►
From YouTube: 2019 08 08 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
Oh,
that's
new.
Alright,
thanks
for
the
warning,
okay,
so
hiring
update
conversation
with
the
internal
candidate
continues
moving
to
the
reference
stage.
Hopefully
we
can
put
together
a
compelling
offer
and
get
going
on
that
pretty
soon
external
candidate
I
have
reference
checks
set
up
for
later
and
next
week.
Actually
so
again
close,
hopefully
we
can
come
up
with
a
compelling
offer
and
we'll
have
two
new
folks
to
join.
The
team
have
other
folks
in
technical
interview
stage,
so
things
looking
good
on
the
hiring
front.
A
The
product
manager
starts
next
week
on
Monday
and
they
will
be
splitting
time
between
distribution
and
memory.
I'm,
not
sure
what
the
allocation
per
team
will
be,
but
we'll
find
out
as
time
goes
on.
Let's
see
so
infrastructure
escalation
process
memory
team.
We
don't
have
any
engineers
that
are
eligible.
You
have
to
have
been
here
at
least
three
months
and
not
be
a
distinguished
engineer
so
that
effectively
disqualifies
everybody
on
the
team,
I'm,
not
sure
if
QE
folks
are
going
to
be
involved
in
that
grant.
Have
you
heard
anything
about
that.
A
So
just
read
up
on
the
process:
it
starts
with
volunteers
and
then,
when
we
have
gaps
we
just
need
to
start
filling
in
the
gaps.
It's
I,
don't
think.
There's
too
much
overhead,
though
all
right
see
reviewing
12,
so
12
coach
Reeves
is
effectively
in
17th,
which
is
later
next
week.
States
next
weekend.
So
just
take
a
look
at
milestones
assigned
to
you.
If
there's
anything
that
we
think
will
slip,
then
let's
talk
about
it,
we
can
move
accordingly.
A
There
were
only
two
issues
when
I
look
this
morning
that
we're
on
signed
and
those
were
issues
I
think
that
were
made
up
during
this
milestone.
So
if
those
need
to
move
on
a
big
deal,
I
think
those
were
owned
by
Camille
and
now
we'll
chat
with
him
when
he
gets
back
any
questions
on
that
or
a
move
on
no
cool,
so
12:3
with
the
kickoff
earlier
this
week.
The
main
goal
for
our
team
is
still
Puma
getting
it
out
into
production.
We
have
deployed
dad.
We've
deployed
the
ops.
A
The
next
stages
are
staging
than
canary
then
production
and
the
goal
is
to
have
it
by
the
end
of
12
3,
which
would
be
effectively
September
22nd
so
seems
reasonable,
but
those
are
all
dependent
steps
right.
Each
stage
it
depends
on
the
previous
stage
being
successful,
so
everything's
looking
good
so
far,
we'll
see.
Hopefully
we
don't
run
in
your
bumps
in
the
road
but
see
412
3.
We
have
issues
the
span
two
different
projects,
so
sadly
we
have
to
look
at
two
different
boards,
but
they're
set
up
and
linked
here.
A
A
There
are
some
late-breaking
hype,
Oreo
high
priority
issues
that
have
come
out.
One
of
them
just
came
out
this
morning
and
I'll
have
any
details
for
it
yet,
but
based
on
the
discovery
of
those
two
issues,
so
we
have
49
51,
seeing
it
overhead
for
our
queues
and
then
there's
a
recently
discovered
issue
where
background
processes
are
repeatedly
getting
created.
A
It's
causing
issues
that
I
need
to
take
a
look
at
right
after
this
meeting
and
start
getting
people
involved
that
could
offset
everything
we're
doing
for
12,
3
and
possibly
even
some
of
the
stuff
we're
doing
right
now.
So
I
need
to
gather
some
more
information
and
I'll
get
back
to
y'all,
let
you
know
what's
going
on
okay
and
then,
as
for
waiting.
So
last
time
around,
we
tried
waiting
everything
in
the
pants
for
12
to
have
everybody.
Take
a
look
at.
A
It
got
a
lot
of
feedback
that
that
wasn't
the
desired
way
to
do
it
so
I
think
this
time
around
we'll
try
waiting
as
folks
pick
up
the
work.
Part
of
the
reasoning
for
that
is.
We
have
weighted
items
that
are
in
12
3
that
we're
waited
a
couple
weeks
ago
and
I
found
that
when
you
point
or
weight
something-
and
then
it
sits
there
for
weeks
and
even
months
to
value,
it's
really
no
longer
relevant
because
new
items
could
have
come
up.
Someone
else
could
take
it
on
so
we'll
try
something
different
this
time.
A
A
See
the
memory
team
retro,
sorry
async,
retro
process,
a
new
issue
is
automatically
created
by
the
bots.
It's
out
there
as
things
come
up
throughout
12,
as
we
still
have
a
couple
weeks
to
go
here.
You
start
adding
your
thoughts
like
we
did
last
time.
I
think
we
did
a
really
good
job
last
time
of
adding
things
as
they
were
top
of
mind,
and
we
had
a
lot
of
good
conversation
points
in
12
and
this
time
we'll
try
and
find
a
time
that
works
to
have
a
synchronous
retro.
A
B
Just
a
quick
paper
form
our
ends.
The
big
thing
we
work
on
this
week
is
switching
up
the
framework
can
do
it
a
big
refactor,
because
we've
had
that
the
original
tool,
though
using
our
Tories
non-performance,
especially
formerly
French,
to
be
searched
online,
is
actually
the
second
works,
a
little
testing
tool
in
the
entire
industry.
B
So
that
was
a
surprise
to
see
that
and
that
with
that
comes
a
lot
of
actual
extra
noise,
so
actually
will
be
what
the
comparators
error
in
this
episode
was
not
only
very
good
at
actually
the
performance
testing,
which
was
ironic,
though
it's
also
very
bad
at
actually
hand
in
itself,
but
wasn't
able
to
report
correct
figures
and
stuff
the
more
pressure
you
put
on
it.
So
the
knowledge
of
Eno
actually
seems
to
be
in
current
or
heavily
implanted
by
multiplier,
say
I
think
is
actually
they
have
your
end
points.
B
We
saw
artillery
pouring
very
high
numbers
and
I
think
now
that
she's
blue
closer
till
he
wasn't
able
to
cope.
So
he
was
switching
one
to
a
no-till
called
k-6,
which
is
much
more
more
more
recent
and
much
more
loved
in
the
industry
and
as
much
more
performant
to
give
some
context
and
the
previous
iterations,
we
had
to
run
ten
separate
brothers
of
artillery
to
get
200
responses
per
second
into
our
biggest
environment,
which
was
ridiculous.
That
I
shouldn't
be
happening.
B
Eternity
s
is
nothing
compared
to
what's
our
tool,
so
the
new
tool
want
to
turn
our
PS
and
one
runner.
So
what's
a
piece
of
marentina
sub
tool
once
this
is
committed
which
hopefully
should
be
soon,
you
should
see
better
and
more
consistent
results
from
little
testing
an
instance.
Of
course,
the
hook
talk
to
Griffin,
which
is
good.
This
is
only
for
the
k-6
to
only
reports
on
what
it's
seen
in
terms
of
test
results
and
still
show
something
that
tells
you
how
silhouetting
points
are
and
will
be
interesting.
B
Those
never
starts
a
lot
bad
least
on
this.
Your
final
incidence
is
that
you
can
maybe
go
back
and
look
at
it
transfer
the
last
week,
community
and
stuff,
like
that,
so
I'm
still
kind
of
working
on
that.
So
once
that's
a
well
let
you
know,
but
next
week
and
the
following
weeks,
we'll
just
because
a
native
iterating
this
is
this
big
switchover
happened.
So
it's
what's
called
a
how.
B
It's
already
all
done,
oh
yeah,
and
so
the
things
that
they're
actually
test
results
of
k6f
for
that
it
was
about
three
few
days.
Effort
from
my
part,
I
don't
give
it
take.
I
still
not
finished
yet
betrayed
dogs.
Those
three
days
like
fill
effort
for
the
scripting,
and
do
we
didn't
learn
in
the
tool?
Knowledge,
as
I
have.
C
B
Well,
we
need
to
review
it.
Well
then,
when
I
say
inflates,
its
water
till
I
was
doing
so.
Teller
was
technically
reporting
stuff.
Currently
it's
just
the
numbers
were
just.
It
was
campus
miscalculated
numbers.
So
when
it
said
there
was
a
slow
point
that
end
point
is
slow
and
they
still
is
slow
in
case
it.
The
walk.
Everything
you
said
was
60
seconds,
P
95
for
a
ten
point
case.
X
has
come
back
as
the
ID.
B
No
it's
more
like
five
seconds
perfect,
but
if
is
drastically
bad,
it
really
was
I
was
surprised
by
the
persistent
numbers
about
double-checked
everything.
Everything
is
correct,
and
so
artillery
was
just
really
not
good.
That's
the
summary,
but
so
the
end
points
that
we've
called
there
are
slower,
so
slow,
so
those
issues
are
still
valid.
I
won't
update
them
with
the
right
numbers
you're
right,
though,
as
it
gets
cold.
Okay,.
A
Good
to
know
some
of
those
endpoints
that
were
called
out
over
like
three
or
four
right:
they
were
kind
of
on
the
scale
you
were
talking
about,
like
I,
think
what
I
remember
I
like
forty
some
seconds
on
some
of
them.
So
if
K
six
comes
back
and
says
it's
more
like
five
or
six,
that
might
a
few
prioritize
some
of
the
work
that
we
need
to
do
you.
B
Could
argue
for
that
to
be
deeper
or
die
spot
five
to
five
seconds
is
still
very
slow
for
an
endpoint
news
either,
and
so
that's
a
fair
question,
though
it's
a
fair
thing
to
discuss
to
say
it's
a
five
second
endpoint
the
end
of
the
world
for
customer
simply
policy.
Yes,
but
you
know,
obviously
things
still
need
to
be
prioritized
and
obviously
too
much
the
priority
here.
I'd
say
so.
Certainly
the
memory
intensive
endpoints
I
think
probably
have
a
double
whammy
to
be
prioritized
higher
I
know,
carelessly,
spoke
and
called
a
few.
B
B
It's
really
starts
just
to
act
very
weirdly,
so
I'm
gonna
have
a
it's
called
the
Federation,
so
I'm
gonna
self,
and
you
from
used
to
server
that
collects
the
relevant
metrics
from
the
various
environments,
all
in
Prometheus
service
centralizes.
It
and
thank
you.
Fine
I-
can
use
that
as
well
as
the
the
tetras
also.
We
should
hopefully
start
literally
begging
them
14.
B
To
give
me
the
access
they're
taking
the
time
for
some
reason,
other
than
a
wife
tiniest
part
of
the
multiple
tankers
yeah,
or
is
this
on
the
pre
puma
environments
and
it
is
to
get
access
to
the
testbed
project,
which
I
am
told,
should
give
me
access
to
the
premium
people
advice
we
already
have
access.
The
problem
is
that
at
katana
Prometheus
needs
on
earth
needs
clean
access
that
right
now,
you
need
to
authenticate
with
from
your
Google
account
chopsticks,
not
good
for
Prometheus
and
katana.
C
Something
to
keep
in
mind
that
Puma
could
be
redeployed
by
kind
of
older
release.
I
know
you
said
that
before
Camille
was
on
vacation,
he
updated
it
with
my
latest
fixes
and
after
that
they
who
is
all
the
release.
So
some
fixes
we
made
to
metrics
could
disappear
so
something
to
keep
in
mind
to
check
the
help
page
and
check
the
version,
because
it
was
kinda
rolled
back
from
the
custom
configurations
that
made
by
Camille
I.
Don't
know
who
is
responsible
for
updating
people,
one
button
and.
B
The
good
thing
in
the
new
script-
I've
written,
is
actually
pulls.
The
new
version,
so
in
the
test
is
also
you
couldn't
slap
links
there.
I
should
call
it
the
specific
version
that
the
environment
is
running
on
freeing
people
my
book
running
on.
Obviously
it
was
a
much
the
same
but
running
on
twelve
point.
One
point
for
enterprise:
there
was
an
option
there.
I
could
I
could
add
the
reference
number
two.
Do
you
want
the
shower
added
to
that
as
well?
Would
that
be
useful
for
you.
B
Yeah
I
mean
which
is
usually,
if
you
like,
to
get
the
help
of
the
department
just
to
call
it
because
it's
it
can
be
confusing.
For
me,
we
don't
know
what
each
environment
is
running.
Certainly,
I've
made
some
comment
on
a
few
issues
about
that
kind
of
thing.
Recently
and
I've
said
when
the
questions
been
asked
about
what
bird
this
environment
has.
Should
it
be
updated
either
I've
set
the
same
asset.
Each
time
said
they
should
be
updated
automatically.
B
It
should
be
a
big
weekly,
if
not
earlier,
maybe
daily,
if
that's
possible
as
much
as
reasonable
possible,
and
if
we
do
need
to
do
a
test
against
a
specific
version.
Then
that's
when
we
come
in
and
manually
or
the
tool
we
pause
it.
We
factor
to
the
right
version
test
it
and
then
bring
it
back
on
to
latest
because
we
want
to
be.