►
From YouTube: [wg-lts] Biweekly Meeting for 20230912
Description
[wg-lts] Biweekly Meeting for 20230912
A
A
But
would
appreciate
anybody
Jordan,
thank
you
for
doing
so.
This
meeting
will
be
recorded
and
on
YouTube
because
we
did
get
the
automation
set
up
super
awesome,
so
we
won't
have
to
upload
it
manually
anymore,
that'll,
be
such
a
Time
Saver
and
make
things
pretty
nice
all
right
with
that.
We'll
move
on
to
our
recurring
topics
and
we'll
open
it
up
to
let
anybody
that's
new,
introduce
themselves
if
they'd
like
to
say
hello
and
I.
D
E
Working
at
our
plastic
provisioning
management
platforms,
I
am
also
a
release.
Manager
at
signal
is
doing
a
lot
of
release
stuff.
So
this
is
definitely
interesting
today,
topic
because
I
would
like
to
break
my
experience
to
someone
who
has
been
working
on
a
cluster
management
tools
and
also
someone
who
has
experience
with
how
really
stuff
works,
so
I'm
definitely
happy
to
be
with
all
of
you
there
and
yeah.
That's
all.
G
A
D
A
A
I
A
I
A
A
There
were
a
bunch
of
terms
added,
so
thank
you
to
whoever
added
those
things
as
as
good
background
information.
It
looks
like
the
questions
we
have
so
far
are:
what
is
your
desired
Cadence
for
feature
updates?
Do
you
have
a
change
freeze
window
lasting
longer
than
a
month
where
feature
updates
cannot
be
applied?
If
so,
how
often?
And
how
long
are
those
freeze
Windows?
A
How
long
do
you
require
to
test
new
feature
updates
before
updating
production
environments
and
what's
the
longest
and
typical
delay
between
Upstream
kubernetes
releases
and
all
third-party
integration
points
declaring
compatibility
with
that
version?
Has
anybody
had
a
had
a
chance
to
look
through
this?
Have
any
other
suggestions
or
feedback
or
data
points
that
they're
really
interested
in
in
learning
more
about?
A
J
J
But
I
don't
know
if
we
want
to
capture
like
what
stops
you
from
upgrading
and
maybe
are
you
not
going
to
touch
a
cluster
once
it
exists
and
you're
going
to
migrate
workloads
instead
or
try
to
resist
change
forever?
Like
yeah,
that's
I'm
interested
in
what
the
group
thinks,
but
I've
definitely
encountered
that.
I
Yeah
they
understanding
like
how
they
consume
security
fixes
would
be
helpful,
but
like
do
you
upgrade
at
all,
is
it
maybe
a
good
sort
of
first
question
and
then,
if
you,
if
you
don't
like
what
does
that
mean
for
your
environment,
like
do
you
consume
security,
fixes
like
how
do
you,
how
do
you
sort
of
stay
secure
while
not
upgrading,
and
then
what
happened?
What
do
you
do
at
the
end
of
your
cost
or
lifetime?
Do
you
do
you
migrate?
I
J
You
know
immutable
kind
of
like
our
LTS
sort
of
scenario,
with
just
like
absolutely
essential
security
patches,
but
nothing
else,
and
then
the
people
who
are
fine
with
doing
the
updates
for
a
specific
version
but
believe
or
have
experienced
such
a
breaking
change
between
you
know,
1.21
to
1.22
or
1.22
to
1.23
or
whatever
that
they
refuse
to
try
to
jump
versions
like
they'll.
Do
the
security
patches
as
long
as
they're
available
for
a
version
and
then
they're
like
welp
you're,
not
updating,
1.24,
anymore
I,
guess
I'll
have
to
start
thinking
about
planning.
J
Like
it
there's
I
I'm
interested
in
when,
when
you
say
upgrade,
do
you
mean
apply
the
security
incidental,
like
patch
releases,
in
a
specific
version,
or
do
you
mean
move
to
another
version
because
it
just
patch
releases,
minor
upgrades,
I
guess
kubernetes
has
no
version
two.
So
there's
no
major,
but.
H
L
I
I
Would
you
be
able
to
consume
upgrades
or
or
would
would
that
resolve
the
problems?
You
have
I'm
because
we've
heard
things
about
people
needing
to
certify
and
like
sort
of
regulatory
or,
and
so
people
like
that,
wouldn't
they
would
still
have
issues
even
if
the
new
version
was
perfect
and
didn't
break
a
thing
and
everything
kept
working
like
for
them.
The
Cadence
is
the
issue,
and
the
certification
process
is
the
issue
I
I.
Don't
know
how
to
ask
sort
of
thought
experiment,
questions
in
a
survey
but
I
I
want.
B
M
Yeah
sorry
I
was
like
joining
today,
yeah.
So
like
some
of
the
some
of
the
customers,
we
deal
with
only
one
upgrade
twice
a
year
right,
so
that's
either
one
or
two
kubernetes
jumps
depending
on
what
the
release
cycle
hits.
So
it
just
gets
complicated
if
you
need
to
do
a
step
step
upgrades
every
time
that
that's
that's
the
kind
of
frustrations
we
deal
with
at.
A
F
N
J
K
J
I
Yeah
yeah
I
mean
like
very
concretely
We
Are,
Burning,
Down
The.
The
categories
of
things
that
have
been
disrupted
in
the
past
so
like
removal
of
apis
served
by
default
is
a
thing
we
no
longer
do
after
the
last
Vedic
API
graduates
like
we,
we
don't
serve
apis.
We
know
people
will
depend
on
and
then
we'll
break
in
the
future.
I
The
The
Proposal
that
was
linked
to
the
safer
upgrades
talk
specifically
about
running
servers
in
a
compatibility
mode
that
keeps
all
the
behavior
the
same
as
the
previous
version
so
that
it
like
it
formalizes
that,
and
then
we've
had
some
success
in
metrics
and
things
showing
use
of
stuff.
That's
going
to
change
in
the
next
release,
so
you
can
do
a
pre-upgrade
check,
so
you
can
say
like
look
at
my
metrics
or
all
my
metrics
clean.
Is
there
any
use
of
the
thing
that
changes
in
the
next
release?
I?
I
L
Yeah
I
think
this
right
now
Focus
very
much
on
kubernetes,
which
rightfully
so
but
I
do
wonder
if
there,
if
one
of
the
reasons
why
people
are
moving
is
also
all
the
stuff
they
have
to
test
that
do
have
to
run
on
top
of
kubernetes.
So
I
don't
know.
Maybe
it's
like
a
sub
question
to
an
existing
one,
I'm,
not
sure,
but
it
might
be
good
to
also
understand
that
as
well.
I
Yeah
for
sure,
like
the
visibility,
so
the
the
API
use
metrics
are
a
good
example
of
that,
like
you
had
random
things
running
on
top
of
kubernetes
using
who
knows
what
and
then
kubernetes
upgrades
in
like
a
beta,
API
turns
off
and
so
having
visibility
to
that
centrally
to
say,
like
looking,
we
can
sweep
the
audit
log.
We
can
sweep
the
metrics
and
show
none
of
the
things
you're
running
on
this
cluster,
we're
using
the
apis
that
are
going
to
go
away.
I
I
N
A
N
I
mean
I,
think
I.
Think.
A
very
relevant
thing
that
happened
in
the
last
couple
of
weeks
is
that
in
Cube
128
there
was
a
very
slight
Behavior
change.
That
meant
that
sometimes
init
containers
will
not
finish
before
the
main
containers
start
up
which
completely
hosts
psyllium
on
some
clusters.
So
if
you
try
to
run
versions
of
psyllium
that
weren't
upgraded
to
handle
128,
then
then
sometimes
that
would
just
fail
in
really
weird
ways:
yeah.
I
I
B
N
Even
you
know,
you
know,
because
we
do
not
live
in
a
perfect
world
and
having
a
way
for
you
know,
I'd
say
it
sounds
to
me
like
one
of
the
things
the
bridge
you're
asking
for
is
like
what
we
really
need
is
a
tool
that
you
can
run
in
the
cluster
that
says:
I'm
Co,
here's,
what
version
you're
currently
running?
What
version
do
you
want
to
upgrade
to,
and
you
say
I
want
to
upgrade
what
you
three
versions
and
it
says:
here's
all
the
stuff
that
you
need
to
know.
N
You
know,
and
one
of
the
things
that
you
need
to
know
for
this
couple
of
weeks
would
be
you
know
until
kubernetes
128.2
comes
out
with
the
fix.
You
know.
If
you
try-
and
you
know,
upgrade
your
cluster
to
128
and
you're
running
psyllium,
then
you
might
have
a
bad
time.
You
know
I,
think
that
is
a
pipe
dream,
because
the
the
the
everything
will
move
far
too
fast
for
that
to
be
like
a
doable
thing,
but
something
that
went
some
of
the
way
to
helping
people
like
that.
N
You
know,
like
I
mean
with
it
again
without
wanting
to
use
dirty
words.
You
know
a
long
time
ago,
when
I,
when
I
was
the
sort
of
person
who
did
consultancy
and
I
managed
Windows
small
business
server
for
small
businesses.
That
was
one
of
the
things
that
you
got
as
part
of
small
business.
Server
was
a
thing
that
said:
hey,
you
know
you
want
to
upgrade
to
from
yo
SPS
2008
to
the
next.
N
A
All
right,
quick
time
check,
we
are
just
about
out
of
time
for
this
this
session.
Do
we
want
to
continue
kind
of
iterating
through
this
asynchronously
after
this
meeting?
How
do
we
get
to
a
point
where
we
are
ready
to
say
we
are
ready
to
ask
people
questions
to
learn
more
and
figure
out
if
we're
over
training
or
if
we
are,
if
we
need
to
go
in
another
direction,
to
look
at
new
things.
E
E
H
I
I
think
we
should.
We
should
ask
people
kind
of
more
generally
what
what
is
blocking
them
from
upgrading
and
what
do
they
need
to
do
to
discover
those
things
we
know
about
certain
classes
like
qualification,
but
our
certification-
we
don't
know
if
we
know
about
all
of
them,
and
so
I
mean
some
of
these-
are
things
that
are
just
kind
of
pipe
periods
that
we
can't
necessarily
solve
for
them
like.
B
First
I
think
we
need
some
cut
and
common
Concepts
that
all
people
understand
the
question.
But
if
you
ask
hon
to
upgrade
the
kubernetes
I,
don't
understand
the
when
should
be
the
start
point
to
count
the
upgrade
the
period
and
the
second
in
the
consistent
which
stop
the
upgrading,
it's
not
from
the
technical
Factor.
Sometimes
it's
from
the
economic
Factor
sample.
H
A
A
A
Perfect
we'll
move
out
to
next
week,
then
so
I
think
we've
captured
a
lot
of
pretty
good
stuff.
There's
a
lot
a
lot
of
things
here.
I
think
it
would
be
useful
to
one
figure
out
when
we
want
to
send
this
so
to
kind
of
back
up
from
that,
to
figure
out
what
things
we
can
get
done
in
that
time.
Rita,
let's
give
the
same
thought.
A
I
think
would
be
useful
later
on
in
the
chat
Bridget
said:
if
we
have
the
survey
ready
in
time
for
kubecon,
it
would
be
good
I
think
that's
a
a
good
plus
one
we
can
mention.
We
can
also
do
a
blog
post,
I,
don't
think
we
have
done
a
blog
post
about
starting
the
working
group
back
up,
and
maybe
if
we
have
a
survey
ready
to
go,
we
can
include
that
as
well.
H
I
I
I
One
we're
just
trying
to
get
sort
of
free
form
input.
That's
one
one
option:
if
there's
a
lot
of
directions,
we
think
we
could
go
and
we
just
ask
people.
Would
you
like
to
see
this
improve?
Would
you
like
to
see
this
improved?
Would
you
like
to
see
this
improved
people
will
say
yes
to
everything
like
improve
everything
if
we
make
them
rank
things
or
spend
points
or
like
here
are
10
things,
10
directions
we
could
improve.
I
You
have
three
points
to
spend
or
you
have
a
ranked
order
or
you
some
some
form
that
makes
them
indicate
which
is
more
important
to
them.
That
might
be
helpful.
I
don't
know
mechanically.
If
what
we're
wanting
to
send
a
survey
out
can
do
things
like
that.
But
if
we
just
say
do
you
want
things
to
be
better,
everyone
will
say
yes
to
everything
and
that
won't
really
help
us.
I
I'm
specifically
interested
where
people
who
would
just
plus
one
like
longer
longer
longer
longer
support
existing
stuff
longer
like
what
is
your
plan
at
the
end
of
that
time,
because
four
months
longer,
eight
months
longer,
12
months
longer,
if
there's
no,
if
the
plan
at
the
end
doesn't
change,
then
all
you've
done
is
buy.
Like
a
one-time
reprieve
of.
However
many
months
and.
N
O
I
I
one
just
as
an
aside
one
thing,
I
hear
a
lot
is:
oh
we're
not
planning
to
upgrade
we'll
do
a
cluster
migration,
to
which
I
say
well
that
that's
great,
like
that,
that's
there
are
a
lot
of
things
that
are
good
about
that
approach.
Are
you
doing
that
today,
once
a
year
and
if
you're
not,
then
that's
a
good
place
to
start
that
would
at
least
get
you
annual
upgrades
and
if
you
aren't
doing
that,
then
like
getting
that
solved,
would
improve
your
life
a
lot
right
away
with
the
support
window.
N
I
mean
like
I
honestly,
it
honestly
does
sound
to
me
like,
because
we're
still
a
bit
about
the
exact
directions,
like
maybe
maybe
some
a
bunch
of
free
text
fields
that
are
like
tell
us
about
about
what
you
think
sucks.
You
know
like
I
mean
getting
people
to
complain
about
things
is
easy
and
like
and
like
I
said
like
it
is
less
hard
than
it
used
to
be
to
take
a
bunch
of
identical
forms
with
a
bunch
of
different
free
text,
things
and
pull
some
themes
out
like
now.
N
N
I
I,
just
I
say
that
to
say
to
make
sure
that
I
don't
end
up
to
trying
to
figure
out
how
to
do
that,
because
I
don't
know,
but
but
I
know.
That
is
the
thing
that
people
know
how
to
do.
Yeah
like
honestly,
like.
Maybe
that's
because
we
need
the
we
like.
We
need
to
sort
of
know
what
we
don't
know
like
asking
people
to
give
us
free
text.
Answers
feels
like
the
best
way
to
surface
information
that
we
don't
yeah.
Yeah
I
mean
things
like
yeah.
N
I
think
because
I
think
you're
right
Jordan
that
the
and
the
real,
no
matter
how
we
spin
the
idea
of
looking
at
LTS,
the
real
problem
is
upgrades
like
you're
like
you're,
like
the
It,
doesn't
matter
how
long
the
support
window
is.
If
the
you
know,
if
the
upgrade
is
ridiculously
hard,
yeah
and
yes,
yes,
I,
agree
Bridget
that
the
the
free
text
part
should
be
optional
at
the
end,
you
know
giving
people
a
place
to
rent
if
they
want
is
like.
A
N
I
think
something
like
that
seems
like,
and
that
because
then,
and
also
you're
keeping
the
responses
in
sort
of
a
keeping
them
set,
we
don't
need
to
do
the
sort
of
the
psychometric
testing
thing
of
asking
people
the
same
question
in
seven
different
ways
to
ensure
that
we're
getting
valid
data
right
like
we
don't
care
about
that
yet,
but,
like
you
know,
we
just
want
to
get
indications
rather
than
like
solidly
valid
data.
I
think
at
this
point.
I
F
I
I
A
H
K
H
A
A
It's
a
good
idea:
I
also
pinged
the
prr
folks
to
ask
if
I
could
get
access
to
the
survey
data
that
they've
done
since
2020.
So
there's
a
like
a
pretty
wide
range
of
stuff
that
they've
captured
specifically
around
upgrades
and
things
people
have
encountered
that
might
be
addressed
with
prr,
so
I
think
that
would
also
be
a
another
good
data
point
to
include
when
we're
doing
some
analysis.
F
A
Anybody
has
anything
else
they
would
like
to
add
for
next
week.
Please
add
it
to
the
agenda.
I'll
just
bump
everything
up,
and
then
we
can.
We
can
pick
up
from
there
if
anything
comes
up
and
you'd
like
to
discuss
things
at
Hawker,
asynchronously
feel
free
to
reach
out
on
slack
and
we
can
discuss
there
or
on
the
mailing
list.