►
From YouTube: 2021 01 04 Memory Team Weekly
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
happy
new
year:
this
is
the
first
memory
team
meeting
of
the
year
and
I
figured
we'd
just
jump
right
into
the
board
and
figure
out
where
we
are
there's
a
lot
of
pto
taken
and
I've
been
out
for
the
last
couple
weeks.
A
First
one
see
nikola
and
I
talked
about
it
a
bit.
This
one's
going
to
remain
open
until
the
feature
flags
removed,
probably
need
to
well
for
all
of
them.
We
need
to
add
either
deliverable
and
or
workload
workflow
labels
on
them,
because
it
has.
C
Does
anybody
know
what's
going
on
with
puma?
If
not,
we
can
just
wait
till.
B
D
Yeah,
no,
that's
right.
I
mean
it's
because
I
mean
I
think,
stan
said
it's
not.
He
doesn't
expect
it
to
cause
any
any
problems,
and
I
I
went
through
all
all
those
issues
and
we
did
run
the
qa
the
tests
and
it
sounded
like
it
was
fine,
but
yeah.
No
one
wanted
to
make
a
change
like
that.
You
know
and
everything
yeah,
but
everyone
yeah
yeah,.
A
For
sure,
should
we
mark
it
as
endeavor
in
review.
D
E
D
Yeah
I
parked
work
on
this
because
yeah
after
a
while,
because
I
spent
a
couple
days
trying
to
get
to
a
point
where
we
can
compare
before
after
results
in
some
kind
of
meaningful
way,
and
I
I'm
still
like
not
really
able
to
do
that,
because
there's
still
like
pretty
wild
swings
in
some
of
the
key
metrics,
even
when
trying
to
establish
a
baseline,
which
is
kind
of
a
big
deal,
because
I
think
we've
kind
of
relied
on
these
perf
test
results
in
in
the
past.
D
To
say:
oh,
it's
fine,
you
know
like
and
like
it's
really,
it's
so
unpredictable
like
I,
I
see
swings
like
with
it
with
no
change
at
all.
I
see
swings
of
like
six
percent.
You
know
in
like
memory
use
or
something
like
this-
that
that's
not
small.
You
know.
D
You
know
north
of
a
gigabyte
per
worker-
that's
not
small!
So
so
I
wanted
to
talk
to
camille
and
alexi
about
it.
They
said
they
might
have
some
ideas
as
well,
so
I
switched
to
other
topics
for
now.
A
Okay,
should
we
mark
this
one
as
block
for
now
then
or
just
leave
it
as
it
is.
D
No,
I
mean
it's
not
blocked
by
anything.
It's
really
just
I'm
looking
for
help
here.
I
need
some
like
I've.
Looked
at
this
for
too
long
as
well.
I
kind
of
would
like
to
get
a
second
opinion
on
like
some
like
fresh
eyes
and
it's
like
maybe
how
how
we
can
approach
it,
because
maybe
we
should
just
approach
it
totally
differently.
D
You
know-
or
maybe
we
should
just
be
a
bit
more
brave
and
just
pick
out
things
that
we
intuitively
feel
might
move
the
needle
in
the
right
direction
and
then
just
find
a
way
to
safely
deploy
into
prod,
rather
than
getting
hung
up
on.
Like
all
these
performance
test.
Suites
stuff,
you
know,
is
another
option,
but
I
don't
want
to
do
this
like
yeah.
I
want
to
talk
over
this
with
the
folks
on
the
team.
Okay
camille
said
like
he
might
have
some
time
tomorrow
to
talk
about
it.
G
B
B
G
I
I
mean
my
my
question
is
really
like
about
our
expectations
for
13.8,
which
is
like
it's
like.
This
issue
describes,
provide
america
needs
to
selectively
load
like
do
you
think
that
this
is
something
that
we
can
achieve
in
13.8
or
it's
like.
We
are
really
like
poc
fights
right
now
on,
investigate
and
investigating
the
direction
that
we
may
execute
in
the
next
milestone.
Like.
B
Well,
I
would
more
say
it's
investigation
and
I
will
stay
with
the
psc.
So
if
we
agree
that
the
mechanism
is
good,
then
we
can
proceed.
Maybe
for
the
next
release
with
having
something.
D
D
B
D
Like
and
I
I
don't
know
the
details,
but
I
do
remember
that
rails
engines
got
a
fair
amount
of
criticism
as
well
back
in
the
in
the
days,
because
they
they
they
might
introduce
other
problems
to
the
code
base
which
have
nothing
to
do
with
like
us.
Looking
to
you
know,
separate
separate
out
grad
loading
it
it
will
affect
the
architecture
of
the
app.
So
I
I
assume,
I
think
the
right
way
of
doing
this
is
if
this
works.
Well.
D
B
D
C
G
G
A
Well,
we
already
talked
about
that
one,
that's
probably
dependent
on
ruby.
I
just
actually
assigned
this
one
to
grant
I'll
follow
up
with
him
on
that
one.
I
don't
it's
not
super
critical,
it's
more
of
a
data
gathering.
A
We
have
anecdotal
information
that
upgrading
to
ruby
27
was
not
hugely
impactful
in
memory,
but
he
wants
to
actually
test
and
measure
it
camille.
You
are
up
next.
F
Fabian
we
have
this
one
yeah,
so
I
think
the
other
team
has
agreed
to
a
handover
session.
I
still
have
to
schedule
that
one
we'll
remind
them
to
read
the
docs,
but
I
think
that
would
be
a
30-minute
affair
of
here
is
this
matthias
can
hopefully
answer
any
immediate
questions,
and
then
I
think
we
can.
We
can
close
this
out.
I
mean
it's
documented.
F
I
don't
think
this
is
an
issue,
but
I
still
need
to
schedule.
It
was
good.
A
D
Yeah,
I
might
within
the
next
two
weeks.
I'm
not
sure
I
mean
I
have
some
context.
G
All
right,
camille
now
really
progress
on
that
one.
I
asked
for
some
clarification
from
the
op
on
like
what
is
happening,
but
I
got
the
answer,
but,
like
not
conclusive,
so
I
need
to
ask
again
with
more
detailed
explanation.
What
I'm
looking
for?
Okay,
matthias.
D
Yeah
and
that's
what
I'm
working
on
right
now,
because
there
wasn't
any
anything
immediately
for
me
to
do
so
so
alexa
handed
over
the
github
exporter
stuff
to
me,
but
he
had
like
three
or
four
open
mr's,
and
there
was
nothing
immediately
to
do
so.
I
looked
for
the
next
step
there
and
I
found
this
to
be
probably
the
next
most
actionable
one
and
it's
a
bit
chunkier.
So
I
wanted
to
make
some
progress
there.
It's
basically
there's
at
least
two
metrics.
D
We
know
of
that
gitlab
exporter
currently
exports
around
sidekick,
that's
stuff.
We
alert
on
as
well.
So
it's
quite
important.
So
I
have
an
mri
out
to
move
that
into
the
application
as
well.
Yeah
there's
some
risk
attached
there,
because
we
need
to
query
for
this
site
for
this
stuff
in
an
efficient
way,
because
it's
now
part
of
the
application,
but
I'm
yeah
I'm
trying
to
make
it
make
it
efficient
and
yeah.
I
I
hope
to
get
some
feedback
on
this
soon.
D
Flaws
with
the
way
I
go
about
it,
then
I
think
that
can
be
that
we
should
be
able
to
do
it.
Yeah
all
right.
F
Fabian
yeah,
I'm
not
quite
sure,
right
out
of
the
top
of
my
head.
What
the
next
step
here
is.
I
believe
I
have
a
meeting
scheduled
actually
for
this
week
with
sarah
to
discuss
the
gitlab
exporter
and
sort
of
the
monitoring
impact
or
the
impact
that
monitoring
has
on
our
memory
usage
and
how
we
can
proceed
there.
D
Okay,
lexi's,
not
here
yeah.
I
have
some
context
on
this.
This
is
also
that
should
be
okay.
So
there
were
two
steps
to
this.
I
think
he
broke
it
into
a
bunch
of
separate,
mrs
that
I
looked
at
and
they
looked
fine
to
me
again.
It's
like
mostly
waiting
for
feedback
and
yeah.
It's
a
matter
of
getting
them
merged.
A
D
Yeah
I
mean
we're
always
like
a
bit
dependent
on
folks
from
infrastructure
and
maybe
scalability
team
to
kind
of
green
light.
These
changes,
because
we
were
very
careful
we're
trying
to
be
careful
not.
D
D
I
know
that
alexei
started
to
look
into
this
ipad
with
him.
On
on
this,
let
me
try
to
recall:
did
we
send
out
an
alarm
for
that?
It
might
still
be
inward
in
progress.
D
We
figured
out
a
way
to
do
it,
but
there
might
not
be
a
specific
mri
yet
put
it
in
develop
in
development.
D
D
Yeah,
this
is
a
fun
one.
So
basically
stan
noticed
that
our
error
rate
went
up.
This
is
the
whole
thing
you
know
where
we
had
to
silence
the
alert
for
like
seven
months
now
and
yeah.
He
said
like
he
had
a
couple
really
interesting
ideas
for
how
to
go
about
this,
and
I
was
first
not
really
in
favor
of
patching
it
on
the
way
out
at
the
serving
layer,
because
it's
really
a
workaround.
F
D
I
was
a
bit
afraid
that
we
would
keep
adding
these
workarounds,
but
I
spent
like
an
afternoon
on
it
and
I
so
he
actually
created
an
issue
against
the
go
image:
imaging
library,
because
it's
kind
of
like
a
bit
of
a
quality
issue
with
that
internal
goal,
library
as
well
yeah
and
the
maintainer
the
go
maintainer,
he
suggested
hey,
you
could
try
work.
You
know
credit
work
around
if
we
did
so.
I
have
an
mr.
D
I
would
stand
to
prove
that
I'm
waiting
for
a
workhorse
maintainer
to
look
at
it
yeah.
So
hopefully,
if
this
works
out
well,
we
should
see
that
these
errors
drop
again
significantly
because
we
strip
out
basically
the
bad
data.
That's
causing
the
overly
strict
go
run
time
to
reject
these
images.
A
F
A
So
that's
it
for
the
individual
issues
themselves.
So,
on
the
bigger
picture
side
of
things.
Where
do
we
think
we'll
be
at
the
end
of
the
milestone?
And
it's
probably
you
know
if
nobody
has
an
idea,
then
we
can
talk
about
it
next
week.
But
if
the
goal
for
13.8
was
to
get
us
closer
to
the
2gig
footprint
or
make
some
progress
on
it,
what
measurable
impact
will
we
have
within
138?
Does
anybody
have
an
idea
and
again,
if
not,
that
we
could
talk
about
it
next
week,
but
showing
progress?
A
You
know
next
week,
we'll
have
one
week
left
in
the
milestone:
we'll
have
to
start
thinking
about
release,
post
and
figuring
out
what
we've
actually
shipped
for
13.8
along
that
goal
with
2gig
footprint,
so
anybody
have
any
thoughts
on
where
we
are
and
what
progress
we've
made.
So.
G
Far
so
I
am
personally
looking
very
forward
like
to
what
matias
is
working
with
the
insults,
because
it's
like
the
first
iteration
like
to
figure
out
if
we
can
reduce
the
memory
you
search
and
like
we're,
gonna,
actually
probably
figure
out
also
a
way
to
get
this
baseline
comparison
then
like
what
is
alexi
working
with
puma
2.5
with
five
five
point,
one,
it's
something
more
like
future
proofing
that
gonna
give
us
benefits
in
the
next
milestone,
not
really
in
this
milestone,
but
this
is
kind
of
like
pre-quest
nicola
poc.
G
So
I
guess
this
is
kind
of
like
the
height
of
a
summer,
like
a
lot
of
now
lights
on
matias
and
figuring
out
this
baseline
and
shipping.
This
very
small
gc
item,
because
this
baseline
gonna
be
like
a
baseline
for
what
other
people
are
working
on.
D
D
Is
still
kind
of
laying
the
groundwork,
I
think
it's
kind
of,
like
you
said
as
well
and
in
in
terms
of
actually
reducing
memory
use,
I
think
actually
the
most
promising
thing
might
be
to
run
gitlab
exporter
on,
because
that
might
actually
shave
off.
G
Guess
it's
quite
kind
of
interesting
because
what
we
are
kind
of
mentioning
it
means
that,
like
we
actually
have
two
themes
of
the
high
level
overview
like
the
team,
which
is
bigger,
which
is
like
the
whole
package,
as
you
run
that
and
the
second
team,
which
is
like
ruby,
optimization
that
really
affects
everyone,
every
installation.
So
I
guess
we
kind
of
need
to
figure
out
a
way
for
these
two
teams
to
be
measured.
F
F
G
First
one
is
like
nakayashi
fork,
which
gonna
reduce
the
mmr
usage
in
general
for
everyone
and
which
is
like
the
ruby,
optimization
team,
and
the
second
one
is
like
puma
single
that
gonna
affect.
This
is
like
the
small
install
of
the
whole
package
thing.
Yes,
matthias
is
working
and
matthias
is
mentioning
about
yeah.
F
D
I
I
just
wanted
to
say
puma
single,
I
I'm
not
sure
we'll
get
it
done
this
month,
because
I
don't
think
I
don't
think
anyone's
working
with
us
right
now
right.
I
think
we
look,
but
last
time
I
was
experimenting
with
it.
There
was
still
a
lot
of
manual
work
too,
like
metrics.
H
G
D
Yeah
yeah,
that's
just
a
config
switch,
but
again
I
don't
think
it
will
save
any
memory.
It
just
kind
of
puts
the
application
into
like
a
cleaner
memory
layout
just
before
forking
into
workers,
but
I
still
like
how
much
that's
going
to
have
a
lasting
effect
on
lower
memory
use.
I
think
it
will
be
very
small,
at
least
that's
what
the
that's
what
we
found
when
we
were
looking
at
this
stuff
in
the
research
week
right
it
basically
deteriorates
very
quickly.
D
F
But
I
think
I
liked
I
like
how
camille
said:
there's
actually
two
themes
going
on
one
is
maybe
more
for
global
changes
right
and
then
there's
one
that
is
specific
for
memory
constraint,
environments
and
they
have
you
know
different
different
impact,
and
you
know,
with
my
infinite
amount
of
time
on
my
hand,
maybe
we
can
for
sort
of
next
week.
Just
tease
these
two
things.
F
You
know,
but
I
think
that
may
also
help
us
sort
of
say:
hey.
These
are
things
that
actually
will
affect
gitlab
as
a
whole
by
then
gitlab.com.
Those
are
specific
things
that
will
have
an
outsized
impact
for
a
small
installation,
and
I
think
that
may
then
also
make
it
clearer
what
the
impact
respectively
is.
A
G
The
one
that
I
posted
on
the
memorial
channel
yeah,
it's
it's
my
plc
on
the
rugby
pm
to
count
memorial
locations.
I
I'm
not
sure
yet.
G
I
need
to
figure
out
if
there
is
like
easier
way
to
see
that,
because
this
could
be
like
one
of
the
ways,
much
or
memory
usage
by
the
future
categories
and
like
have
let's
say
some,
some
some
inform
some
indication
like
how
much
like
the
given
aspects
of
the
product
how
unoptimized
they
are.
So
I'm
not
sure
if
you,
if
you
understand,
like
the
cons,
the
concept
behind
this
issue
and
behind
this,
mr,
and
what
it
offers
and
like
how
we
could
tire
tie
that
with
the
feature
category.
A
Guess
and
while
we're
talking
about
all
these
things
and
breaking
them
down,
like
we
talked
about
the
proposal
issue
earlier
we're
talking
about
graphql,
it
reminds
me
we
need
to
I'll
create
a
13-9
planning
issue,
so
we
can
start
talking
about.
What's
next.
F
A
Yeah,
lots
of
catching
up
to
do
with
all
the
pto.
My
to-do
list
for
today
is
quite
long,
so
that
is
all
we
had
on
the
team
agenda
today.
Any
other
topics
we
need
to
cover.
A
In
review
yeah,
that
is
something
that
I
had
planned
to
talk
about
for
13.9.
So
reviewing
our
processes,
you
have
that
proposal
for
that
demonstration
of
other
boards
that
you
use
on
other
teams.
A
So
I
have
it
scheduled
for
13
nine
since
alexi's
out
this
week
and
we're
all
kind
of
catching
up
but
yeah,
something
we
could
talk
about
next
milestone,
but
for
now
we'll
keep
using
it.
H
H
H
H
Months,
yeah
yeah
that
sounds
about
right,
cool
yeah,
I'm
just
looking
like
all
all
the
time
during
the
meeting
I'm
wondering
like
like
fabian
is
talking
about
this
memory
stuff
and
I'm
like
what
what
does
like.
I
don't
know,
monitoring
and
memory
has
to
do
with
geo.
Like
maybe
he's
volunteering,
I
don't
know.
I
have
no
idea
now,
I'm
looking
at
the
page
and
cool,
that's
great
cool
the
question
I
had
so
I'm
looking
for
it.
I
know
fabian
did
the
same
thing
for
for
geo.
H
So
by
the
way
I
mean,
if,
if
you
didn't
see
I'm
here
from
from
support
side,
I'm
I'm
the
support
stable
counterpart,
so
fabian,
whatever
you
need
from
support
in
terms
of
memory
group,
feel
free
to
reach
out,
and
so
I
know
fabian,
you
did
the
thing
with
geo
that
that
you
created
a
view
and
you
used
some
keywords
to
actually
follow
tickets
and
stuff.
So
basically
I
I
just
wanted
to
follow
up.
So
aside
from
from
memo
for
memory
and
basically
ram,
do
you
do
everyone
like
do?
H
Do
you
know
any
any
other
keywords
that
might
be
useful
that
I
can
actually
track
tickets
for
because
right
now
you
know
we
don't
really
have
a
way
to
say
you
know
this
ticket
is
about
memory.
It
can
end
up
being
like
that.
But
you
know
there
is
no
clever
way
of
doing
it.
One
of
the
things
and
then
fabian
knows
that
because
he
did
it
for
geo,
not
sure
if
you,
if
you
did
anything
for
memory
in
in
that
sense,
fabian.
F
Yeah,
so
I
I
look
at
zendesk
quite
a
lot,
because
geo
generates
a
lot
of
support
tickets,
it's
joyful
and
so
essentially
there's
a
way
to
set
up
keyword
alerts.
You
know
if
a
customer
writes
a
support
ticket
and
says
my
geo
upgrade
failed
right.
F
I
get
the
notification
because
it
mentions
geo
and
giro
is
really
easy
to
find,
because
it
has
a
distinct
name,
and
I
think
what
moment
is
asking
is:
are
there
any
specific
good
keywords
for
for
memory,
because
it
makes
it
easier
to
filter
out
all
of
the
the
support
questions
that
people
may
have
because
of
it?
Is
that
right?
Did
they
understand
correctly
yeah.
H
That
that
that's
perfectly
right,
yeah,
so
jen
I
mean
anyways.
I
am
going
to
to
create
a
view
for
for
memory
and
ram
for
at
least
for
these
two
keywords,
because
I
I
did
see
like
in
in
couple
of
like
these,
like
few
months
that
I've
been
looking
at,
there
were
some
tickets,
but
also
like
this
is
I
mean,
even
if
you
don't
have
anything
on
your
mind
right
now,
being
me
with
with
anything
that
might
come
up
yeah,
I
don't
know
in
the
future.
H
Maybe
you
are
working
on
something
at
the
moment
and
you
know
like
you
expect
tickets
or
something
like
that.
Just
being
me-
and
I
can.
I
can
basically
filter
these
out
for
you
and
see
if
there
is
anything,
useful
and
I'll
definitely
come
up
with
tickets
for
you.
If
I
think
you
know,
I
don't
know,
you
might
be
interested
or
useful,
but
it
might
be
an
interesting
data
point
for
for
the
future.
F
I
think
memory
is
probably
a
little
bit
difficult
to
filter
for
with
keywords,
because
it's
more
generic,
but
maybe
there
are
better
words.
What
I
I
mean
you
know
for
geo.
The
flow
that
I
I
have
is
if
there
are
specific
questions
from
support,
you
know
just
go
to
the
memory
channel
and
raise
them
there,
and
you
can
ping
me
on
them
and
then
we
can
look
through
it.
If
there
are
specific
zendesk
things,
you
can
also
ping
me
there
and
I'll
I'll.
Take
a
look.
F
A
All
right,
you
can't
hit
the
mute
button.
Yeah
couple
non-verbalized
topics
in
there
to
read
in
some
requests
for
feedback
so
other
than
that
everybody
have.