►
From YouTube: 2020-09-08 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
Okay,
the
recording
has
started-
and
this
is
the
september
8th,
2020
rook
community
meeting-
let's
jump
into
our
latest
milestones,
so
we've
done.
We've
done
a
couple
of
past
releases
recently,
but
looking
forward,
we
don't
currently
have
a
1.3
patch
release
scheduled,
but
we
do
are
considering
a
patch
release
for
1.4
later
on
this
week.
So
let's
take
a
look
and
see
what's
on
the
board,
there.
B
Yep,
if
it's
this
thursday,
it
would
just
match
our
bi-weekly
schedule
for
patch
releases.
Since
1.4
came
out,
I'm
just
catching
up
since
I
was
out
of
the
last
week.
I
think
yeah
it'd
be
good
just
to
plan
for
that.
B
A
Does
anybody
have
an
update
on
that
one?
Is
that
a
consistent,
repro
travis
that
you
know
of
or
is
it
intermittent.
C
It
currently
seems
a
bit
like
it
might
be
something
with
1.18,
at
least
all
the
people
like.
I
stumbled
upon
the
issue
today
as
well
at
a
customer,
so
we
all
had
1.18
or
higher
or
well
patch
release,
not
1.19
yeah,
so
it
might
be
even
like
this
kubernetes
issue
there,
which,
for
whatever
reason
like
there,
you
have
it
in
the
kubernetes
issue.
C
C
C
There,
oh
might
have
been
a
different
kubernetes
issue,
but
I
think
I
saw
one,
but
it
was
yeah.
The
kubernetes
garbage
collection
seems
to
remove
like
monetary
deployments
and
ocu
deployments
and
all
sort
of
things
for
some
reason.
Where
is
this
column?.
C
Which
kubernetes
issue
are
you
in?
Oh,
this
is
a
different
one.
Yeah
there
is
the
I'm
not
in
the
should.
I
just
tell
you
the
number
jared.
C
And
the
number
four
of
the
kubernetes
issue
is
eight
eight
zero,
nine
seven
that
is
eight
eight
zero,
nine
seven
exactly
and
the
the
last
comment
there
have
the
same
problem
with
one
point:
nineteen
zero
also
seems
to
1.19
is
even
yeah,
yes
and
one
deployment
services,
manager
deployments
and
all
those
seem
to
be
disappearing
from
time
to
time.
C
A
Yeah
and
this
this,
it
was
interesting,
though
this
hasn't
gotten
a
lot
of.
You
would
think
that
the
implications
or
the
outcome
of
an
issue
like
this
of
the
garbage
cleaner,
garbage,
collective,
deleting
resources
that
it
shouldn't
be
you'd,
think
that
would
be
like
a
major.
You
know
fire
drill,
craziness,
and
this.
B
Has
it's
been
inactive.
A
For
for
you
know
a
couple
months,
let's
say
so
I'm
so
that
so
that's
that's
interesting.
Maybe
there's
a
different
upstream
issue.
That's
tracking
this,
which
is
getting
more
noise,
but
it
doesn't
seem
like
this
issue
back
in
february,
is
getting
the
attention
you
would
think
it
would
be.
If
it
was
a
widespread.
B
A
Yeah,
it
was
the
one
thing
to
look
into
is
like
like
this
issue
is
indicating
here
is
that
you
know
references
across
namespaces
are
invalid
and
it's
a
deterministic
or
undeterministic
when
or
if
they'll
get
garbage
collected.
So
it
seems
almost
a
little
bit
random,
but
but
they
are
invalid.
References
that
will
get
garbage
clear
are
very
likely
to
get
garbage
collectors.
I
don't
know
if
there's
been
a
logic
change
in
things
with
the
cs.
A
This
csi
implementation
here
for
us,
but
if
we
are
doing
anything
like
we're
doing
owner
references
that
go
across
name
spaces,
then
that
is
definitely
something
to
look
into
that
could
be
causing
this
issue.
B
A
A
Oh,
this,
oh
jordan,
jordan
says
that
it's
likely
the
same
as
65
200..
Can
you
go
to
that
issue,
or
maybe
not
that's.
That's
the
other
one
we
had
here.
Oh,
that
was
yeah,
but.
A
All
right
yeah,
so
thanks
for
looking
to
there
for
sure
other
other
notes,
anybody
on
the
other
blocking.
B
D
We
don't
we
don't,
we
cannot
make
it
optional,
but
yeah.
I
will
look
into
this
now
that
we
have
more
logs,
very,
very
yeah.
C
C
C
C
D
C
B
D
I
guess
the
thing
is
that
before
adding
cluster.local,
there
was
nothing
it's
not
like.
I
wasn't
really
under
the
impression
I
was
breaking
something
at
least.
C
D
C
C
Are
a
bit
special
regarding
dns,
like
I
think,
especially
like
the
alpine
muscle
library
also
had
some
well
has
some
troubles,
I
think
maybe
the
a4
finally
finally
fixed
it,
but
the
muscle
libraries
of
alpine
especially
had
some
issues
regarding
dns
resolution
when
you
didn't
provide
at
least
like
three
four
five
away
or
something.
B
A
All
right
and
then
yeah
still
making
progress
towards
1.5
and
still
planning
to
do
a
the
release
for
that
before
kubecon
north
america,
which
is
you
know,
mid
to
late
november.
B
A
Okay,
all
right,
we
can
move
on
to
the
community
topics
section
now,
so
we're
up
to
six
binding
votes
now
for
the
graduation
proposal.
What
do
we
need?
Seven
right,
travis.
A
So
one
more
vote
left
hooray.
We
don't
have
any
minus
one
votes,
which
is
good.
All
six
binding
votes
that
have
been
cast
have
been
positive
or
affirmative.
So
that's
that's
good.
Hopefully
we'll
get
that
last
that
last
vote
sometime
soon
and
that
would
be
wonderful
to
finally
get
there
just
waiting
and
waiting
forever
more
votes,
all
right,
yeah,
no,
no
action
items
associated
with
that
right
now,
so
google
summer
of
code
has
now
officially
come
to
completion
for
the
summer.
A
It
was
a
fantastic
summer
that
we
had
here
with
ahmad,
who
is
on
the
call,
I
believe
today,
yeah,
so
ahmad
had
a
lot
of
contributions,
an
effort
all
around
the
nfs
operator,
and
you
can
see
ahmad's
final
submission
of
work
here,
the
artifacts
for
all
the
things
that
he
did
here
at
this
link
in
the
community
agenda.
If
you
go
to
the
official
google
summer
of
code
website
also
you'll
see,
this
is
the
same
link
here
as
published
as
the
final
submission
for
the
project.
A
The
announcements
are
today
for
the
students
that
have
passed
or
failed,
but
I
think
ahmad,
if
he
hasn't
yet
gotten,
that
announcement
is
very
confident
about
which
ways
which
what
the
outcome
will
be
so
no
worries
or
concerns.
There
were
drama
there
whatsoever
and
then
also
ahmad
went
ahead
of
the
schedule
for
the
summer.
A
When
we
at
the
beginning
of
the
summer,
we
planned
out
what
we
were
hoping
to
accomplish
and
ahmad
did
more
than
that,
and
so
he
does
have
some
prs
that
are
still
outstanding
right
now,
around
nfs
disc
quota
that
have
not
yet
been
merged.
But
those
were
you
know,
90,
we
didn't
even
think
we'd
get
to
those
or
those
weren't,
even
on
the
schedule
when
we
started
the
summer,
so
that's
all
all
good
there
for
sure
being
able
to
exceed
what
the
expectations
of
accomplishments
were
ahmad.
A
Let
me
know
if
you
have
any
any
words
or
anything
to
share
here
on
this
call,
since
the
summer
is
now
now
completed.
For
you.
B
This
is
a
good
summer,
good
experience
for
me
in
the
rock
community,
thanks
for
everyone
support
me
in
this
google
summer
of
good
program.
A
Right
on
awesome,
yeah
that
was
yeah,
it
was
it
was
a
great
summer
and
you
know
we're
hoping
that
ahmad
will
find
the
time
to
continue
contributing
to
to
rook.
As
he's
you
know,
doing
more
university
at
school
this
year.
Of
course,
that's
going
to
be
the
primary
focus,
we'll
be
graduating
and
getting
done
with
with
university,
but
you
are
always
welcome
in
the
community
ahmad
for
any
other
contributions
that
you
are
you're
wanting
to
do.
B
A
Yes,
sir,
thank
you
aman
all
right.
Let's,
let's
head
head
on
to
ci
issues,
fun
fun
times
there
I
know
adam
had
made
some
changes
recently
and
we're
looking
at
the
you
know,
reliability
and
time
for
each
build
and
stuff.
You
guys
want
to
kick
off
the
conversation
there.
D
Yeah,
I
don't
have
much
update
regarding
ci
issues.
I
did
a
few
things
after
we
switched
instances
flavors,
because
the
disk
naming
naming
in
this
new
flavor
was
different.
So
some
scenarios
in
the
ci,
such
as
oysters
on
pvc,
were
not
working
anymore,
so
I
fixed
that
I've
also
tried
to
reduce
the
number
of
intimate
intermittent
issues
we
were
having,
especially
especially
with
the
manager's
suite.
D
B
D
A
Okay,
that
sounds
that
sounds
reasonable.
Did
oh
yeah,
adam
adam,
mentioned
that
he
found
a
fix
or
realized
why
it
looked
like
some
bm
instances
were
getting
reused.
Oh
more
stability
from
that,
or
is
that
he's
still
he's
still
looking
into
that,
I
guess
maybe
today
I.
D
A
B
A
Yeah
that'd
be
exciting
for
adam
to
have
more
conclusions
on
that
one.
I
know
adam
also
made
a
change
to
the
instance
type
that
we're
using
for
the
slaves.
I
think
eric
test
notes.
I
guess
so
that's
that's
good
too,
in
terms
of
cost
were
there
any
other?
Were
there
any
other
conclusions
around
like
why
the
first
the
month
is
insanely
expensive
compared
to
all
the
other
days
and
stuff,
like
that,
any
any
other
input
or
out.
B
Not
that
I
know
I
asked
adam
about
that
too,
but
I
don't
remember
him
responding
to
that
either.
A
Yeah,
that
was
weird
right.
A
Well,
actually,
the
weird
part
with
that
was
that
he
was
making
a
statement.
That
august
was
cheaper
than
the
you
know
like
may
june
and
july
were
where
you
know
the
costs
had
doubled.
But
then
august
was
like
not
that
bad
again,
and
that
was
before
he
made
the
change
because
he
made
the
change
to
the
instance
type
on
the
last
day
of
august.
A
A
Yeah,
I
think
I
think
in
general
the
entire
month
of
august
was
was.
If
you
look
at
the
bar
graph,
there
is,
it
was
lower
than
june
and
july,
which
are
also
on
the
graph.
A
So
I
don't
know
I
mean
that's
good,
but
it's
you
know
just
another
one
of
the
we're
not
sure
here
that
we
don't
really
have
a
great
explanation
for
cool
any
other
updates
on
ci
related
stuff.
The
you
know
adam's
working
on
figuring
out.
Why
we're
not
getting
fresh
or
clean
vms?
We
did
already
lower
the
instant
size
for
our
our
test
nodes,
which
is
which
should
have
an
expected
60
cost
reduction,
which
is
good
anything
else
here.
Right
now,.
A
Can
you
can
you
explain
that
further
sebastian
like
what
what
is
going
to
have
actions
do
so.
D
I
think
it's
something
that
I
introduced
six
to
eight
months
ago
on
github,
it's
a
way
to
integrate
ci
cd
within
the
github
pipeline.
So
it's
it's.
If
you
will
it's
like
jenkins
from
getup
they
more
or
less
share
the
same
capabilities.
D
So
we
use
we
use
already
actions
for
doing
comet,
linting,
also
running
gosek,
and
it
has
proven
to
be
super
reliable.
So,
like
those
vms,
are
like
quite
decently
fast
for
this,
the
flavor
they
have
and
yeah.
I
think
it's
just
about
about
some
investigations
there,
because
it's
well
integrated
into
getup
it's
kind
of
ci
for
free
right
now,
because
we're
not
really
limited
in
the
number
of
slaves
we
we
can
allocate
to
a
given
pr.
D
A
D
Yeah,
there
are
no
limits
right
now,
unless
you
run
private
repos
or
something
you
have
to
pay
for
this.
But
right
now
it's
just
free
like
free
of
charge.
It's
like
yeah,
I
think
they're,
just
also
kind
of
killing
travis
ci
right
now,
because
it's
more
or
less
what
they
have
been
offering
with
obviously
less
compute
power.
D
But
it's
just
that
if
we
can
get
a
much
more
ci,
a
much
more
stable,
ci,
slash,
reliable
ci,
and
even
though
I
would
I
would
go
even
further
saying
that
even
if
the
runtime
takes
just
a
little
bit
longer,
then
I
will
be
really
in
favor
of
migrating
more
much
more
scenarios
that
we
have
on
github
actions.
To
be
honest,
yeah.
A
Yeah,
I
totally
support
that
as
well.
Seb
like
reliability
is,
probably
you
know
the
highest
priority
sort
of
thing.
A
Well,
besides
cost,
but
if
it
was
like
you
know,
20
000,
to
run
every
month,
then
that
would
be
crazy,
but
but
yeah
so
I
mean
it's
definitely
I
absolutely
think
it's
worth
looking
into
and
if
like,
if
we
get
a
good
feel
for
it
and
it
is
reliable,
it's
stable
it's,
you
know
fast,
like
it's,
it's
reasonable
in
terms
of
the
offering
for
what
they
have
it
like
free,
tiers
versus
or,
like
you
know,
some
cost
to,
especially
because,
like
at
the
end
of
the
day,
we'll
need
to
compare
it
to
what
we're
already
paying
for
aws,
which
is
you
know
like
our
baseline
there.
A
A
D
Well
so
far,
I
didn't
manage
to
run
any
suite,
yet
one
was
failing
right
at
the
end,
so
I
need
to
see
if
what's
going
on,
but
but
still,
what
was
really
encouraging
is
that
by
just
the
strapping
mini
cube
using
this
ssd
on
the
vm
and
just
doing
cube,
ctr
create
common,
yellow,
then
operator,
yeah,
mold,
then
cluster
test,
yellow.
D
B
D
B
A
A
D
A
Cool,
that's
encouraging
sebastian
cool
yep,
nice,
nice.
I
think
I
had
another
comment
on
that.
I
can't
remember
now,
though,.
A
Yeah
well
yep.
Thank
you.
Thank
you,
microsoft,
deeppockets,
for
providing
free
resources
to
github
projects.
D
All
right,
oh
yeah,
I
think
the
the
only
downside
to
this
is
that
people
that
are
looking
at
running
our
pipeline
anywhere
else
won't
really
be
able
to,
because
we
this
would
mean
that
we
will
stop
maintaining
the
jenkins
pipeline,
slash
file.
We
have-
and
maybe
ultimately
this
might
be
outdated
where,
right
now
we
have
users
running,
I
guess
the
pipeline
pipelining
pipeline
in
their
own
ci,
it's
probably
easier
for
them
to
integrate.
That's,
I
guess
a
small
downside,
because
I
don't
think
we
have
that
many
people
doing
this.
C
I'm
using
that
for
some
private
well,
some
well
public,
slash
private
projects
as
well.
Like
just
writing
up
the
workflow.
I
see
yaml
for
most
ci
tools
and
well
it
does
its
job
so
releasing.
B
A
Okay-
and
we
didn't
have
any
prs
to
add
here,
which
is
discussed
specifically
travis-
you
are
back
welcome
back
and
I
see
this
action
I'm
carrying
over
from
from
your
time
time
away.
A
Oh
all
right:
well,
that's
everything!
We
had
on
the
agenda
anything
else
that
anybody
wants
to
bring
up
in
this
forum
right
now,.