►
From YouTube: 2020-08-25 Rook Community Meeting
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
All
right
the
recording
has
started-
and
this
is
the
august
25th
2020
rook
community
meeting.
We
will
all,
as
always
start
with
the
milestone
checkups.
So
we've
got
a
couple
of
patch
releases
that
are
intending
to
go
out
today
and
tomorrow.
Travis
do
you
want
to
catch
us
up
on
those
starting
with
1.3.
B
A
C
B
And
then
for
1.4
we're
coming
up.
Well,
maybe
today's
two
weeks
already
or
was
it
that
thursday
last
time
anyway,
it's
about
time
to
refresh
for
1.4
as
well.
We
do
have
some
important
fixes
there,
there's
one
more.
I
I
do
want
to
understand,
though,
before
releasing
that
one
in
the
blocking
release
column
with
nfs
ganesha,
not
working
in
1.4.
B
A
B
It
has,
has
anyone
been
able
to
reproduce.
C
A
Okay,
then
so
at
least
we'll
look
into
that
at
least
understand
the
scope
or
the
severity
or
the
likelihood
of
it
impacting
others,
etc.
Before
we
move
forward
with
1.4
release.
B
Exactly
and
I'd
really
like
to
get
the
release
out
by
tomorrow,
either
way,
at
least
if
we
don't
have
a
fix
or
if
it
takes
longer
to
get
the
fix.
We
can
potentially
get
that
in
the
next
release,
because
that
feature
is
less
than
or
not
as
common
as
others,
but
yep
I'll
I'll,
confirm
or
the
severity
of
the
issue
before
the
release.
E
Yeah,
I
think
I
addressed
your
last
comments
as
well.
It's
just
running
one
ci
now
yeah,
I
add
a
deep
separate
label
and
annotations
for
the
prepared
job
I
just
went
down,
so
I
had
to
and
edit
it
for
the
annotations.
A
Sounds
good
alex
thank
you,
okay
and
then
1.5
we're
still
working
towards
that
goal.
The
time
frame
of
kubecon
north
america,
which
virtual
event,
which
is,
I
think,
sometime
in
november
right
now.
The
final
excuse,
the
final
schedule
for
that
of
the
released
yet
like
which
days
it
is.
A
Yeah
talks
have
not
been
announced
yet
because
I
I
haven't,
got
any
rejection
notices
for
the
ones
that
I
submitted
so
right.
I
don't
think
the
talks
have
been
announced
november
17
to
20.
It
looks
like
right.
B
A
E
I
have
one
one
small
thing:
just
realized
that
41.41
release,
maybe
now
for
the
1.42
release.
Maybe
if
I'm
able
to
get
a
fix
for
at
least
like
a
partial
fix
for
that
in
travis.
If
you
remember,
like
the
611
operator
iterating
over
all
nodes,
instead
of
just
the
osd
placement
valid
nodes,
and
my
idea
would
be
to
as
I
wrote
it
to
as
a
first
kind
of
quick
fix,
just
reducing
the
just
to
reduce
the
amount
of
nodes
it
has
to
iterate
over
by
switching.
E
E
E
A
Oops
cool
thanks
all
right,
great
nice.
All
right,
then
we
could
move
on
to
the
community
topics
and
questions
section
I
believe
so
graduation.
We
are
now
up
to
four
votes,
which
is
all
trending
in
the
right
direction.
There
are
no
that
I'm
aware
of
negative
or
minus
one
votes
all
so.
100
of
the
votes
coming
in
from
the
coc
have
been
affirmative.
A
So
that
is
good
at
least
to
continue
that
trajectory
and
no
other
update
on
that
that
just
continue
waiting
for
the
doc
to
finalize
or
complete
their
voting,
no
follow-ups.
A
All
right
so
kubecon
was
last
week
that
was
fun
and
interesting.
I
suppose
thanks
to
everybody
who
helped
out
with
talks
or
the
booth
or
chat
and
slack
and
all
that
sort
of
stuff,
it
was
great
to
get
some
conversations
going
and
have
some
people
there
in
the
intro
section
that
alexander
and
I
did
there
are
a
fair
amount
of
questions,
so
people
seem
to
come
prepared
with
some
questions
that
they
wanted
to
have
answered
and
some
other
questions
too
around.
A
You
know
what
is
the
value
or
like
messaging
for
rook
like
how
is
it
different
from
like
other
storage
systems,
and
it
seems
to
be
that
there's
a
somewhat
of
a
lack
of
understanding
about
rook's,
orchestration
focus
and
benefits
versus
you
know,
being
a
storage
system
itself.
So
there's
some
clarity
that
was,
you
know,
needed
around
those
things
alex.
Do
you
have
any
other
thoughts
about
the
intro
session
or
seb
seven
travis
any
thoughts
from
your
deep
dive
session.
E
Only
for
the
intro
part
energy,
if
you've
just
mentioned
it,
but
like
disaster
recovery
and
backup
restore,
were
big
topics.
Yeah.
A
E
And
I
can.
I
can
just
also
confirm
that
at
the
new
project,
I'm
working
on
that
it's
basically
the
same
like
a
project
where
we
implement
rooksef
at
the
company
and
there's
the
same
like
today.
We
were
like
yeah
and
how
what
what
would
happen
if
kubernetes
well
breaks
and
we
need
to
reinstall,
would
it
just
pick
everything
up
and
like
well
yeah?
A
I
don't
know
alex:
what's
up,
what's
the
status
or
status
of
our
documentation
or
a
guide
for
like
disaster
recovery
to
we
have
come
back
from
that
we
have.
E
We
we're
gonna
find
out,
probably
till
the
end
of
the
week,
we're
gonna.
Try
that
one
out,
oh.
A
E
A
Test
environment
yeah
a
trial
run
as
it
were,
yeah
that'd
be
awesome.
I
mean
that's
it
like.
That's
such
a
good
thing
to
do.
You
know
like
with
philosophy,
like
chaos,
engineering
and
you
know
be
prepared
for
emergencies
before
they
happen
type
of
thing.
So
the
lessons
from
that
or
any
issues
you
run
into
will
be
fantastic
to
like
either
document
and
have
you
know,
guidance
around
how
to
work
past
those
things
or,
if
there's
fixes
that
need
to
go
in
the
code
base.
B
B
B
Yeah,
we've
also
had
a
long-standing
work
item
around
disaster
recovery
to
simplify
the
steps
for
disaster
recovery,
because
if
you
actually
lose
your
whole
kubernetes
cluster
and
lcd
and
everything
right
now,
our
steps
walk
you
through
how
to
do
it.
But
it
is
a
lot
of
tedious
steps
to
recover
and
we
can
do
things
to
simplify
that.
So
that's
that's
still
on
my
radar
and
yeah
alexander.
Let's
definitely
chat
to
see
how
your
experience
goes.
A
Yeah
sounds
good.
You
can
throw
that
in
there.
What
were
folks
experiences
with
the
the
virtual
booth
and
and
like
the
virtual
experience
of
connecting
with
attendees
and
stuff?
Did
anyone
have
a
positive
experience
with
that?
They
thought
it
was
a
good
platform
and
great
for
that.
B
B
We
only
had
like
three
and
a
half
or
three
minutes
at
the
end
for
verbal
q
a
so
it
was
really
short
that
way,
but
we
we
only
got
about
a
dozen
questions
and
they
were
all
really
good
questions,
but
we
answered
them
all
in
chat
and
then
talked
to
a
couple
of
them
in
the
in
the
verbal
q.
A
but
overall
I
mean
it
was
an
interesting
experience
not
near
as
as
good
as
being
in
person,
but
it
was.
It
was
something
at
least
said
what
you
want
to
add.
C
Yeah
not
much
stuff,
I
don't
think
it
was.
Obviously
it
wasn't
as
pleasant
as
being
like
on
site,
but
we
got
like
over
110
people
at
a
at
the
peak
of
the
presentation.
It
was
good,
I
think,
decent
enough,
almost
as
as
what
we
get,
I
think,
maybe
a
little
bit
less,
but
it
was
good.
C
We
got
questions.
We
got
like
twice
of
third
time
questions
over.
How
can
they
migrate?
How
can
this,
how
can
they
start
using
rook
by
consuming
existing
clusters?
How
can
basically
adopt
existing
clusters
and
having
them
managed
by
rook?
So
it
was
also
interesting
yeah
but
didn't
make
any
talks.
Well,
I
didn't
make
it
into
any
talks.
A
Yeah
yeah,
I
think,
regarding
questions.
I
think
that's
something
that
alex
and
I
kind
of
learned
too,
and
then
this
I
was.
I
was
moderating
a
number
of
sessions
as
well,
and
so
the
lesson
from
that
is
that
leaving
more
time
for
questions
to
go
over,
questions
live
with
the
video
q.
A
would
be
a
better
approach,
but
it
sounds
like
both
of
our
recordings
in
the
35
minute
session
were
like
over
30
minutes,
so
leaving
barely
any
time
to
do
questions
at
the
end.
A
I
think
there's
a
lot
of
value
that
comes
from
that
too.
So
I
think,
maybe
next
time.
Well,
I'm
going
to
stuff
the
intro
in
a
deep
dive
into
a
single
session.
Maybe
maybe
it
won't
be
very
easy,
but
shorter
recordings
and
leaving
more
time
for
q,
a
seems
to
be
good
and
I
think,
there's
more
engagement
on
q,
a
as
well
from
the
audience,
and
the
virtual
virtual
talk
experience,
because
it's
a
lower
barrier
to
entry
that
people
just
type
a
question
they
don't
have
to
like.
B
A
A
So
yeah
it
sounds
like
yeah.
We
should
probably
take
that
into
account
and
do
a
shorter
recording
for
north
america
so
that
we
could
have
time
for
q,
about,
like
video
q,
a.
E
Now,
to
be
honest,
like
I'm,
I'm
not
sure
even
the
people
getting
the
response,
even
if
they
leave
the
the
the
talk,
because
I
think
some
people
had
like
a
green
or
something
icon
shown
showing
all
the
time
and
some
people
had
the
green
icon
and
then
well.
It
went
away
or
something
do
you
know
if
the
people
got
like
the
response,
even
if
they
left
the
talk.
A
B
A
Yeah
I
mean
that
experience
is
terrible,
really
because,
like
there's
the
the
ability
to
actually
have
an
interaction
with
somebody
who
comes
to
the
booth
is
very
minimal,
because
it's
not
like
in
person.
You
know
they
walk
up
they're.
Looking
at
the
stuff
like
they're
right
there
in
front
of
your
face,
you
can
say
hi
and
then
like
it's
fairly
rude.
If
they
don't
reply
back
to
you
right,
but
in
a
virtual
thing
like
you,
send
it
like
the
only
option
really
that
there
is
is
to
send
a
chat
request
to
them.
A
So
you
know
say:
hey
how's
it
going
have
you
heard
of
rook
whatever
and
but
a
lot
of
the
times
it
just
says
you
know,
person
is
chat,
is
waiting
for
a
person
to
accept
your
request.
So
a
lot
of
people
are
just
like.
Oh
no,
I
don't
want
to
actually
talk
to
somebody
and
then
so
you
don't
get
very
many
engagements
at
all.
A
Yeah
tried
it
didn't
didn't,
have
didn't,
have
success,
but
I
say
but
but
though
in
in
hindsight,
though,
that
speaks
to
the
the
necessity
to
have
rich
content
for
the
booth
itself,
because
people
aren't
going
to
be
engaging
a
lot
with
you
and
getting
its
conversations
so
at
the
booth
we
need
to
have
like
the
tabs
filled
out
with
like
good
links
to
good.
You
know
to
places
like
you
know,
like
a
q,
a
or
questions
like
facts.
A
Yeah,
did
we
see,
do
we
can
look
in
us
in
a
bit,
but
do
we
have
the
like
an
overall
number
of
slack
members
on
the
on
the
general
rook
slack
increase
last
week,
much.
A
That's
like
as
a
goal.
Overall,
you
want
the
overall
community
to
grow
right,
and
so
people
that
are
engaging
just
during
kubecon.
That's
one
thing,
but
people
that
join
the
rook
slack
and
like
actually
become
part
of
the
community
to
do
to
persist
their
participation
beyond
kubecon
is
the
the
real
goal
there.
I
guess
so
seeing
if
the
slack
members
went
up
would
be
useful
to
see.
A
A
A
Don't
know
when
okay
cool
so
google
summer
of
code,
this
is
the
final
week
at
the
this
week.
Ahmad
is
intending
to
submit
his
summary
of
the
work
from
the
summer
and
then
next
week,
our
final
evaluations.
A
Let
me
write
a
note
for
that.
Ahmad.
I
think
you
are
on
the
call.
Do
you
want
to
give
us
a
quick
update
about
what
you're
working
on.
D
D
A
Yep,
that's
this
pull
request
here
for
the
design
of
implementing
a
quota
in
nfs
and
yeah.
We
got
some
comments
on
it
and
we
will
we'll
follow
up
to
get
that
feedback
incorporated-
and
you
know
finish
off
this
design,
and
so
yeah
as
the
mod
was
saying
here
is
the
we
have
the
link
in
the
agenda
document
here
for
ahmad's.
Like
summary
of
the
final
work,
final
submission-
and
this
will
be
used
as
an
input
to
the
final
evaluation
as
well
so
yeah,
it's
the
summer,
went
by
really
fast.
A
Okay
and
then
so
yeah,
so
we'll
do
the
evaluation
next
week,
and
so
I've
asked
ashish
and
rohan
as
well
to
make
supporting
ahmad
this
week
and
the
final
week
of
google
summer
of
code
a
priority
so
that
anything
that
a
mod
needs
to
get
pull
requests
in
or
get
feedback
on
things
that
we're
prioritizing
that
during
his
final
week,
so
that
he
can
wrap
up
new
signs
and
get
to
the
finish
line
on
things.
A
And
then
the
evaluation
next
week
will
will
go
great.
I
am
confident
okay,
so
there
is
a
agenda
item
around
ci
issues
as
well.
B
Yep
yeah
satoru.
I
brought
this
up
in
slack
yesterday
and
you
know
I
definitely
agree
with
him
that
cia
is
a
continuing
issue,
so
some
questions
he
brought
up
that
I
want
to
just
point
out
in
reverse
order.
Maybe
first
of
all
I
mean
we
have
an
ongoing
issue
with
the
reliability
of
the
tests,
so
intermittent
tests
are
failing
too
often
and
then.
B
Like
an
intermittent
test,
so
we
merge
and
then
occasionally
master
gets
broken
because
it's
not
actually
intermittent
sometimes,
but
it
looks
intermittent
anyway.
So
that's
I
don't
have
a
good
solution
to
that.
Other
than
that.
We
have
to
continue
to
be
diligent
in
tracking
down
intermittent
test
failures
and
that's
definitely
one
of
my
priorities
and
we
need
the
help
of
everyone
to
make
sure
the
quality
stays
high
for
the
ci
reliability,
which
also
translates
to,
of
course,
better
product
and
release
reliability.
A
Yeah,
I
just
think
that
that's
a
really
difficult
problem.
You
know
travis,
as
you
know
that
you
know,
even
though
we
can
as
a
human
identify
which
ones
are
intermittent
or
which
ones
that
we've
you
know
seen
from
time
to
time,
and
then
it's
not
a
new
issue.
It's
not
a
regression,
so
to
speak
that
it's!
It's
still,
you
know
very
difficult
to
like.
I
isolate,
identify,
isolate
and
fix
the
causes
of
the
intermittent
behavior.
So
it's
it's
nothing
very
challenging.
You
know
that's.
B
Exactly
it's
an
ongoing
thing.
Every
time
we
change
code,
we
risk
new
intermittent
issues
too,
so
they're
not
staying
constant
anyway
and
then
the
other
topic
is
really
around.
How
can
we
speed
up
the
ci
pr
builds
are
taking
about
an
hour
in
general
right
now
and
the
one
thing
that
so
sebastian
opened
a
pr
still
in
progress.
Do
you
want
to
talk
to
that?
One
of
the
improvements
you're
looking
at.
C
C
Then
I
started
looking
into
that
too,
and
just
because
every
new
pr,
every
new
world
is
a
fresh
one
and
nothing
gets
cached
or
anything
like
that,
but
also
what
I.
What
I
realized
is
that,
even
though
I'm
sending
a
pr
against
ceph
operator,
the
ci
is
building
all
the
other
backends
as
well,
which
is
not
necessary.
C
So
I've
started
looking
at
that,
which
should
definitely
improve
the
time
we
spend
on
building
images
first
before
testing
the
pipeline
with
that
image.
But
still
I
need
to
go
back
to
it.
Also
so
this
comes
with
different
flavors,
one
is
only
building
a
specific
architecture
and
the
others
is
just
building
a
specific
operator.
A
And
we
have
support
for
the
for
that
in
the
make
file
already
right.
That
would
just
be
like.
C
A
A
C
So
because
of
the
jenkins
filed,
yep
yeah.
A
Cool
yeah,
I
think
those
are
great
investments
sebastian.
You
know
like
the
headaches
that
are
caused
from
you,
know:
ci
cd,
taking
a
long
time,
and
you
know,
and
having
flakiness
etc,
is
like
definitely
something
that
causes
pain.
So
investments
in
that
is
really
appreciated.
B
A
Yeah
yeah,
I
think
I
think,
honestly
too
travis.
I
think
that
the
majority
of
the
cost
it
comes
from
maybe
ebs
volumes.
It's
just
storage
related
costs.
I
don't
think
I
don't
think
we
should
actually
follow
up
and
look
at
the
cost
breakdown,
but
I
don't
think
that
the
cost
majority
of
it
at
least
comes
from
instances
or
compute
time.
I
think
it's
storage
related
costs.
So,
okay,
let
me
let
me
take
a
look
at
that
and
see.
Do
you
know
which
specific
instance
you
would
want
to
use?
A
A
Me
look
I'll,
go
ahead
and
fire
up
the
cost
analysis
too
just
to
get
that
ready
and
stuff.
B
Yeah
yeah
I
mean
basically,
the
topic
is
so
for
the
I've
been
doing
most
of
the
releases
myself.
Blaine
has
done
some
in
the
past.
It's
been
a
while
since
he's
been
involved
either,
though
so
as
steph,
and
I
were
talking
about
how
to
have
him
start
doing
that,
but
I'd
like
to
really
spread
it
even
beyond.
B
You
know.
If
we
had
three
or
four
of
us
where
we
could
rotate,
we
can
share
the
load,
and
I
think
that
would
be
nice,
so
I'm
gonna,
I'm
gonna,
go
ahead
with
the
releases
this
week,
though,
and
while
I'm
doing
them,
I'm
gonna
create
more
detailed
instructions
with
screenshots
and
created
google
doc
and
things
I
think
said
for
the
idea
just
so
we
could
make
it
easier
to
to
share
that
load
when
we
don't
do
releases.
B
Already,
of
course,
we
could
do
better,
but
I
think
we
do
pretty
well
about
that.
The
we
haven't
had
a
lot
in
the
pending
release.
Notes
honestly,
because
we've
had
more
of
a
philosophy
of
oh.
This
is
just
a
small
feature,
let's
backport
it
to
the
previous
release,
and
then
it's
not
in
the
pending
release.
Notes
for
the
because
we
only
use
pending
release
notes
for
the
next
minor
release,
not
the
patch
releases.
A
A
B
B
Yeah
and
then
in
the
patch
release
notes
we
always
put
individual
pr's
in
the
list,
so
you
see
all
those
features
anyway,
so
yeah.
I
think
it
would
be
nice
to
automate
how
we
generate
those
release,
notes,
which
means
we
need
to
make
sure
pr.
Titles
are
descriptive
enough
for
that
yeah
sub.
You
found
a
project
that
generated
it
from
the
pr
titles
in
the
release.
That
sounds
pretty
cool.
I
forget
which
project
the
control
run.
B
A
Okay,
so
that
is
the
end,
then,
of
the
agenda
items
or
in
the
dock,
as
anybody
have
any
other
further
topics
for
discussion
or
agenda
items
that
did
not
make
it
into
the.
A
Dock
all
right,
then,
we
can
go
ahead
and
wrap
up
for
the
week
and
we
will
see
each
other
in
two
weeks
again
at
the
community
meeting.