►
From YouTube: 20200311 SIG Arch Prod Readiness
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
A
Okay,
so
gender
today
is
to
look
at
some
of
the
survey
results
from
the
survey
we
did
while
back
a
few
other
comments,
I
guess
before
we
get
started.
This
is
our
first
meeting
I.
Think
since
the
the
contributor
summit
and
the
cube
calm
were
postponed,
we
were
gonna,
do
something
in
the
contributor
summit
to
discuss.
A
A
A
Here
are
some
of
the
questions
we've
discussed
in
the
past,
so
I
think
I've
answered
most
of
these,
but
there's
still
a
few,
a
few
more
that
I
haven't
quite
gotten
to
yet,
but
let's
just
go
through
it
and
then
we'll
we'll
discuss
things
as
we
go
and
if
somebody
else
can
take
notes
in
the
agenda
maybe
directly.
That
would
be
helpful
because
I
can't
do
both
at
the
same
time
sure
I'll
type.
Thank
you
so
for
general,
though
the
way
I
guess
it
doesn't
show
you
the
overview
of
this
I'm.
A
C
A
It
disablement,
some
information
on
alpha
enablement
and
some
discussion
of
the
troubleshooting
questions.
Those
are
the
have
used
events
logs
metrics.
How
recently
I
didn't
get
to
the
monitoring
tools-
section,
which
is
another
kind
of
bigger,
bigger
area
that
we
need
to
explore.
But
let's,
let's
go
with
what
we
have
and
we
can
refine
our
questions
as
well.
If
these
aren't
giving
what
we
need
now.
A
A
A
Now
those
are
percentages,
that's
why
so
it
doesn't
give
the
actual
I
don't
have
the
numbers
in
this
chart,
but
since
they're
78
right,
you
can
see
that
20%
of
or
say
30%
of
of,
that
is
what
24,
if
we,
you
know
24
with
more
than
ten
clusters
breaking
that
down
by
node.
So
we
have
two
separate
questions.
A
The
other
thing
we
sort
of
have
discussed
is
that
that's
what
we
want
the
focus
to
be
that
we
don't
think
that
the
googles
and
red-hats
of
the
world
necessarily
will
benefit
from
anything
that
comes
out
of
this
program.
But
really
we
want
to
make
sure
that
the
operators,
the
end
users,
are
benefiting
as
much
from
the
program
as
we
can
so
that's
kind
of
the
focus
of
what
we
want
to
do
anyway.
So
that's
that's
not
bad
that
really.
A
So
any
questions
on
any
of
this
okay.
That
gives
some
context.
So
let's
look
at
the
versions
that
people
are
running
so
the
versions
here.
So
we
asked
a
question
that
was
basically
what
you
know:
what
versions
do
you
have
running
110
and
earlier
up
through
117
and
of
course
it's
not
mutually
exclusive,
so
so
what
I
did
is
map
those
into
the
oldest
version
running
for
a
given
respondent
and
the
newest
version
running
for
a
given
respondent,
which
gives
us
a
little
more
more.
A
Look
sort
of
mid-tier
here-
let's
say
these
are
these
are
an
order
based
on
the
counts,
have.
D
C
D
A
Version
so
basically
what
old
versions
do
you
have
lingering
around
and
you
haven't
Paul,
you
don't
pulled
up.
There's
a
surprising
10%
folks
have
a
have
one
that
ten
and
look
at
this
in
the
this
is
surprising
right.
People
have
this
older
version
completely
out
of
support.
That's
like
a
third
of
of
the
people
with
between
a
hundred
and
a
thousand,
and
then
it
jumps
up
to
113
and
115
I.
Don't
seems
a
little
surprising
to
me
that
that's
because
that's
pretty
old
version
but
I
guess
people
just
don't
want
to
move.
A
A
D
A
A
C
A
D
A
A
A
A
A
Same
pattern,
but
yes,
they
did
change,
not
that
many
people
have
rolled
back
in
production
and
the
large
ones
here.
These
are
the
two
that
came
out
of
that.
That's
why
you
see
that,
but
there's
the
count.
The
absolute
count
of
those
is
very
low,
which
is
why
it
looks
more
dramatic
in
this
one,
where
it's
it's.
A
percent
of
of
that
category
versus
a
percent
of
the
whole.
A
And
then
based
a
number
of
clusters
under
management
when
you've
got
more
than
a
thousand
clusters.
You've
done
it,
although
it's
not
that
many
people
again.
So
here
are
the
reasons
the
reasons
of
those
who
roll
back.
Then
these
are
the
reasons
Farah
by
far
and
away
individual
components
and
some
components
were
failing.
The
complete
the
cluster
didn't
completely
go
down.
It
wasn't
like
a
cascading.
The
way
I
interpret
the
question,
and
hopefully
other
people
did.
A
The
people
answering
it
did
is
that
basically
complete
cluster
failure
means
like
there's
some
sort
of
cascading
failure
that
just
like
causes
massive
you
know,
takes
the
whole
thing
out
as
opposed
to
like.
Oh
my,
you
know
this
particular
I.
Don't
know
it's,
you
know
the
scheduler
or
this
particular
controller,
or
this
particular
you
know
if
it's
the
API,
so
ever
that
probably
takes
the
whole
cluster
out,
but
other
things
thrown
away.
That's
the
that's
the
biggest
reason,
I
think
what
that
means
for
us
is.
A
B
A
B
A
Don't
know
that's
okay,
another
problem,
I
just
want
to
make
sure
you,
you
didn't
think
you
were
off
you,
so
the
we
probably
want
to
go
talk
to
people
and
see
whether
it's
in
what
specific
component
state
they
were
talking
about.
It
wasn't
like
say,
update
version,
skew
between
components,
the
cost
of
failure
like
what
was
the
you
know.
We
need
to
investigate
that
more
detail.
I
guess,
but
it's.
A
A
D
A
A
A
A
C
A
So
what
are
the
problems
we
have
with
them
problems
same
problem,
which
is
interesting
and
what
yeah
that's
patch,
this
patch
yeah,
that's
weird
like
that
is
weird.
That's
another
one
I
think
we've
got
actually
we
could
put
on
here
a
metric
that
looks
at
that,
compares
that
comparing
the
minor
in
patch
to
each
other
right
now,
I'm
looking
at
at
broken
down
my
clusters
and
notes,
and
but
compare
them
to
each
other,
because
that
isn't
interesting,
I.
A
B
A
This
can
becomes
something
we
go
and
we
talk
to
folks-
and
this
is
the
first
time
we're
doing
a
survey
so
we're
we're
in
some
ways
discovering
better
questions
to
ask.
So
we
need
to
understand
this.
Maybe
is
too
broad
a
category
to
maybe
we
you
know,
maybe
it
maybe
why
it's
it's
it's
it's
like
this
many
times,
but
but
but
yeah
understanding.
Why
hatch
just
roll
back
more
often
I
mean
there
is
something
to
that.
It's
an
operation,
that's
done
more
often,
operation.
I
think
I'm,
not
sure
how
to
get
at
that.
A
A
I
checked
them
multiple
times,
but
they
were
identical.
The
the
allowing
beta
and
allowing
all
GA,
almost
everybody,
does
and.
A
A
A
A
C
A
A
A
D
A
A
C
C
D
C
A
D
A
This
is
disabling
a
particular
feature
in
GA
which,
like
you
said
it's
hard
to
do.
I
think
that
a
better
measure
of
that
question
is
roll
back
and
the
fact
that
this
many
people
have
been
rolling
back
again,
though,
these
are
small,
they're
round
round
percentages,
so
there's
probably
we
could
look
at
absolute
numbers.
I
could
probably
should
probably
put
those
on
at
some
point.
A
A
A
A
D
A
A
Okay,
now
slicing
and
dicing
the
same
information
cuz
you
you
were
wondering
you
mentioned
David
about
the
people
who
run
newer
versions.
Are
they
more
likely
or
less
likely
to
disable
or
to
allow?
And
actually
this
is
pretty
pretty
nice-
the
people
with
the
oldest
version
of
since
113,
all
of
them
allow
allow
algae
in
beta.
Whereas,
prior
to
that,
you
know
there
were
a
fair
number
who
limited
that
this
is
a
little
bit.
I
would
call.
A
Wishy-Washy
correlation
with
it.
But
but
because.
D
A
If
we,
but
what
this
is
saying
is
if
we
exclude
versions
112
and
earlier
people
who
are
running
versions
112
and
earlier,
they
all
say
that
they
allow
yeah,
which
is
which
is
meaningful,
I
think
yeah
for
those
who
have
disabled
a
beta
feature.
A
A
A
A
A
A
You
know
what
it
might
mean.
It
might
mean
that
the
people
who
newest
version
I
117
at
the
time
we
ran
this
survey
and
the
people
who
aggressively
go
to
the
next
version,
and
so
they've
had
to
do
all
this
stuff,
because
they've
been
doing
for
the
last
three
years,
they've
been
aggressively
going
to
every
new
version.
That's.
A
A
D
A
And
merge
it
should
be
graduation
telling
me
yeah,
it
should
be
which
we
well.
It
should
be
graduation
through
through
stages,
not
just
beta
right,
not
just
beta
decay
like
it
needs
to
move
from
alpha.
You
know,
I
think
we
talk.
Maybe
we
talked
about
that
on
the
on
the
in
the
comment,
so
it
needs
to
move.
A
One
thing
I
was
thinking
about
with
respect
to
that
is
you
know
we,
we
probably
should
provide
some
tools
to
help
teams.
Do
that,
so
that
might
mean
like
we're
talking
about
revising
a
lot
of
the
cap
functionality
and
if
we
need,
if
the
cap
is
known
to
be
in
alpha
the
feature
in
alpha,
there
may
be
some
structured
ways.
We
can
help
people
gather
data
through
these
forms,
or
whatever
is
right
now
like.
A
What
do
you
do
you
you
put
it
in
outside,
you
put
it
out
there
and
you
hope
somebody
like
contacts
here
right
and
that's
very
passive
way
to
gather
data,
so
I'm
trying
to
think
about
it.
Whether
there's
a
way
that
we
as
a
you
know
as
a
an
organization
could
offer
some
things
to
prompt
people
or
so,
even
even
just
like
a
here's,
how
you
should
collect
feedback,
you
know
and
give
some
agents
there.
I
think
that
might
help
anyway,
tangent.
A
Okay,
so
this
is
troubleshooting
broken
down
the
broken
down
by
the
number
of
nodes
under
management.
So
basically
the
way
this
question
is
I
asked
near
the
top.
Have
you
used
Prometheus
metrics
in
the
last
month,
quarter
year,
more
than
a
year
or
never,
and
then
some
people
didn't
answer
to
watch
is
where
the
novel
comes
from.
So
I.
A
Don't
find
that
I
find
it
I
guess
a
little
troubling,
but
that
there's
that
there's
so
many
with
never
using
metrics
but
on
the
other
hand,
like
I'm,
not
surprised
that
you
know
as
you
move
to
the
right
in
this
chart,
it
goes
down.
So
in
that
sense,
it's
not
that
surprising.
But
to
me
the
the
number
of
people
who
actually
are
using
logs
is
kind
of
surprising
and.
A
B
A
Yeah
I
might
give
us
more
and
we
can
actually
also
we
can
filter
out
certain
role
like
if
we
want
to
say:
okay,
let's
just
let
you
know
it's
filter
out,
DAV's,
for
example,
dams
are
gonna,
have
a
really
different
workflow,
then,
because
they're
working
on
some
specific
thing,
then
operators
or
even
DevOps
but
like
can
we
look
at
just
operations.
You
know,
then
we
might
see
something
more
like
what
we're
expecting
yeah.
A
D
A
C
A
D
A
D
A
A
Yes,
a
bleeb,
a
this
rare
scary
number
of
people
enable
alpha
in
production.
That's
something
I'm,
not
sure
what
to
do
about.
C
A
B
D
A
But
you
know
you
can
just
put
it
in
the
words
you
can
just
say
you
know
inversion
this.
Did
you
need
immediate
student
version?
This
did
you
do
any
of
these?
Like
we
talked
about
with
roll
back,
did
you
roll
back
117
to
116?
Did
you
roll
back
116
to
115?
Did
you
roll
back?
You
know
we
can
ask
those
questions.
Yeah.
D
A
C
A
A
D
A
I'll
check
it
out
after
I'm,
actually
my
battery's
running
out
so
oh,
do
we
have
anything
else
we
want
to
discuss
today
or
shall
we
call
it
five
minutes
early,
I'm,
good,
you
good
everybody
else,
good,
okay,
all
right!
Well!
Thank
you!
Everybody
I
think
we
got
some
useful
information
out
of
this
I
think
we,
you
know
we
we
can
do
better
with
our
next
survey
and
we
can
also
salvage
this
to
some
extent
by
going
back
and
actually
talking
to
the
people
and
so
maybe
put
together
a
plan
for
that
too.