be
a
nightmare
that
the
coordinates
yeah.
A
B
So
we
we
have
developers
solution
for
us,
so
we
didn't
go
into
the
details
on
this.
You
will
find
more
in
the
slides
or
on
our
web
pages,
but
we
think
that
you
shouldn't
expose
your
a
configuration
directly
to
the
front-end,
so
we
developed
something.
We
called
unleash
proxy
for
this
and
we
created
a
single
page
or
a
JavaScript
client
for
that,
so
that
we
only
expose
the
evaluated
value
for
a
specific
user
to
the
browser,
and
this
is
because
we
want
to
be
as
secure
as
possible
after
expose
intervals.
Okay,.
A
A
Not
so
good,
that's
perfect,
then
the
I'm
sure
there's
also
blog
post
on
that.
But
the
question
is:
what's
the
strategy
behind
keeping
these
flakes
in
my
code?
How
long
should
they
be
kept?
Should
we
clean
it
up,
but
any
best
practice
is
to
keep
it
for,
let's
say
a
sprint
and
then
clean
it
up
again.
So
what's
the
deal
with
that,
I
think.
B
And
this
varies
a
lot
from
team
to
team
and
how
long
it
makes
sense
for
them
to
keep
around.
We
actually
invest
together
a
lot
around
this.
What
so
we
have
access
to
the
future
target
history
for
a
few
hundred
applications
over
almost
five
years,
so
we
investigating
we
found
is
that
50%
of
the
toggles
are
deleted
within
the
first
50
days,
but
there
is
a
long,
long
tail
their
morale
and
we
found
them.
B
What
does
that
mean
and
I
think,
typically,
when
you're
kind
of
satisfied
with
a
new
feature,
and
you
have
enabled
it
for
everyone
and
had
it
there
for
maybe
a
few
days,
you
should
probably
clean
it
off,
but
it
needs
to
be
so
part
of
your
practice
and
tippling.
You
forget
about
it.
You
get
different
priorities
and
you
run
off
to
new
stuff
and
you
just
leave
it
around.
A
But
in
the
end,
I
mean
what
you're,
enabling
with
features
like
this
really
continues.
Experimentations
and
experiments
should
run
a
short
period
of
time,
and
then
you
make
a
decision
completely
get
rid
of
it
or
you
know,
expose
it
to
the
full
world.
Here,
that's
great,
then
you
also
showed
in
the
metrics.
It
seems
you.
You
only
show
at
least
that's
what
you
see
here.
How
often
a
feature
flag
was
evaluated
in
a
positive
or
negative
way.
B
No,
not
in
English
itself,
but
unleash
provide
Matrixyl
Prometheus.
Oh
so
you
could
easily
correlate
at
somewhere
else.
This
is
kind
of
or
stand
on
it
right
now,
at
least
that
we
provide
all
the
data
from
a
leash.
When
was
a
toggle
enable/disable
stuff
like
that
now
you
should
probably
measure
your
business
KPIs
elsewhere.
Mm-Hmm.
A
Okay
and
then
there's
questions.
We
obviously
have
a
large
number
of
people
on
the
line
that
are
always
different.
That
are
also
familiar
with
dynaTrace,
and
the
question
is:
can
we
monitor
feature
flag
usage
in
dynaTrace?
That
means
do
you
already
have
thought
about
defining
rules
in
dynaTrace
so
that
when
dynaTrace
captures
individual
transactions,
we
call
them
view
paths
that
we
actually
know
exactly
which
requests
particular
flag,
enabled.
B
A
B
A
A
B
A
Perfect
then,
there's
another
one:
it's
around
the
captain
workflow-
and
this
may
also
goes
to
your
go-to
you,
so
you
showed
Jurgen
in
your
demo
that
you
enable
the
feature
flag
in
the
UI
and
that
trigger
then
sent
an
event
to
captain.
Is
this
the
intended
way?
I
mean
it's
the
intention
that
you
turn
on
a
feature
flag
through
the
unleash
UI
or
shouldn't
this?
Be
awesome?
What
triggered,
through
the
same
workflow?
That
captain
has
for
deploying
a
new
artifact.
C
Yeah
I
think
there
are
like
both
both
ways,
a
valuable
so
for
the
unleash
you
I
might
be
that
it's
really
not
so
much
tech
person,
maybe
just
want
to
enable
this
feature
flag,
because
they
know
how
to
use
the
UI
and
captain
can
then
still
capture
this
change
or
with
the
captain's
Eli
or
the
captain
rest
api.
You
could
also
trigger
the
change,
so
this
is
what
captain
was
doing
for
the
remediation
part.
We
use,
then
the
unleash
API,
together
with
with
the
captain's
internal
functionalities,
to
call
this
ApS.
C
A
I
mean
I
think
it's
great
to
see
that
both
options
are
possible.
As
you
said,
some
people
live
we're
comfortable
with
your
UI
and
then
they
would
change
it
there
and
then
address
may
just
automate
it
through
the
CLI.
Then
the
other
thing
I
think
you
answered
this
already
earlier.
There
was
a
question:
how
does
cap
map
the
problem
to
turning
off
that
feature
flag,
but
I
think
you
showed
the
remediation
yamo
file,
so
it's.
C
Exactly
the
remediation
yamo
file,
so
it's
basically
a
mapping
between
what
is
the
error
and
what
is
the
problem
and
what
is
the
solution
to
this
problem?
And
maybe
I
can
emphasize
here
again
here
like
between
what?
What
is
the
problem
and
what
what
I
want
to
change?
It
does
not
say
like
in
which
state
is
it
like
for
pre-production
or
is
it
for
production?
It
does
not
say
what,
when
we
talk
about
what
to
do,
but
actually
it's
it's.
How
captain
works,
how
we
kind
of
organizing
our
configuration?
C
We
will
tell
here
it's
in
our
production
environment
and
it's
for
our
project
sock
shop
and
it's
for
our
service
cards.
So
this
is
how
we,
how
we
make
sure
we
have
maximum
reuse
of
this
first,
how
captain
works,
but
also
it
works
in
a
way
because
we
want
to
enable
reuse
of
those
artifacts
of
those
confirmations
you
right
so
there's
there's
no
need
to
write
everything
inside
the
actual
remediation
yamo,
but
you
just
put
it
in
the
right
place
and
captain
knows
what
to
do.
B
This
is
say:
configuration
on
the
client
side,
so
we
actually
do
pole
based
to
makes
it
simple.
We
just
mostly
most
of
the
time,
is
just
three
or
four
headers.
Nothing
changed
how
often
you
polish
that
configuration,
so
you
can
configure
that
as
slow
as
you
like.
The
default
is
every
15
seconds,
but
you
do
whatever
you
like.
B
A
B
So
for
unit
test,
most
of
our
SDKs
have
something
we
call
a
fake
and
leash
or
similar
approaches
where
you
can
actually
just
set
it
state
for
the
different
toggles
for
the
test.
That
makes
it
a
bit
easier
for
the
unit
test,
but
for
integrations
test
and
stuff,
like
that,
the
usual
pattern
that
we
recommend
is
to
test
what
you
expect
most
users
to
see
in
production,
because
the
complexity
starts
to
be
very
high.
If
you're
going
to
test
all
the
different
permutations
mm-hmm,
you
can
of
course
do
that.
A
B
C
C
Let
us
move
on
to
to
the
next
part.
So
what
was
going
on
the
legs?
The
the
last
two
weeks,
actually
a
lot.
We
have
our
a
new
integration
in
for
for
captain
it's
a
new
SL,
a
provider
for
knee
Otis,
so
Henrik
was
Henrik
from
from
Mutis
was
actually
developing
this
it's
a
live
provider
for
for
new
your
load.
Actually,
so
what
can
it
do?
C
It's
basically
and
service
level
indicate
a
provider
that
can
reach
out
to
Neo
load
and
fetch
all
the
indicators
there,
the
metrics
and
give
it
back
to
captain
for
evaluation
within
the
quality
gates.
So
this
one
is
really
based
on
or
like
for
the
captain
quality
gate
use
case
if
you're
using
Neo
load.
For
this
make
sure
you
check
this
out,
it's
a
kind
of
brand
new.
He
just
did
it
a
couple
of
days
ago
and
you
have
all
the
the
instructions
here
how
to
install
it.
It's
basically
just
what
you
want
to
do.
C
Is
you
define
your
credentials
how
to
reach
out
to
new
your
load,
and
then
you
just
want
to
run
the
installer
script,
which
basically
installs
this
integration
into
your
captain.
Cluster
captain
commands
cluster
and
then
you're
good
to
go.
So
this
is
how
you
can
interact
with
new
load
in
terms
of
fetching
all
the
service
level
indicators
from
the
other
next
part
is
we
got
a
shout
out
from
Cho
beta
from.
He
is
one
of
the
founders
of
kubernetes,
which
makes
us
really
proud,
and
he
was
mentioning
captain
on
his
webcast,
CGI
kubernetes.
C
C
He
will
talk
about
a
little
bit
like
he
gives
his
thoughts
on
captain
not
in
too
much
detail,
but
yeah
just
make
sure
you
you
check
this
out,
and
we
are
really
happy
that
we
are
expanding
like
the
the
community
and
we
expanding
that
our
visibility
of
captain
and
yeah
ciaobella
is
one
of
the
founders
and
one
of
the
main
guys
in
the
kubernetes
ecosystem.
I
would
say.
C
A
Maybe
worth
to
mention,
because
ondrea's
was
just
asking
the
question
about
the
SLI
provider:
are
these
things
then
also
going
to
be
shown
in
one
of
the
community
meetings?
Yes,
they
are,
and
also
do
us
a
favor.
If
you
have
your
own,
you
know
extension
or
if
you
have
your
own
project,
we
try
to
keep
an
eye
on
what's
going
on
in
the
community,
but
you
know
a
lot
of
people
are
doing
a
lot
of
things
and
not
always
to
be
here
about
everything.
So
also,
please,
proactively.
A
Let
us
know
when
you're
working
on
something
or
when
you
see
something
well,
as
they
always
say
in
the
states.
When
you
see
something
say
something.
So
that
means,
if
you
see
an
a
plug
in,
if
you
wrote
something,
let
us
know
happy
to
give
you
some
airtime.
Also,
if
things
like
this
blog
post
of
the
YouTube
tutorial,
when
we
don't,
you
know
if
he
sees
and
I
mentioned,
let
us
know
as
well.
This
is
always
great
for
us
to
also
keep
the
collection
of
of
references.
That's
just
no
good
for
us.
Obviously,.
C
So
in
our
captain's
Lac
workspace,
what
we
have
done
is
basically
we
removed
the
number
of,
or
we
reduce
the
number
of
channels
a
little
bit
and
we
have
now
and
I
think
it's
here
yeah.
We
have
now
a
description.
What
each
channel
should
be
about
so
make
sure.
First,
when
you
enter
as
a
new
user,
you
will
get.
You
will
get
anyway
a
welcome
message,
and
hopefully
you
will
find
it
useful
to
to
find
to
find
out
the
right
channels
and
also
here
in
the
channel
section.
C
We
always
make
announcements
and
one
of
the
latest
ones
was
just
how
to
use
the
new
captain.
It's
like
workspace
and
it's
kind
of
easy.
So,
basically,
all
the
questions,
every
kind
of
start
of
a
discussion.
You
can
always
start
in
the
captain
project
and
then
we
either
direct
you
more.
If
it's
a
more
development
issue
or
it's
more
an
integration
issue,
we
might
direct
it
to
another
channel.
C
If
you
really
just
feel
you
need
some
help
and
it's
very
clear
that
you
have
any
question
you
can
always
go
to
the
help
channel
and
also
we
discussed
the
captain
specification
in
our
dedicated
captain's
back
channel.
So
that's
basically
a
rough
overview
on
how
the
new
structure
is.
So
we
reduce
the
number
of
channels
we
moved
most
of
the
people
or
actually
removed
all
the
people
that
have
been
in
previous
sections
that
we
archive
we
moved
into
the
new
ones
that
are
the
replacements.
C
So
if
you
already
are
a
user
of
the
captain's
like
workspace,
no
worries,
you
will
find
yourself
now
in
the
right
channels
and
you
can
start
there.
You
can
just
continue
with
your
discussions.
There
well
yeah
I!
Think
that's
it's!
So
far
in
the
useful
link
sections
we
will
also
put
in
the
links
we
already
put
in
the
links
to
unleash
we'll
also
put
in
the
links
to
the
blog
posts.
C
Maybe
we
can
have
more
questions
here
before
we
do
a
little
bit
of
the
outlook
if
or
I
just
I
just
start
with
outlook.
If
there
are
more
questions
any
you
can
always
drop
me
so
in
the
next
community
meetings.
Actually,
we
want
to
discuss
also
the
captain
enhancements
proposal,
so
we
got
a
couple
of
them.
Let's
take
a
look.
What
is
the
current
one?
C
So
if
you
have
any
Pro,
if
you
have
a
proposal
how
to
enhance
captain,
then
you
can
just
go
to
our
captain
enhancement
proposal,
each
repository
and
you
just
open
their
an
issue
full
requesting
you.
Let
us
know
about
your
proposal
for
enhancement,
so
this
one
is
actually
about
the
service
level
indicators.
How
this
should
be
improved,
then
we
will
have
next
the
next
time.
Let's
make
sure
we
we
will
update
the
dates
if
we,
if
we
move
our
community
meeting,
but
next
time
we
will
have
so
I'd
from
Citrix.
C
He
is
the
original
contributor
to
the
captain.
Slack
chat
pod,
so
you
can
actually,
with
this
chat
bot,
you
can
interact
with
captain
and
you
can
fetch
evaluations
and
results,
and
you
can
start
evaluation
of
your
services
directly
in
slack.
So
this
is
pretty
cool
when
you
want
to
know
how
your
service
is
behaving.
If
the
service
will
pass
the
quality
gate,
you
can,
then
doubt
you
can
then
ask
captain
directly
by
your
slack,
so
that
that's
that's
pretty
cool
I'm
really
looking
forward
to
this
then
March
16th.
C
We
will
have
Christian
from
ERT
talking
about
captain
quality
gates
with
an
integration
in
it
labs,
so
without
not
having
captain
as
like
as
a
full
installation,
but
having
like
the
smaller
installation
with
a
smaller
footprint
of
captain
and
then
using
the
captain
quality
it's
in
get
labs.
This
will
something
Christian
will
talk
about
and
then
Henrik
will
provide
more
information
on
the
SL,
a
provider
from
notice
for
near
load,
end
of
March.
C
A
C
A
C
The
questions
so
thank
you
ever
thanks.
So
much
for
for
being
part
of
this
captain
community
meeting,
I'm
really
looking
forward
to
see
more
from
unleash
and
I'm
really
happy
for
for
this
unleash
integration
with
Captain,
because
I
think
that's
a
very,
very
strong
use
case
and
also
unleash
is
very
easy
to
use
and
captain
is
easy
to
use
it's
just
like.
C
If
you
have
more
feature
toggles,
you
have
a
lot
of
complexity,
and
this
just
makes
it
very
easy
and
like
it's
kind
of
a
safety
net
when,
when
you
want
to
change
your
future
dollars
in
production
and
captain
can
make
sure
that
you
don't
that
you
don't
break
the
great
experience
for
your
end
users.
Thanks
for
having
me
what's.