►
From YouTube: kubernetes aws 20190809
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
Hello,
everybody
today
is
Friday
August
9th.
This
is
the
AWS
meeting.
I
think
we
are
technically
no
longer
a
cig.
We
are
now
a
sub
predictive
signal
provider,
but
this
is
the
sig
ish
AWS
meeting
this
meeting
is
being
recorded,
will
be
put
on
the
Internet,
a
reminder
to
be
mindful
of
our
code
of
conduct
and
be
a
good
person.
I
pasted
a
link
in
the
chat
to
our
agenda.
A
There
are
only
two
items
on
it
right
now,
both
of
which
I
added
and
they
should
be
fairly
quick,
I
think
they're
both
fairly
informational
and
please
do
feel
free
to
add
items
to
the
agenda.
I
feel,
please
also
feel
free
to
add
your
name
to
the
agenda.
It's
helpful
helpful
for
people
watching
the
video
at
a
later
date
to
be
able
to
contact
you
or
something
like
that.
A
A
Or
they
encountered
an
issue
with
the
the
kubernetes
number
of
max
or
limit
on
the
number
of
volumes
you
can
attach
to
an
instance,
the
scheduler
and
the
kubernetes.
A
scheduler
enforces
this
limit
based
on
sort
of
traditional
limits
that
were
basically
constant
across
all
instance.
Types
I
think
it
was
26.
Authors
of
my
head
on
the
newest
a
diverse
instance
types.
A
So
that's
particularly
challenging
the
workaround
that
I
think
people
are
playing
with
for
now
is
basically
to
say.
Well,
you
know
we
can
set
the
volume
limit
to
something
like
10
or
something
like
that
and
that
sort
of
fits
most
of
our
use
cases,
but
it
will
be
problematic
for
people
that
are
pushing
it
a
little
bit
further
like
either
running
with
Atos,
V,
PCC
and
I,
and
running
a
lot
more
pods
or
running
trying
to
run
more
than
workloads
with
volumes
sort
of
at
a
denser
rate
than
whatever
cap
they
pick.
A
So
there
are
some
links
there
to
the
the
kubernetes
issue,
the
cops
issue
and
just
a
link
today,
there's
documentation
that
describes
the
details
of
this
I.
Don't
think
we
have
any
real
answers
on
it
right
now,
but
I
am
trying
to
make
people
aware
of
it
and
hopefully
in
Google
we
will
figure
out
I
think
the
first
thing
is
to
figure
out
where
the
right
place
is
to
solve
this,
because
we
don't
have
any
great
answers.
B
Question
on
that
one
Justin,
okay,
so
would
we
were
kind
of?
We
were
just
kind
of
shooting
ideas
about
the
cni
and
and
trying
to
avoid
putting
that
static
map
of
maxi
and
eyes.
That's
kind
of
you
know
built
into
the
C&I
and
how
we
paid
it
around
that,
and
one
of
the
ideas
was
to
add
instance,
limit
for
Ian's
eyes
to
the
instance
metadata.
B
B
A
That's
a
great
change
that
you're
proposing
to
expose
that
information.
I
think
the
I
think
that
may
be
a
harder
is
going
to
be
figuring
out
the
plumbness
information
into
the
scheduler
and
having
it
having
it,
be
aware
of
the
idea
that
it's
now
dynamic
I
like
the
idea
that
of
like
yeah
putting
it
into
metadata
somewhere.
We
can
also
put
it
into
kubernetes
metadata
to
like
so
I'm
having
it
be
dynamic
without
the
scheduler
or
having
any
so.
Currently,
the
scheduler
has
I
think
I.
A
B
A
B
So
there's
a
the
scheduler
accustomed
resources
allows
you
to
you.
Could
you
can
create
an
IP
resource
right
and
schedule
according
to
that?
It's
kind
of
weird,
because
the
container
actually
has
to
have
the
resource
and
not
the
pot
itself?
So
it's
like
I,
don't
think.
There's
such
thing
as
pod
level
resources,
even
though
IP
as
technically
would
be,
but
is
that,
like
I
see
what
your
see
your
point
to
like
the
actual
issue
here
is
scheduling
not
actually
getting
the
data
on
the
instance
itself,
but
I
mean
getting.
A
A
A
A
I,
don't
know
if
there's
a
better
name
but
like
so
lots
the
this
28
slot
thing
and
the
idea
that,
like
some
some
NYX
well
yeah,
it's
like
some
pods
consume
them
and
some
pods
don't
consume
us
a
lot
based
on
like
whether
they
were
the
first
one
or
not
it
it's
a
difficult
accounting
but
I
think
is
a
little
different
from
the
way
most
of
the
other
accounting
works.
So
like
it's
not
as
it's
not
as
simple
as
like.
A
You
have
a
gig
of
memory,
and
this
one
eats
so
much
of
it
right
or
you
have
a
GPA
right.
It's
it's
a
little
it's
little
weirder,
but
I
do
think
that
that
is
the
likely
the
right
mechanism.
It's
also
yes,
I
think
that
is
likely
the
right
mechanism
how
we
would
actually
achieve
that
I,
don't
know
like.
Would
we
dynamically?
Would
we
don't
Emily
tweak
the
availability
of?
Would
we
forget
about
slots
and
just
dynamically
tweak
the
Lambor
of
volume
attachments
right
or
what
would
we
do
there?
A
Is
there
is
a
PR
or
there's
a
for
a
dynamic
volume
which
I
think
you
pointed
out,
or
someone
pointed
out
in
tell
provider
meeting
last
on
Wednesday
that
is
I'll
put
a
link
to
the
enhancement
about
that
I'm.
So
they're
still
trying
to
dig
into
that
and
find
out
what's
going
on
there,
it
doesn't
look
like
it's
an
exact
match.
I
was
certainly
not
exact
match,
it
doesn't
look
like
it.
It
looks
like
it
any.
A
A
A
You
yeah
I'm,
trying
to
I'm
guessing
cube
scheduler
is
gonna,
have
to
be
the
first
port
of
call
for
this,
its
scheduler
as
the
one
that
enforces
this
limit
and
yeah
I'm
guessing
they.
Hopefully
we
can
get
their
attention
on
it
and
then
get
some
input,
but
I
want
to
make
sure
everyone
in
this
group
was
aware
of
it
from
a
kubernetes
not
applying
necessary
to
correct
limits
on
some
instance
types
anymore.
A
B
A
B
A
A
Yeah
permissions
on
the
docks
like
I
know,
actually
that
I
know
that
okay
I
will
try
to
rename
this
stuff.
I
know
that,
like
I,
like
cops
for
whatever
reason
lost
permissions
on
its
docks,
so
I
presume
that
was
something
to
do
with
the
rename.
But
it
looks
like
I
I,
don't
know
whether
people
have
tried
editing
the
minutes.
Yeah
Nick,
you
put
your
name
in
so
I'm
guessing
you
still
have
access
that
didn't
break.
So
that's
good!
Oh.