►
From YouTube: Intro to Lodestar - Cayman Nava
Description
This talk was given at IPFS Camp 2022 in Lisbon, Portugal.
A
So
about
me
my
name
is
Cayman
I
work
on
lodestar
at
chainsafe
and
at
Twitter
and
GitHub.
These
are
some
pictures
that
you
might
see.
If
you
see
me
online
so
lodestar
lodestar
is
a
project
started
and
maintained
by
Chain
Safe
Systems
we're
a
we're,
a
development
company
out
of
Toronto,
but
we've
got
developers
and
people
globally.
So
this
is
basically
this
our
company's
project
for
for
helping
ethereum
trying
to
build
a
public
good
and
word
heavily
using
Js
lipp2p.
A
So
this
project
load
star
it's
a
typescript
and
assembly
script
ecosystem.
We
run
everything
open
source,
it's
very
Community
oriented
project
with
all
of
our
planning
and
all
of
our
meetings
are
done
in
the
open
public.
If
anyone
anyone
wants
to
join,
you
know
we
we
let
them
join.
We've
got
a
ethereum
consensus,
client,
so
Beacon
node
and
the
validator
that
are
running
in
production,
but
I,
don't
know
so
some
some
a
number
of
validators,
maybe
in
the
hundreds
now
that
are
running,
running
load
star
on
on
the
ethereum
mainnet.
A
We
got
a
lot
of
packages
we
try
to
on
npm.
We
try
to
like
I,
don't
know
view
the
project
as
a
public
good
and
try
to
try
to
be
putting
putting
back
into
open
source.
Like
you
know,
if
there's
like
a
I,
don't
know
something
that
some
package
that
needs
maintaining
or
needs
to
be
built,
or
you
know
we
basically
do
it
whatever
we
need
to
do
to
to
to
make
it
to
make
a
our
product
work.
A
You
can
see
us
on
GitHub
there's
our
link
so
before
I
get
into
how
we're
using
and
using
and
using
jslip
P2P
I
just
want
to
talk
about
some
of
the
unique
challenges
of
our
of
our
project.
A
So
this
is
not
like
a
normal
web
app
or
you
know.
This
is
not
your
normal
use
of
JavaScript,
so
this
is
like
high
performance
JavaScript.
If
you
can
call
it
that
we
have
a
lot
like
pretty
decent
bandwidth
requirements,
we've
got
a
lot
of
like
you
know.
Networking
that's
happening
a
lot
of
different
peers,
sending
a
lot
of
different
types
of
messages
and
there's
a
lot
of
computation.
A
A
lot
of
things
are
being
processed
all
at
once,
and
you
know
all
I
just
say
if
we,
if
it
goes
well
where
our
our
software
is
making
money
and
if
it's
going
poorly,
it's
losing
money
and
I
mean
a
lot.
A
lot
of
you
know:
I
guess
that
could
be
said
for
a
lot
of
like
businesses
and
business
software.
A
Here's
you
can
see
some
graphs
of
of
kind
of
our
bandwidth
over
time
over
the
course
of
a
past
day,
it's
roughly
1.3
megabytes
per
second
inbound,
and
probably
half
of
that
is
just
gossip.
We're
getting
you
know
possibly
bombarded
by
a
bunch
of
gossip.
A
So
some
of
the
networking-
here's,
like
total
number
of
connections,
this
some
graph
Suite.
This-
was
on
a
beta
beta
node,
so
that
the
connections
are
not.
Things
are
looking
a
little
choppier
than
they
would
normally.
But
you
know
roughly
50
50
connections.
A
A
If
you
can
see
in
this
this
first
graph,
this
is
looking
at
the
utilization,
CPU
utilization,
I,
guess
yeah
CB
utilization,
based
on
what
things
happening
inside
of
our
running
system.
It's
mostly
like
I,
want
to
say
upwards
of
80
is
like
BLS
signature
verification,
and
you
can
see
on
the
right
here.
We've
got
a
few
different
cues,
where
we,
where
we
validate
different
types
of
gossip
and
we're
processing,
you
know
thousands
of
attestations
per
slot.
A
You
know
hundreds
of
aggregated
attestations,
there's
just
a
lot
of
a
lot
of
things
happening
kind
of
all
at
once
here,
and
you
know
if
it
goes
well
then
we're
when
then,
then,
you
know
this
blue
line
goes
down,
look
lower
and
lower.
We
want
to
be
missing
a
small
number
of
we
want
to
be
getting
as
much
money.
That's
on
the
table
for
our
validators
as
possible.
A
So,
where
that
comes
into
like
abusing
Json
P2P,
so
the
the
you
know
due
to
kind
of
these
unique
challenges
of
our
application.
We
stress
test
loopy
to
pee
a
lot
and
I.
Think
I'd
say
that
our
project
probably
is
doing
I
mean
probably
the
the
heaviest
user
of
Jason
b2p.
As
far
as
like
pushing
the
boundaries
of
what
what
it
does
I
I,
don't
think
we
use
all
the
features
of
jslip
P2P.
We
don't
use
the
the
DHT,
we
don't
use
the
relay.
A
We
only
use
TCP
transport,
but
what
we
do
use
we
like
really
really
use
and
we
contribute
back
to
jslip
utility,
a
lot
where
we're
always
pushing
PRS,
and
you
know,
and
myself
I'm
a
I'm
actually
a
maintainer
of
of
JS,
the
P2P
so
we're
we're
committed
to
pushing
the
ecosystem
forward.
What
we
found
is
that
it's
not
really
easy
to
to
load
test.
A
A
But
as
of
right
now
like
we,
don't
we're
not
really
able
to
test
other
we're
not
really
able
to
test
it
in
a
in
like
a
unit
test
kind
of
a
thing
like
we
need,
we
actually
need
the
load
of
a
real
Network
to
be
able
to
see
what
the
problems
are
or
we
you
know
we
are
not
able
to
see
the
same
things
in
in
this
like
sandbox
environment,
so
the
two
tools
we
really
use
really
rely
on
us
hooking
into
the
software.
A
That's
running
in
production
and
those
two
things
are
metrics
and
CPU
profiling,
so,
like
metrics,
that
was
basically
I've
been
showing
these
different
graphs.
These
are
all
different
metrics
that
we
have.
We've
got
a
bunch
of
different
dashboards,
but
these
metrics
are
really
invaluable
for
providing
visibility
into
like
complex
software.
That's
running
so,
if
you
have
any
kind
of
software,
that's
has
any
scale
to
it.
A
You're
really
going
to
want
to
this
is
something
that
you
I,
really
highly
recommend
you
you
invest
in,
because
you're
you're
not
going
to
be
able
to
get
everything
through
logs
or
aggregating
logs
you're
going
to
need
to
see
really
everything,
that's
happening,
and
so
things
like
caching
cache
sizes.
That's
that's
a
really
Prime
candidate
for
a
metric
things
like
how
long
certain
tasks
take.
A
That's
that's
another
pretty
prime
prime
thing.
So
once
you
start
measuring
a
bunch
of
things,
it's
a
lot
easier
to
know
when
something
does
go
wrong.
How
it's
going
wrong!
You
can
see.
Oh
there's
a!
We
have
a
chart
that
measures,
the
duration
of
the
heartbeat
and
heartbeat
is
now
spiking.
It's
like.
Oh
something
happened,
and
if
you
didn't
have
these
metrics,
you
wouldn't
know
where
something
is
heartbreaking
or
where
a
regression
is
happening.
A
So
it's
it's
pretty
vital
and
then
the
other
thing
that
we
we
do
a
lot
of
is
we.
We
are
doing
a
lot
of
CPU
profiles.
A
So
once
once
a
metric
has
kind
of
alerted
us
that
something
is
wrong,
then
we
start
digging
in
and
if
it's
a,
if
it's
a
performance
problem,
we
can
run
a
CPU
profile
and
see
see
where
the
hotspot
is
so.
This
is
an
example
of
something
that
actually
happened.
A
Five
days
ago,
it
turned
out
that,
seeing
if
a
string
is
a
valid
IP
who
became
a
performance
issue
due
to
some
dependencies
upgrading,
unfortunately,
in
the
process
of
trying
to
prevent
a
denial
of
service,
they
created
a
denial
of
service
in
our
application.
A
So
I
guess
the
the
just
the
you
know
the
work.
The
kind
of
the
headline
is,
you
know:
JavaScript
lip
p2b
is
getting
faster,
better
and
stronger
in
some
part,
we're
helping
in
some
part
to
make
that
happen
and
yeah
we're
kind
of
pushing
pushing
what
it
what
it,
what
it
was
initially
designed
to
do,
we're
trying
to
kind
of
productionize
it
as
much
as
we
can
so
I
I
guess:
I
can't
really
take
any
questions
but
yeah.
A
That's
that's
all
I
got
where
if
any
of
this
sounded
really
interesting,
we're
hiring
for
new
Engineers.
Anyone
interested
in
the
networking
side
we're
hiring
the
networking
engineer
drop
into
our
Discord
or
just
like,
send
me
a
DM
on
on
on
Twitter.
These
are
Twitter.
We
also
have
a
low
star
Twitter.