►
From YouTube: 2019 11 04 Memory Team 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
cool,
so
Joshua's
here
today
for
those
who
have
met
he's
our
interim
product
manager
and
his
camera's,
not
working,
so
he
can't
wave
to
everybody.
So
he
run
through
the
agenda.
So
is
Matias
pointed
out.
I
have
some
holiday
PTO
already
called
out
Tom,
just
because
I
know
I'm
travelling
on
those
days,
I
threw
it
out
there,
but
all
these
are
coming
up.
So
please
put
your
PTO
in.
Let
your
let
your
managers
know
how
those
things.
A
Joshua's
gonna
rejoice:
okay,
ours,
I,
just
linked
in
there
take
a
look
they're
very
similar
to
last
time,
so
we
still
have
the
throughput.
Okay
are.
There
was
another:
okay
are
for
proactive
development,
some
more
tooling
and
balancing
out
our
reactive
work
open
the
board.
So
one
can
see
we
are
under
the
enablement
section
and
obviously
memory.
So
we
met
our
hiring
goals
for
q3
with
Nicola
and
Matthias
joining.
Thank
you.
A
We
essentially
met
our
goal
for
increasing
monthly
throughput
targets,
70
percent
on
on
ok
RS,
and
we
had
just
above
eighty
percent,
so
this
last
month,
I
think
we
hit
40
and
March
for
the
month
and
that
excludes
the
onboarding
M
ours
from
Nicola
and
Matthias,
but
still
did
great
for
the
month
and
then
Puma
was
the
third
one.
So
we
have
poom
out
on
canary
I'm,
not
there
yet
on
production,
so
that
will
carry
over
for
the
next
quarter
and
actually
that's
a
very
good
transition.
A
There's
another
areas,
large
town,
so
pretty
unacceptable
latency
at
this
point
in
time,
so
we're
trying
to
follow
along
on
the
issue,
there's
a
lot
of
discussion
about
reconfiguring
the
number
of
CPUs
per
thread
that
we
have
and
I'm
trying
to
balance
it
out,
but
there's
still
some
issues
and
some
late-breaking
stuff
that
happened
this
morning.
I,
don't
can
you
know?
Maybe
you
can
get
everybody
up
to
speed
on
what
was
going
on
with
the
metrics
endpoint.
B
Nothing
really
special
like
we
had
two
like
two
things.
One
is
like.
We
know,
like
I
notice,
that
we
have
this
matrix
and
find
that
is
very
slow
and
the
Ducks
while
it
makes
Puma,
is
arrival
and
we
have
this
matrix
on
point.
Everything
is
implemented
to
switch
that
to
separates
endpoint.
Just
by
that,
so
we
should
basically
try
I
mean
production
team
should
try
and
test
it
and
switch.
B
That's
it
for
now.
That's
the
current
state.
We
we
happily
run.
We
also
started
writing
it
on
two
production
notes
to
out
of
I
think
30
in
to
production
notes.
So
we
have
some
good
comparison
of
the
production
traffic
distribution
as
well
on
Puma.
That
is
so
like
we
run
on
the
canary
and
on
some
small
percentage
of
the
production
feet
today.
A
Yeah
and
the
reason
I'm
calling
this
out
most
of
the
testing
and
configuration
while
all
of
the
testing
and
configuration
right
now
is
on
the
infrastructure
team.
But
if
there's
anything
that
comes
back,
that
we
need
to
jump
on
will
have
to
reprioritize
and
make
sure
that's
our
top
priority.
So
if
there's
any
any
projects
we
need
to
work
on,
then
we'll
have
to
pause
on
like
import
or
sidekick
whatever
to
help
out.
A
C
A
C
What
just
I'm
gonna
current
state?
So
it's
so
thanks
to
me
for
working
with
infrastructure
and
tracking
down
there.
The
Redis
delay
and
things
like
that
as
far
as
path
to
get
miss
on
by
default
do
be
sense.
Camino
of
how
long
you'd
want
to
let
it
sort
of
run
and
product
wreck
and
run
full
wine
production
before
we
make
it
default
to
all
of
our
self-managed
customers,
I
I.
B
Don't
know
the
answer
for
that,
so
I
would
assume
that
initial
plan
was
a
piece
like
one
from
my
stun
to
meet
the
confidence
between
a
deployment
procedure,
but
it
really
depends
on
a
lot
of
things
that
are
not
dependent
on
us
on
how
things
can
not
be
deployed
and
how
it
they
cannot
be
running
so
I
mean
so
far
like
there
is
no
surprises
in
running,
but
we
also
expect
that
there's
gonna
be
some
fine-tuning
connected,
so
maybe
one
my
son
gonna
be
enough.
I
have
no
clue
ready.
B
I
can
just
I
can
help
us
try
to
answer
you
that
question
might
in
two
weeks
from
now
when
we
actually
rama
we've
some
production
traffic.
We
go
through
a
few
deployment
cycles
because
we
use
a
lot
of
new
things
right.
We
use
the
new
zero
zero
deployment
procedure.
It
runs
so
far
right
and
it
sees
the
production
team
is
very
happy
of
that
change
because
we
don't
have
to
like
oversubscribed
notes
as
we
were
doing
before,
and
it
gives
like
a
lot
of
flexibility
in
changing
configuration.
Like
20
configuration
right
now.
B
B
C
B
Like
like
so
from
my
perspective
and
an
actually
like
I'm
kind
of
refreshing,
Maori
perspective
as
well,
I
think
that
on
by
default,
should
be
with
the
next
major
release
because
on
by
default
for
Puma,
it's
mostly
backgrounds
change
for
our
customers.
But
it's
not
fully
background
change
because
you
have
to.
We
have
to
update
our
customers
as
well
on
h8
documentation
and
how
to
perform
zero
downtime
deployments.
We
are
now
complete
with
the
documentation
of
H
a
and
zero
content
deployments.
Yet
this.
B
C
So
we
should
have
that,
so
it
might
be
good
to
have
that
in
that
note,
in
the
blog
post
like
for
that
for
this
month's
blog
post,
that
goes
out
to
say,
hey
start
using
the
employee,
everyone
you
are
using
the
endpoints
and
gain
I
configured
because
we'll
have
cumin,
which
boat
it'll
be
more
important
for
right.
I'm,
just
kidding
folks
murder
heads
up
to
get
that
configured
so.
B
So
I
I
think
like
right
now
we
should
be
looking
like
take
a
look
again
at
the
Puma
documentation
on
how
to
configure
Puma
and
ensure
that
that's
part
of
the
documentation.
There
is
also
like
a
TA
section:
how
to
configure
Puma
for
AJ
and
I.
Think
that,
like
this,
what
we
should
we
are
our
like:
blog,
post
and
customers
to
that
the
documentation
document
ik
also
how
to
configure
AJ.
B
C
Yeah
there's
something
on
a
change
specifically
on
Puma,
so
we'll
want
to
add
that,
but
we'll
also
want
to
make
sure
it
gets
added
to
the
standard,
but
like
the
standard,
consolidated,
AJ
documentation
as
well,
because
that's
what
people
tend
to
follow
so
I'll
make
an
issue
we
can
discuss.
That's
all
the
code.
B
So
like
there
is
also
one
architectural
terms
like
this
kind
of
behavior
also
applies
to
the
Unicode
so
like
we
don't
have
a
difference
in
the
behavior
of
handing
zero
downtime
after
Puma
it's
the
same.
Behavior
is
implemented
for
the
Unicode
and
for
Puma,
so
technically
right
we
could
have
a
documentation
for
AJ
that
this
is
how
you
should
use
H
a
it
should
be
based
on
the
health
check,
not
on
the
specific
behavior
of
the
application
server,
as
it
is
today
and
this
this
this.
C
B
Yes,
yes,
because
like
now,
if
we
use
this
kind
of
trick,
we
could
then
figure
out
I
mean
implement
the
same
mechanism
in
the
workhorse
and
kind
of
have
a
graceful
like
we
start
an
update
of
the
work
as
well
for
basically
for
free
without
doing
any
major
architecture
or
like
reconfiguration
of
user
instances.
If
this
is
needed,
so
they
think
he
like
that.
We've,
the
new
operative
they're,
using
her
check,
we
can
actually
quite
well
support,
is
0.9
deployment
on
more
than
one
note,
if
more
than
one
notice
in
the
cluster.
A
B
A
D
B
A
C
B
B
Don't
know
if
it's
gonna
be
nice
all
aspects
of
that
because
he's,
if
has
problem
on
so
many
different
angles,
this
end
point
and
and
he's
right
now
as
part
of
that
I
I,
don't
know
what
is
the
other
lovely
that
another
each
but
he's
he
created
one
match.
It
was
for
like
very
specific
change
for
now,
like
a
very
specific
improvement,
so
someone
I,
don't
know
Jeanne
are
you
my
tears
could
take
a
look
yeah.
D
So
what
we
came
out
of
the
call
saying
was
that
ego
and
he
actually
had
a
pretty
good
lead
on
where
we
might
be
losing
a
lot
of
time
right
now,
because
there
was
an
inefficient
database
queries.
So
he
said
why
don't
you
just?
Let
me
try
it
at
first
and
see
how
much
better
better
it
gets
so
that
we
don't
like
look
at.
You
know
multiple
things
simultaneously
and
then
we
can't
really
correlate
like
cause
and
resolution.
D
So
so
yes,
so
we
said,
okay,
that
that
smells
great,
actually
and
so
I
think
we
want
to
kind
of
wait
until
those
changes
are
live
to
see
how
much
better
it
gets
and
can
ya
measure
again
and
then
maybe
it's
good
good
enough.
You
know
that
we
can
close
this
out
before
we
like
put
more
time
into
this.
Okay.
A
A
A
D
A
A
A
B
B
There
was
a
comment
from
Andrew
like
kind
of
rephrasing,
like
my
comment
and
the
conclusion
was
that,
like
it's
not
so
important
right
now,
but
we
should
spend
more
time
on
that,
even
though
it's
like
quite
well-defined.
So
unless
we
have
like
more
important
items,
it
would
be
fine
to
deprioritize.
That's
for
now.
A
D
D
D
D
So
it's
said
in
the
description
here:
this
is
to
request
psychic
jobs
being
profile
right
here.
That
implies
running
the
profile
and
safety,
but
we
don't
have
that
capability.
Yet
there
was
another
epic
I
think
for
this,
and
we
investigated
this
for
a
while,
and
then
we
put
this
back
in
the
backlog
because
we
weren't
quite
sure
sure
how
to
go
about
that.
Actually.
D
B
A
B
B
One
here,
yes,
sir
I'm,
where
I
have
to
like
redefine
the
description,
but
I
am
very
close
to
closing
this
as
well,
because
I
think
here,
like
like,
we
have
like
the
one
operas
that
he's
working
with,
have
to
be
find
it
and
update
documentation.
D
Yeah
it's
we
had
really
really
interesting
and
good
discussions
over
the
past
week
or
so
generally
about
sidekicks.
So
there
were
a
couple
of
things
that
came
up
that
this
was
another
one
I
think
this
is
also
labeled
I
think
this
might
be
a
duplicate
of
an
issue
we
already
had.
If
you
scroll
down
a
little
bit,
it
points
to
a
pretty
old
issue.
No,
no
sorry
in
my
continent
that
one
in
my
comments
we're
further
down
yeah.
D
D
Move
this
whole
psychic
psychic
cluster
script
into
core
or
like
our
basic
distribution
and
I,
think
this
is.
This
might
not
be
properly
attacked
because
I
just
created
it
using
an
issue,
template
and
I
might
not
have
used
to
write.
I,
don't
think
is
a
milestone
tech,
but
that
is
the
that
I
started
to
look
into
today.
I
cannot
send
your
link
tonight.
A
D
I've
worked
in
it
today,
I
mean
it's
still
kind
of
yeah.
We
like
looking
into
what
does
it
mean
I?
Think
it
needs
a
proper
breakdown.
I
need
some
more
grooming,
so
it's
I,
don't
know
if
we
call
it
in
deaf
already
or
if
it's
still,
you
know
and
feeling
press
it's
not
quite
ready
for
depth
because,
like
I,
don't
think
we're
at
a
point
yet
well,
we
wait
can
knock
out
code
but
yeah
we're
trying
to
figure
out
what
what
we
could
do
about
it.
Okay,.