►
From YouTube: Envoy Community Meeting - 2018-07-17
Description
Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io
Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects
C
D
Think
in
parallel
to
this
discussion
is
like
Basile,
remote
cashing,
which
would
give
us
like
a
huge
win,
I
think
or
even
just
getting
circle
CI
to
cash,
build
objects
and
it
can
actually
do
you
know
cash
and
restore
and
I
feel
it
just
needs
someone
to
go
in
and
investigate
and
we'll
probably
you
know,
save
like
I,
don't
know
80%
of
build
time.
If
we
do
that,
how.
B
D
D
Yeah,
it's
like
a
content.
Addressable
storage
I
mean
turn
anything
basil,
that's
kind
of
a
cow
as
a
represents
its
own
cash,
although
you'll
be
forgiven
for
thinking
otherwise,
given
how
it
deals
with
you
can
use
to
build
flags,
but
yes
in
principle,
it's
exactly
how
it
works.
It's
like
the
command
line
and
the
status
of
every
have
all
its
inputs
and
hashes
and
hash
out
of
that.
Basically,
so.
D
D
These
unsaid
he's
actually
had
some
experience.
You
do
his
personal
build
this
way
and
it's
pretty
good.
We
have
an
internal
meeting
with
you
know,
not
internal
me.
John
will
be
there
as
well,
but
with
the
folks
interested
in
basil
on
comparing
that
with
the
actual
remote
build
execution
and
caching,
which
is
going
to
helpful
from
Google,
and
we
could
potentially
use
that,
but
I
think
the
performance
I
think
of
using
GCS
would
still
be
probably
way
better
than
what
we
have
today.
So,
okay.
E
D
B
D
E
B
B
You
know
it
seems
like
we
would
even
benefit
from
some
basic
graphs
or
or
metrics
around,
like
how
many
lines
of
code
are
we
building
over
time
like
how
long
is
CI
taking,
because
you
know
I
suspect
that
we've
increased
the
number
of
lines
of
code
by
a
ton,
and
there
may
have
been
a
couple
of
regressions
here
and
there
around
build
performance,
but
I
think
combination
of
them
is
essentially
making
the
build,
probably
2
to
X
or
two
and
a
half
acts
is
slow,
is
when
we
first
switch
to
two
circle.
One.
D
B
B
Is
that
circle,
their
instant
size,
basically
max
out
right
now
at
what
what
we
have
and
those
are
effectively
eight
core
I
think
they
have
like
16
or
8
8,
gigs
of
RAM,
and
obviously
just
the
way
that
we
do
the
builds
like
we
could
saturate
a
32
core
machine
easily
and
my
my
concern
is
that,
like
in
the
future,
as
the
lines
of
code
go
up,
we
really
have
I.
Think
so.
There's
caching
for
sure,
but
then
I
think
we
have
two
options
like
option.
B
D
So
one
thing
that's
coming
into
alpha
right
now
is
remotes
build
farm,
which
will
essentially
be
something
that
we
supply
to
put
it
to
the
on
board
project
and
we'll
just
be
able
to
use
remote,
build
workers
running
off
in
a
cloud
environment
outside
of
circle
CI
to
do
the
work
and
I
think
back
to
basil
and
hand
back
objects,
and
things
like
that.
I
see.
B
B
I
mean
that
that
sounds
amazing.
My
only
concern
is
last
time
that
we
did
this
I
think
we
talked
to
them
and
there
was
all
these
security
concerns.
We're
like
we
had
to
like
opt
in
to
the
builds
for
people
and
like
it
sounded
really
horrible,
so
I
just
didn't
know
if
that's
gonna
be
fixed
or
not,
I
think
yeah.
D
B
D
D
B
I
can
work
with
him
directly
to
help
him
set
up
whatever
he
needs
like
that.
That
part
is
super,
easy,
so
he's
not
on
the
call,
but
we
can
talk
to
him.
So,
if
he's
willing
to
do
that-
and
he
already
kind
of
knows
what
to
do,
that
sounds
really
great,
because
I
also
tend
to
agree
that
I
feel
like
in
most
of
our
builds
the
actual
object
files.
Don't
don't
change
so
like
I'm,
guessing
caching
would
be
very
effective,
specifically.
B
D
B
I
I
was
just
actually
so
the
the
reason
that
I
bring
that
up
is
I'm
I'm
curious.
Actually,
if
that
would
change
bill
performance
at
all,
because
if
we're,
if
we're
generating
intermediate
code,
only
like
for
the
individual
compiles
and
then
only
optimizing
or
linking
basically
at
the
final
link
time
like
I,
guess,
I'm
curious.
B
If,
for
example
like
even
in
the
release
build
like,
if
we
were
compiling
with
FLT
Oh
for
the
tests,
you
could
basically
potentially
compile
them
without
optimizations,
but
for
the
binary
you
could
link
the
binary
with
like
full,
full,
full
optimization
or
something
like
that
and
I'm
just
wondering
if
that
would
take
less
time.
I.
D
D
B
Know,
but
that's
that's
that's
something
that
indirectly
occurred
to
me
that
we
might
like
it
just
seems
like
we
should
turn
on
FLT,
oh,
and
we
don't
build
with
that.
Currently.
But
that's
a
that's
a
separate
thing.
So,
ok,
all
right!
Well
so
I
guess.
B
No
great!
Oh,
let's
see
before
you
leave
Chris
did
you
want
to
talk
about
on
vikon
at
all
or
no.
C
B
Ok
will
do
envoy,
announced
I
think
is
still
broken,
I
think
only
a
couple
people
can
send
email
to
it.
So
I'm
a
lot
I'll
fix
that
No
I'm
sure
more
people
are
in
Devon
users-
anyway,
that's
probably
true
yeah,
so
maybe
just
forward
it
to
users.
Okay,
cool
all
right
looks
like
Ben
wanted
to
discuss.
E
I'm
here,
yeah
hi,
everybody
I'm
Ben
plotnick
from
Yelp
we're
in
development
with
envoy,
where
we're
a
smart
stack,
H,
a
proxy
shop
looking
to
kind
of
swap
out
the
synapse
H,
a
proxy
portion
with
its
Envoy
yeah.
So
I
opened
this
issue
talking
about
the
problem
of
hot,
restarting
across
hot
restart
versions.
E
We've
written
a
few
blogs
about
how
we
do
this
in
synapse
H
a
proxy.
But
basically
the
idea
is
that
you
do
originally.
We
did
a
server
use
port,
there's
some
race
conditions
potentially,
and
so
we
have
some
kind
of
hacks
around
it,
but
basically
like
it
seemed
like
other
people
are
not
running
into
this
problem,
which
was
surprising,
I,
guess
but
yeah
again.
B
Yeah
I
mean
like
I,
can
tell
you
what
we're
doing
it,
lift
which
is
not
like
it's
not
optimal,
but
it,
but
it
works
and
and
and
then
we
can
discuss,
you
know
other
options
at
lyft.
What
we
do
and
I'm
I'm
happy
to
share
this
script,
though
it's
pretty
obvious
is
effectively
when
we
go
to
deploy
envoy.
It
compares
the
hot
restart
version
of
the
current
running
binary
with
the
with
the
new
binary
and
a
couple
other
things,
and
it
basically
determines
whether
it
can
hot
restart
or
or
not.
B
If
it
cannot
restart
it,
it
does
that
if
it
can't
hot
restart,
we
have
a
rolling
deploy
system,
so
it
basically
goes
into
a
mode
where
it'll
lock
itself
using
at
CD
and
then
it'll
basically
do
a
rolling.
You
know
it'll
drain
Forex
period
of
time,
restart
and
then
that
that
role
will
go
through
and
you
know
for
for
what
it's
worth.
Given
that
you
know
I
would
say.
On
average
lately
we
probably
have
a
incompatible
hot
restart
version,
every
six
to
nine
months,
like
it
just
hasn't
matter
that
much
and
and
that
system
works.
E
That,
frankly,
we
have
a
similar
system
for
rolling
deploys
and
we
could
definitely
do
this
operationally
it.
It
makes
it
a
little
bit
complex,
because
one
is
so.
We
can
do
this
like
hot
restart
version
compare
but
the
we
we
use.
You
know
system
D
for
our
to
run
envoy
and
it's
a
little
bit
hard
to
kind
of
tell
it.
Ok
now
actually
call
back
and
do
a
hot,
like
the
coordination
of
all
these
moving
parts
is
a
little
bit
difficult.
B
Well,
there's
yeah,
sorry,
so
there's
there's
two
there's
two
parts
just
coming
back
to
system
D
the
the
one
thing
that
I
would
say
is
we
we
have.
We
have
similar
problems
and
the
way
that
we
get
around.
That
is
that,
because
we
effectively
run
envoy
through
the
kind
of
hot,
hot
restart
rapper
in
in
the
hot
restart
case,
like
system
D
or
for
us
we
use,
we
use
run
it
like
it's.
It's
only
aware
of
that
rapper.
So
it
doesn't
really
know.
B
What's
what's
going
on
under
the
hood
and
then
only
in
the
case
where
it's,
where
it's
basically
doing
that
full
restart.
You
know
you
would
do
the
drain
and
then
we
effectively
just
say
run
it
run
it
restart
or
you
you,
you
would
do
a
system.
Do
you
restart
and
just
restarts
the
entire
stack
so
like
I,
you
know,
it's
probably
out
of
scope,
but
like
I
would
be
curious
to
know
what
the
problem
are
with
with
systemd
there,
because
it
seems
like
it
would
work
and
the
only
reason
that
I
bring
that
up.
B
Is
that
like
what
what
we're
about
to
talk
about
with
the
light
hot
restart,
you
know
it?
It
makes
sense,
but
it's
still
like
it's
gonna
be
non-trivial
like
the
right
and
it's
like
for
something
that
happens
like
every
like
six
to
nine
months,
like
it's
not
clear
to
me
that
it's
a
good
use
of
time,
but
like
I'm,
not
I'm,
not
really
opposed
to
it
right.
So
that's
that's
kind
of
why
I'm
asking
those
questions.
Yeah.
B
So
not
not
necessarily
because
what
we
do
with
this
script
is
we
do
everything
through
the
local
envoy
admin
port,
so
will
basically
tell
envoy,
through
the
admin
port
to
start
draining,
then
will
basically
sleep
and
then
we'll
tell
envoy
to
quit,
and
then
the
supervisor
process
realizes
that
it
quit
and
it
just
and
it
just
restarts
it.
Oh
do.
B
We
we
just
start
sorry,
we
call
health
check,
fail
on
the
local
envoy
to
have
envoy
fail
incoming
connections
so
that
all
of
the
callers
stop
sending
requests.
We,
you
know
we
have
that
time
box
based
on
our
time,
outs
and
various
other
things.
So
for
internal
mesh
traffic,
we
not
for
I,
don't
know
what
the
current
time
is
3060
seconds
or
something
like
that
and
then
and
then,
when
that
time
elapsed,
is
we
basically
just
just
kill,
kill
the
envoy
and
then
so
we
don't
even
involve
run
it
our
system
V.
B
E
B
E
D
E
It's
it's
kind
of
prone
to
failure.
So,
even
though
this
is
like
a
rare
thing,
it's
it's
definitely
something
that
could
like
take
out
the
entire
sure,
the
entire
fleet.
If
it
doesn't,
it
doesn't
work
well
and,
and
even
more
importantly,
like
its
its
board
like
a
sales
and
marketing
thing,
we
have
the
system
working
right
now,
so
we
have
to
be
able
to
sell
to
the
rest
of
the
operational
the
community,
like
in
our
Yelp,
that
that
this
works
yeah.
B
Okay,
I
mean
like
from
the
from
the
light
restart
perspective.
I
think
that
a
couple
people
had
comment
on
that,
including
Harvey
and
Greg.
But
you
know
I
guess
we
can.
We
can
open
it
up
for
other
people
to
chat,
but
you
know
it
makes
sense
to
me
to
have
like
a
simpler
version.
Where
you
know
we
don't
do
any
of
the
shared
memory
stuff.
We
can
have
like
a
simpler
protocol.
If
we're
gonna,
do
this
one
more
time
agree
with
Harvey.
B
A
You
know
I,
don't
think
it
would
actually
be
that
hard
to
implement
based
on
exactly
the
code
we
have
now.
Just
don't
do
the
stats
part
yeah
and
just
you
don't
even
need
to
change
the
protocol
really.
Just
that's
quite.
D
A
B
E
B
Like
and
it
and
it
doesn't
have
to
be
long
like
you
know,
one
one,
two
pages,
just
kind
of
like
what
what
the
goals
are
like
what
you
would
change
and
then
you
can
just
send
it.
You
know
to
that
ticket
and
then
we
can
all
comment
on
it:
cool,
yeah
and
then
in
order
to
get
it
approved,
you
have
to
add
your
logo
to
the
website.
E
B
B
Cool
that's
exciting
Greg,
Harvey
Alissa.
Anyone
else
have
anything.