►
From YouTube: Kubernetes SIG Apps 20190304
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
Welcome
to
kubernetes
cig
apps
for
Monday
March
4th
2019
I'll
start
by
putting
the
meeting
minutes
in
here.
We
actually
have
a
pretty
light
agenda
today.
That's
the
meeting
minutes
of
the
first
thing
hi,
my
name
is
Matt
Farina
I'll
be
chairing
today's
meeting.
I'm,
not
sure
Oh
Adnan
is
also
here.
He
is
one
of
the
co-chairs
and
he'll
be
helping
out
Thank
You
co-host
and
today's
meeting
will
start
off.
A
We've
got
two
announcements:
one
is
in
116,
we're
gonna
start
removing
some
old
components
and
it
looks
like
it's
gonna
start
with
extensions
v1
beta
1.
So
if
you
are
using
those
in
your
kubernetes
resources-
and
here
we
talked
a
lot
about
workloads-
you
shouldn't
you
definitely
shouldn't
I
know
there
are
a
number
of
folks
actually
still
do
create
new
ones
with
that
out
of
habit
and
tooling
that
do
it,
but
as
of
116,
we
plan
on
no
longer
accepting
those
to
the
API.
So
this
is
another
one
of
those
fair
warnings.
A
I
use
the
release
version,
apps
v1
use
things
like
that.
Instead
of
the
very
old
extensions
view,
1
beta
1,
the
second
announcement
is
code.
Freeze
is
this
Thursday
V?
Was
it
the
7th
and
so
that's
for
the
114
release?
So
if
you've
got
something
that
needs
to
be
in
be
aware
of
the
timetable
and
those
are
the
two
announcements,
the
next
thing
that
we
have
up
is
a
demo.
A
C
All
right
yeah,
so
you
guys
will
get
to
hear
from
me
two
weeks
in
a
row.
I'm
Dave
I
shared
a
Cutlass
with
you
last
week,
but
wanted
to
share
another
project
that
I've
been
working
on
with
you.
I
have
another
of
my
colleagues
paying
here
and
I
just
turn
my
camera
towards
you
all
right.
So
we'll
jump
into
sharing
a
little
bit
about
the
project.
B
C
We're
gonna
give
a
quick
demonstration.
Well,
overall,
formats
is
a
application
health
monitoring
service,
specifically
built
for
infrastructure
deployed
in
kubernetes
we're
going
to
go
through
a
pretty
simple
use
case
of
an
example.
You
know
along
the
lines
of
you,
deploying
a
new
service
or
a
new
version
of
a
service,
and
suddenly
there
being
some
error
in
that.
In
this
case,
it's
gonna
be
an
error
that
causes
all
service
responses
to
return
with
a
500
error,
and
certainly
we,
you
know
the
whole
goals.
C
The
goal
of
the
project
is
to
minimize
the
time
to
discovery
and
the
time
to
recovery
for
these
sorts
of
erroneous
states
that
we
can
find
ourselves
in
what
we're
gonna
see
in
the
demo
is
an
the
remediation
strategy
of
automatically
rolling
back,
but
that's
something
that
will
be
configurable
as
well.
It'll
just
work
well
for
this
demo,
let's
see
actually
ping,
do
you
have
a
good
sense
of
what
we
want
to
share
on
this
slide?
Yeah.
B
This
is
why
we
like
a
semester,
Palmas
canina
monitoring,
which
is
connected
a
real
helpful
code
in
some
metrics
around
application,
and
also
we
can
like
analyze
the
logging
and
the
tracing
together,
which
gave
a
visibility
after
like
as
the
application
health
status.
So
we
loved
this
like
say
he
tried
to
turn
the
boat
on,
especially
like
I,
say
during
the
community
department,
and
you
know
maybe
test
and
also
like
an
incremental
department.
All
these
this
case,
like
for
master,
is
here
to
fit
for
all
this
use
case.
A
C
C
This
is
a
component
that
actually
gets
deployed
inside
of
the
kubernetes
cluster,
that
you're
monitoring
and
it
has
the
ability
to
not
only
grab
some
of
the
data
from
the
applications
that
we're
looking
at
like
these
three
down
here
at
the
java
go
and
other
java
application
that
are
in
this
architectural
diagram.
It
also
has
the
ability
to
trigger
remediations,
rollbacks
and
and
modify
the
state
of
the
cluster
yeah.
B
B
B
More
like
thinking
about
committees
as
a
service,
it's
kind
of
essentialized,
you
know
who
misses
you
know:
committees
can
like
a
sitting
under
different
like
a
kubernetes
cluster
clustering
cortex
is
centralized,
one
which
is
like
I
say
so,
like
you
know,
in
order
like
a
current
purpose
so,
and
also
we
also
like
a
support
away
from
and
other
metric
system
as
well.
Okay,.
C
C
And
there
have
some
of
their
dashboards
and
wavefront,
and
so
that's
definitely
an
integration.
That's
gonna
be
important
for
our
project
moving
forward,
so
today
we're
gonna
look
most
directly
at
the
the
dashboard
because
there's
a
UI
component
to
look
at,
but
we'll
talk
a
little
bit
more
about
where
the
brain
and
barrel
men
come
in
throughout
that
this
is
just
a
slightly
more
detailed
view
of
what
demo
we're
going
to
show
here.
C
First,
we're
gonna
start
off
with
a
extension
to
control
the
watch
command
to
tell
the
system
to
start
start
ingesting
data
for
a
particular
application.
From
that
point
on
that
service
is
being
watched,
we'll
go
ahead
and
do
a
deployment
which
will
be
emulated
with
a
KU
control,
replace
with
a
different
file.
That
version.
C
A
C
Just
a
one,
more
quick
overview
of
roughly
what
we're
gonna
see
here
in
a
minute
in
the
demo.
So
let
me
jump
to
that
alright
and
before
you
even
I,
guess
starting
the
demo.
I'll
I'll
give
a
quick
overview
of
what
we've
got
going
on
here.
This
is
a
custom,
build
dashboard
though
it's
you
know
we're
like
I,
said
we're
talking
about
the
wavefront
integration,
though
we
we
definitely
knew.
We
needed
a
at
least
a
little
bit
of
a
custom
built
dashboard
for
to
support
the
open
source
use
cases
as
well.
C
C
So
right
now,
so
the
blue
series
is
the
measured
series,
the
green
series
and
the
green
range.
That's
our
models,
expectation
for
these
data
series,
and
so,
as
we
see
right
now,
everything's
within
the
model
expectations
and
that's
what
we'd
expect
for
the
first
version
we
have
deployed
down
below.
We've
got
this
kind
of
well,
we've
got
a
3d
chart
we,
but
this
is
kind
of
more
just
like
an
open
space
that
we're
playing
around
with
different
things.
C
C
Yeah
I'm
not
sure
what
it's
complaining
about
there,
though
the
majority
of
what
the
watch
command
is
trying
to
do,
is,
is
still
seeming
to
happen.
So
I
should
be
told
this
is
this.
Application
has
already
been
watched.
That's
why
we
have
the
model
data,
but
just
to
show
quickly
how
straightforward
the
process
of
getting
the
watching
started
can
be.
So
beyond
that
we're
going
to
take
a
quick
look
at
the
current
state
of
our
demo.
Application
in
this.
C
B
C
C
A
C
B
B
B
C
Quick,
it's
just
open
this
real
quick.
This
is
just
the
duration
of
time
that
we're
seeing
and
as
I
mentioned
in
the
charts
that
is,
and
as
I
mentioned
a
moment
ago,
I
had
run
this
before
and
so
within
the
last
45
minutes.
So
we're
get
to
see
that
here,
but
it's
obviously
not
as
exciting
as
seeing
it
happen
in
front
of
our
eyes.
We
keep
playing
a
little
bit
more.
C
B
C
C
B
C
C
Back
the
sooner
that
I
tried
before
actually
now
that
I
have
it
see
me
working.
Let
me
flip
back
to
just
the
last
15
minutes,
so
we
get
to
see
the
data
at
a
finer
granularity
that
will
look
a
little
bit
better.
All
right
so
I
had
just
triggered
the
the
version,
2
deployment
and
so
I
think
it
looks
like
it's
starting
because
some
of
our
data
application
metrics
start
to
trail
off.
We
don't
have
the
latest
data
points
from
some
of
those,
but
would
you
see
so?
C
First
off
we've
got
an
annotation
of
the
different
version,
changes
that
this
dashboard
is
able
to
show
us.
We
see
that
the
500
error
count
has
indeed
spiked
we're
still
some
of
the
latency
data.
It
takes
a
little
while
to
come
through.
It
looks
like
there's,
probably
been
a
reversion
version,
remediation
back
to
b1
and
I.
Think
we'll
see
in
a
moment
here
where
the
classification
engine,
the
formats.
C
Here
was
the
anomalous
data
point
that
triggered
the
version
rollback.
So
you
know
that
is
the
the
scenario
I
wanted
to
show
I'm.
Sorry
it
took
a
little
while
to
get
to
that
point,
but
I
think
that's
the
majority.
Let
me
quickly
flip
back
to
the
terminal
just
to
show
as
well.
What
I
didn't
show
was
what
the
version
was.
As
far
as
the
coop
described
employment
was,
it's
probably
not
even
worth
doing.
C
Back
to
version
1,
but
what
we
would
have
seen
is
that
it
had
indeed
flip
diversion
to
and
flip
back
to
version.
1
is
just
kind
of
killed
in
another
way
that
the
demo
is
behaving
as
we'd
expect.
So
that's
the
majority,
the
demo
I'm
gonna
flip
back
to
our
presentation
now
and
open
the
floor
a
little
bit
more.
To
pink
is
this
where
we
want
the
slide,
the
QA
or.
B
Did
you
want
to
go
to
this
one
first
yeah?
So
you
know
you
may
question
about:
what's
the
a
I'm
all
the
way
leverage,
so
this
page
is
telling
you
like
you
know
we
have
a
different
type
of
them
for
different
use
case,
for
example,
for
the
communion.
The
comment:
oh,
we
have
like
all
the
hours
and
the
pier
wise
to
analyze
the
algorithm
on
the
left
side,
but
they're
four-poster,
like
a
department
or
continuously
monetary.
We
use
the
rest
right
side
of
algorithms.
B
Some
of
the
illusion
is
for
like
a
ribbon
array
like
a
time
series.
Data
center
of
the
algorithm
is
for
the
similarity
model
and
if
we
come
with
like
multi-barrel
to
use
case,
which
is
like
weep
with
supporter
for
golden
side
matrix
or
that
for
CPU
memory,
which
is
the
saturation
and
the
latency
error
and.
B
C
C
C
This
is
a
one
that
just
kind
of
at
a
high
level
describes
as
well
the
congestion
of
metrics
logs
and
traces
roughly
where
for
maths,
it's
and
then
some
of
the
remediation
strategies
that
can
be
taken
from
there,
that's
probably
worth
highlighting
there,
and
then
you
know,
of
course
it's
a
I
didn't
mention
it
up
front.
So
this
is
a
fully
open
source
project
that
into
it
is
sponsoring
or
stewarding
I'm,
not
quite
sure
what
the
term
is.
C
B
A
A
C
C
A
No,
no
thank
you.
This
was
great
getting
to
see
this
work.
I
couldn't
think
of
any
real
questions,
cuz
most
of
the
ones
that
I
came
up
with
you
answered
along
the
way
and
some
of
the
other
stuff
I
know
I'll
come
back
to.
It
really
reminds
me
a
little
bit
of
things
like
automated
run
books
with
remediation
based
on
some
machine
learning,
which
I
appreciate
and
the
ability
to
do
things
like
roll
backs
or
find
problems.
That's
entirely
useful,
so
I'm
gonna
go
poke
around
more
at
this.
Thank
you.
Thank.
C
A
A
The
customized
folks
are
adding
generator
plugins
via
a
cap
and
I
know
that
we
are
folks.
You
tend
to
use
the
tools
here
and
dig
in
and
have
opinions
on
how
to
build
them,
and
so
I
wanted
to
point
out
this
cap
that
is
coming
along
and
I'll
share.
My
screen
real
quick,
there's
been
a
bit
of
feedback
on
it,
but
really
this
was
more
just
informational,
sharing
than
anything
else
to
talk
about
it.
A
The
current
path,
they
believe,
is
go
plugins,
but
there
has
been
discussion
about
that
because
it
doesn't
do
things
like
support
Windows.
What
I
would
ask
here
is
folks
who
are
interested
to
go.
Have
the
conversation
over
there
more
than
anything
else,
but
this
is
one
of
the
things
they're
looking
at
customized
secret
generator,
plugins
I,
don't
know
if
anybody
had
any
thoughts
on
this
or
not
I'm,
not
exactly
sure
how
this
plays
into
the
customized.
A
That's
in
kubernetes
SIG's
versus
the
customized
that
was
merged
into
coop
control,
because
there
are
some
differences
there
and
I'm
not
fully
aware
of
the
differences,
but
there
are
folks
who
are
diving
into
this
and
app
developers
wanting
to
dig
in
to
use
customize.
This
applies
to
you.
That's
all
I
really
had
on
this.
Did
anybody
have
anything
they
wanted
to
do
on
this
or
any
form
of
open
discussion?.