►
From YouTube: Kubernetes - AWS Provider - Meeting 20220304
Description
Recording of the AWS Provider subproject meeting held on 20220304
Subproject Updates:
CCM
- new e2e tests https://k8s-testgrid.appspot.com/provider-aws-cloud-provider-aws#e2e
- tagging controller
LBC
- new release v2.4.1
Discussed https://github.com/kubernetes/autoscaler/pull/4588
A
A
Well,
subproject
of
subtract
project
updates
I'll
go
ahead
and
take
the
first
one
for
the
ccm
we
recently
merged
some
new
ede
tests,
so
you
can
check
those
out
and
we'll
be
we'll
be
adding
to
those
as
well
took
advantage
of
cops,
cubetest2
deployer,
currently
just
the
deployer
right
now,
we'd
like
to
get
the
tester
working
there
as
well
needs
a
little
bit
more
work
but
yeah.
I
will
I'm
sure
the
cops
folks
will
hear
from
me
as
that,
as
it
happens,.
A
The
other
thing
for
ccm
is
that
there's
a
new
sort
of
pull,
request
and
issue
proposal
for
a
tagging
controller.
So
this
is
something
that
eks's
customers
have
asked,
for.
They
want
a
way
to
track
cost
allocation
kind
of
things,
and
so
this
will
be
optional,
where
you
can
disable
the
controller
completely.
You
can
also
pass
in
custom
tags,
and
I
think
it's
just
going
to
start
out
with
with
instances
and
then
maybe
volumes
and
load
balancers
as
well
at
some
point.
A
A
C
So
looking
forward
to
that,
and
the
only
other
topic
I
had
was
I'm
wondering
if
the
cluster
auto
scaler
provider
aws
would
be
a
topic
for
this
particular
meeting.
I
mean
I
attend
sick,
auto
scaling,
but
I
don't
necessarily
want
to
like
hijack
the
auto
scaling
meeting
to
talk
about
like
aws
provider
stuff
inside
of
cluster
autoscaler.
All
the
time
I'm
wondering
if
this
would
be
a
more
appropriate
place
to
discuss.
A
A
C
That
work
on
the
like
the
ec2
compute
launch
services,
teams
that
do
pull
requests,
and
I
do
reviews
in
that.
Oh
I
got
it.
Myanna
harris
from
the
nodes
team
has
pushed
up
a
few
pr's.
I
just
I
didn't
know
whether
I
I
don't
I
don't
want
to
like.
Like
I
said,
I
don't
want
to
hijack
the
auto
scaling
meeting
to
talk
about
like
aws
specific
things,
but
I
also
didn't
know
of
of
a
better
sig
to
have
those
types
of
discussions.
C
Well,
there
was
recently
a
pull
request
from
austin
shu
in
the
compute
launch
services
team.
That
added
attribute
based
instance,
selection.
I
don't
know
if
you're
all
familiar
with
this,
but
if
you,
if
you
think
about
like
a
an
asg,
an
ec2,
auto
scaling
group,
you
can
have
a
launch
template
that
has
overrides
in
it
and
those
overrides
can
select
either
a
set
of
instance,
types
that
the
the
asg
should
should
have,
or
should
the
instances
should
belong
to
one
of
those
instance.
C
Instance
types
or
you
can
do
these
things
called
instance
requirements
which
is
like
instead
of
saying
instance
types.
You
say:
okay,
I
want
a
min
or
max
amount
of
vcpus
and
a
min
max
amount
of
memory,
and
you
know
some
gpu
attributes
and
things
like
that,
and
then
the
asg
slash
fleet
api
determines,
which
instance
types
match
those
requirements.
C
So
it's
just
a
different,
just
a
different
way,
a
less
coarse-grained
way
of
launching
instances
right.
So
you
don't
specify
the
instance
type.
Instead,
you
specify
the
fine-grained
resource
requirements
of
that
you're.
Looking
for
anyway,
that
was
recently
added
to
the
provider
aws
inside
cluster,
auto
scaler,
just
yeah,
it
was
a.
It
was
an
interesting
pull
request.
I
thought
it
was
actually
pretty
good,
but.