►
From YouTube: 2022-03-09 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
B
A
A
D
A
Just
pinging
shawn,
I
think
he
wanted
to
talk
about
collector
stuff.
A
A
Cool
well,
I'm
happy
to
to
just
get
started
just
like
an
update
about
the
whole
tc
sponsorship
thing
that
I
was
talking
about
last
week.
So
I
made
a
pr
to
get
this
officially
added
to
the
spec.
The
idea
that
sigs
and
also
really
any
like
oh
type
or
major
change,
should
like
get
a
sponsor
before
gets
too
deep
into
it.
A
A
He
has
to
rearrange
some
meetings
in
order
to
actually
show
up
to
this
meeting,
but
he's
going
to
and
he'll
start
getting
up,
speed
so
feel
free
to
reach
out
to
him
on
slack
ping,
him
in
the
the
instrumentation
channel
about
stuff
and
he'll
help
get
things
rolling
forwards
bogdan's
on
vacation
this
week,
but
he
is
interested
in
talking
more
about
links
and
things,
unfortunately
he's
in
europe.
A
So
I
think
it'll
be
difficult
for
him
to
come
to
this
meeting,
but
I'm
trying
to
get
him
to
go
to
the
thursday
eu
instrumentation
meeting
and
I
can
try
to
rep.
We've
got
some
crossover,
I
think,
between
this
group
and
that
group
lumila.
I
know
you
go
to
both.
A
So
hopefully
we
can
have
some
discussion
there
about
like
some
of
the
general
issues
that
are
being
brought
up
so
yeah
just
in
general.
Hopefully
all
of
that
should
make
things
go,
go
a
lot
smoother
going
forwards.
C
D
A
Yeah
yeah,
we
have
to
hear
back
from
the
tc.
I
think
they
have
their
meeting
tomorrow,
they're
going
to
like
discuss
the
stuff
more
in
depth,
but
riley
said
definitely
for
sure
he's
willing
to
come
to
this
meeting
in
particular
and
bogdan.
I
don't
think,
can
come
to
this
meeting,
but
he's
definitely
interested
in
getting
all
like
the
link
stuff
resolved.
So
hopefully
we
can
get
get
that
stuff
kind
of
kicked
out
soon
right,
but.
C
A
Yeah
yep
so
yeah,
that's
that's.
I
think
all
all
there
is
to
say
about
that.
A
A
I
threw
retries
and
spam
links
on
here,
because
I
know
those
are
two
two
open
issues
that
we
have.
I
think
the
main
retards
thing
is
ready
to
get
merged
at
this
point.
C
It
isn't
yeah
bogdan
was,
was
ready
to
actually
merge
it,
but
cjo
just
came
like
a
recently
last
minute
and
prevent
him
to
to
press
the
button.
A
Well,
I
think
you
could
ping
tigrin
should
should
be
able
to
merge
this,
for
you,
tigran's.
A
And
then
for
links
there
was
some
discussion
going
on
there.
It
looked
like.
A
So
I
don't
know
honestly
how
much
progress
we
could
make
on
that
until
we
like
sit
down
and
talk
to
him,
but
you
might
want
to
respond
to
some
of
the
stuff
like
might
want
to
just
keep
the
conversation
going
in
that
that
pr,
because
he
won't
be
able
to
come
to
this,
like
particular
meeting.
C
Sure
yeah,
I
will
I
will
make
this
currently
it's
draft,
so
I
will
publish
it
once
the
like
a
previous
or
the
another
one
merged
and
yeah.
We
will
try
to
to
get
this
conversation
on
going
in
in
this
pr,
but
definitely
it
will
be
good
to
to
have
like
an
in-person
conversation.
I
mean
this
zone
conversation
happening,
maybe
this
week
or
next
week
I'll
try
to
do
that.
A
Yeah
yeah,
he
he
had
some
yeah
some
questions
in
there
about
documenting,
like
like
link
relationships,
which
is,
I
think,
an
idea
that
we've
definitely
brought
up,
which
I
think
is
a
good
idea
personally,
because
we're
going
to
be
using
links
for
like
pretty
different
things
like
these,
like
using
links
for
retries
and
using
links
for
like
had
I
put
links
for
things
that
are
basically
within
the
same
trace
versus
links
that
are
being
used
to
join
across
a
bunch
of
different
traces.
A
A
So
one
reason
why,
having
like
a
relationship
switch
or
you
know,
link
type
would
would
probably
help
back
ends
be
able
to
change
change
their
behavior
based
on
what
kind
of
a
link
it
is,
though,
I'm
curious,
I
think
I
don't
know
ludmilla
if
you
know
the
details,
but
I
know
just
azure
is
like
the
only
group
I
know
of
that's
actually
like
implemented
links.
B
I
I
cannot
comment
on
costs.
I've
tried
to
find
someone
to
comment
on
this.
Maybe
I
should
push
harder,
but
the
main
cost
is
that
I
think
with
links
you,
you
have
to
have
some
sort
of
indexer,
so
you
don't
query
a
lot
so
and
this
service
is
not
super
straightforward.
It's
a
it's
a
big
deal
to
implement
it.
Let
me
see
if
I
can
find
someone
who
can
comment
on
this
better.
A
Yeah
I
feel
like
since,
since
y'all
have
like
done
the
work,
presumably
to
do
this
kind
of
robustly
before
we
set
anything
in
stone,
you
know
if
the
the
back
end
people
over
at
azure,
who've
implemented.
This,
like
you
know,
have
wishes
about
links
and
like
how
they're
implemented.
If
there's
like
anything,
any
commentary
they
have
about.
Like
you
know
how,
like
anyway,
we
could
change
the
data
structure
that
would
actually
make
things
significantly
better
for
them.
A
It
would
be
good
to
do.
Do
a
last
call
on
that.
B
Do
you
feel
like
if
somebody
can
come
here
to
this
meeting
and
share?
Would
it
be
helpful.
A
I
think
so
I
mean
I
would
certainly
find
it
helpful,
because
you
know
we
have
not
tried
to
implement
links
over
lightstep
yet
so
I
don't
have
anyone
personally
to
talk
to
who's
actually
like
tried
to
do
the
dirty
work
of
bolting
this
on
to
like
their
existing
tracing
system,
and
so
I'm
curious.
If
there's
anything
about
it,
that
is,
like,
I
don't
know
insufficient
in
some
way.
A
You
know
I
assume
you're,
basically
making
like
a
giant
join
table
of
some
kind,
but
I'm
curious
whether
there
are
situations
where,
like
unidirectional
links,
make
it
a
little
tricky
for
some
reason.
B
And
they
can
come
with
some
overview.
Basically,
the
analysis
on
costs
and
anything,
that's
not
great
and
could
be
better
any
pitfalls,
and
then
I
think
they
will
be
open
to
any
questions
and
answers
session.
C
Great,
please
include
me
to
this
conversation.
I
also
was
planning
to
talk
about
rich
rice
with
with
him
so
like
we
are
really
close
to
the
face
when
we
can
get
all
this
stuff
merged.
So
basically
it
would
be
good
to
also
plan
some
implementation
for
application
sites
for
advice,
that's
cool,
it
can
be
a
joint
conversation.
A
This
is
also,
I
would
say,
in
general,
like
a
place
where
we
don't
have
a
lot
of
precedent
in
open
telemetry,
because
all
of
the
other
data
types
for
lack
of
a
better
word
that
we've
all
the
data
structures
we've
implemented
so
far
have
been
like
really
robustly
tested
in
reality.
Right,
like
our
model
for
tracing,
is
not
new.
Our
model
for
logs
is
not
new.
Our
model
for
metrics
is
like
has
some
like
cool
nifty
stuff
in
it.
A
But
you
know
in
general,
is
is
not
new,
but
you
know
things
like
links,
and
you
know
these
big
asynchronous
messaging
systems
and
all
of
this
other
stuff
that
that
just
feels
new.
So
you
know
I
have
some
that's
why
it
would
be
great
to.
A
A
Yeah
anyways,
that's
that's
all
I
have
other
people
have
agenda
items
they
want
to
discuss.
D
A
Yeah
start
start
with
the
ticket
and
then
raise
it
in
like
the
specification
slack
channel.
So
it's
been
six
days
and
I
know
he's
around
so
yeah.
I
would
suggest
going
ahead
and
following
up
with
that
in
the
in
the
slack
channel
and.
D
A
D
Yeah,
I
the
feeling
I
get
from
this
pr
like
the
client
errors,
is
that
there
is
a
bit
too
much
resistance
to
it.
It
seems
like
there's
a
few
instrumentation
like
nikita
christian.
D
D
A
A
Well,
yeah,
so
so
I've
I've
been
like
expanding
this
section,
there's
a
section
in
here
about
how
to
get
your
pr
merged.
You
know
so.
We've
we've
got
this
thing
in
here
that
we
could,
I
think,
want
to
make
more.
I
actually
need
to
add
a
line
to
make
this
more
explicit,
which
is
that
the
only
things
which
count
as
blockers
are
things
that
are
done
explicitly
with
the
request
changes
feature.
A
So
you
know
if
something
has
got
approvals,
enough
approvals
to
be
merged.
Obviously,
like
finish
up
discussions
if
they're
happening,
but
but
the
only
thing
that
that's
like
a
hard
block
is
that
a
code
owner
has
put
in
like
request
changes,
but
it
does
talk
about
like
pr
being
stuck.
There's
lots
of
debates.
People
can't
agree
with
each
other.
A
It
puts
all
the
onus
on
the
owner
right
now,
like
the
owners.
Try
to
get
people
lined
try
to
consolidate
perspectives
by
putting
in
a
summary
so
like
that
would
be
like
one
next
step
to
be
like.
So
it
seems
like
this
pr
is
kind
of
like
50
50
right
now,
and
I
I
tried
to
put
a
summary
in
there
before
so
I
could
do
it
again.
A
You
know
trying
to
to
tag
more
experts
in
the
field,
but
I
feel
like
they're
already
there
you
know
reaching
out
on
slack
looking
to
see
if
there's
a
way
to
narrow
it
down.
So
isn't
in
this
case,
but
this
is
like,
I
think,
a
thing
I
want
to
expand
upon,
which
is
like
oh
yeah
and
then
like
bring
it
to
the
spec
meeting.
It's
just
like
that's
nice.
A
This
is
spec
meeting,
but
the
thing
I
want
to
add
here
is
like
there's
no,
this.
This
is
like
100
on
the
contributor
to
like
sort
all
this
out,
and
I
I
want
that
to
get
changed
to
be
like
you
know
the
basically
like
the
person
assigned
to
the
issue
and
a
sponsor,
if
there
is
one
like
need
to
to
help,
get
it
unstuck.
D
Yeah,
I
feel
like
once
we
get
sponsor
in
this
group
that
will
definitely.
A
D
A
I
hear
your
feedback
that
it's
like
when
it's
mixed,
it's
it's
very
hard
for
the
the
contributor
who's
like
pushing
the
changes
to
be
able
like
yes,
that
person
should
be
trying
to
like
sort
out
all
the
feedback,
but
when
it
hits
like
a
no-go
point
like,
I
think
the
ball
has
to
get
like
bounced
back
over
to
the
tc
or
somewhere
else.
Just
because
it's
you
know
yeah,
I
I
don't.
A
I
don't
see,
I
don't
totally
see
how
other
than
yeah
the
person
just
choosing
to
give
up,
but
it
just
it
feels
like
like
there
should
be
some
some
help
there,
even
if
the
help
is
just
to
look
at
it
and
say
like
yeah,
give
up
like
it's.
This
is
too
contentious
to
merge
right
now.
It's
just
helpful
to
have
somebody
on
the
tc
or.
D
If
someone,
if
someone
from
the
tca
was
like
give
up
on
this
pr,
I'd,
be
I'd
be
over
the
moon.
I'd
be
so
happy
because
I'm
just
like
this
is
a
decision.
This
is
a
decision
that
we
said
we
needed
to
make
for
the
ocean
and
by
someone
from
the
tc
saying
give
up.
That's
the
decision
made
and
we're
one
step
closer
to
being
stable,
which
I
think
exactly
exactly
yeah
cool,
so
yeah,
okay,
I
will
we'll
I'll
monitor
it.
There's
no
rush
I'll
ping,
car
or
some
stuff,
like
that.
Yeah
yeah.
A
Sean,
I
know
you
don't
have
it
on
the
agenda,
but
you've
been
working
on
on
schema.
Versioning!
Apologies!
If
you
can
hear
my
cat
he's
very
hungry.
E
Yeah,
that's
cool
I'd
like
so.
I
just
picked
it
up
like
pretty
much
yesterday,
but
t
grant
has
also
started
his
own
implementation
detail
of
how
we
do
schema
migrations
inside
the
collector,
but
he's
out
for
two
weeks
so
I'll
be
taking
that
over
yeah.
It's
not
really
much
to
discuss
the
other
than
that.
It's
happening
and
it's
going
to
be
really
cool
but
it'll
be
published
inside
the
contrib
collector
first
as
an
experimental
feature,
but
yeah
once
it's
out.
E
I'll,
obviously
share
it
back
here
and
like
continue
on
and
also
try
and
make
sure
that
the
pr's
are
here
so
that
people
are
also
aware
of
what's
happening
and
what
you
can
consume.
A
A
E
Yeah
so
like
the
collector
just
allows
you
to
like
assign
processes
to
certain
pipelines
or,
like
certain
processing
groups.
E
The
big
thing
that
I
haven't
ideally
worked
out
is
like
the
specification
says
that
you
can
upgrade
and
downgrade
the
semantic
convention.
So
if
you're
sending
with
instrumentation
version
1.9.0
but
you're
technically,
targeting
like
1.7.0
like
how
do
you
actually
define
that
and
then
there's
other
technical
difficulties,
whereas,
like
you
can
define
the
schemas
online,
but
also
cache
them
and
then
also
cache
the
latest
version,
with
the
collector
on
build
as
part
of
the
specification.
E
A
E
If
it
doesn't
have
a
locally
sourced
one,
it
would
find
fetch
online
and
like
that
would
be
the
case
in
atlassian
because,
like
you
say
that
we
want
a
schema
version,
we
probably
like.
We
don't
run
our
containers
with
persistent
storage
right,
so
more
than
likely
we'll
be
like
yeah
we're
just
going
to
go,
download
it
and
then
all
of
a
sudden,
it's
like
open,
telemetry.I
o
is
down
that's
out
for
maintenance.
E
A
hundred
percent
but
like
how
we
specifically
run
outside
carson
in
production,
is
we
bundle
all
the
things
together,
but
we
might
not
necessarily
have
the
right
schemer
in
there
and
if,
if
that's
the
case,
go
fetch
it
for
each
time
a
new
instance
starts,
but
we
don't
back
any
of
that
storage
and
we
just
treat
all
storage
as
ephemeral
in
that
host.
A
E
B
A
Yeah
yeah,
I
would
I
mean
I
I'm
glad
you're
you're
trying
to
solve
that
that
remote
fetch
issue.
But
I
would
definitely
recommend
bundling
that
bundling
that
in
your
build
process,
because
that
sounds
like
yeah
like
a
great
a
great
way
for
your
collectors
to
fail
to
redeploy
because
they
don't
have
access
to
some
cdn.
A
Essentially
like
like,
like
any
single
juncture
where
it's
like.
I
need
to
talk
to
another
computer
and
do
something
it's
like.
No,
you
do
not.
You
do
not
need
to
do
that
unless
it's
like,
like
chop
chop
all
those
connections
out
of
there
as
much
as
you
possibly
can,
because
you
know,
because
they
it's
not
only
that
they
can
fail,
it's
that
they
create
these.
Like
now,
you
have
this
third
state
to
deal
with,
which
is
like
some
of
them
are
in
one
version,
and
some
of
them
are
on
another
version.
D
A
Blocked
you
know,
did
you
already
tear
down
the
old
collectors
that
would
be
a
bummer
yeah,
anyways
anyways.
E
A
little
bit
tangential
digression,
but
yeah
it's
currently
what
I'm
looking
at
the
moment
and
the
basically
the
biggest
problem
is
how
you
do
the
transactions
of
like
going
up
and
down
from
targeted
versions.
E
Which
will
be
the
interesting
part?
But
I
don't
know
if
this
group
technically
owns
those
schema
migration
files,
because
I've
not
seen
them
brought
up
here
before.
A
Well,
I
mean
we'll
be
the
ones
doing
the
work
on
of
on
them,
so,
yes,
I
do
think
we
get.
It
would
be
interesting
to
see
if
that's
in
the
code
owner's
file
correctly.
I
don't
actually
know
where
those
things
live,
but
yeah.
A
We
might
want
to
update
the
code
owner's
file
because
I
do
think
this
group
is
like
the
set
of
represents
the
set
of
code
owners
that
that
should
get
assigned
to
those
tickets
and
and
review
them
and
it'll,
be
us
holding
the
tickets
up
right,
but
yeah
right,
like
if
people
are
proposing
schema
chain
like
how
do
I
put
it
the
process
once
something
is
stable
has
to
be,
and
I
don't
know
where
we
write
this
down.
This
is
just
I
think,
part
of
the
people.
A
Processing
spec
changes
around
the
semantic
conventions.
Once
you
know
we
don't
have
any
of
it
stable
yet
so
it
doesn't
matter
right
now,
but
when
some
of
it's
stable,
the
rule
has
to
be.
You
can't
change
that
you
you
cannot
submit.
You
cannot
accept
a
change
to
a
stable
part
of
the
semantic
conventions
without
inc,
bundling
a
schema
translation
right
like
like
that
it
has
to
it
has
to
like
be
there
and-
and
it
also
you
know
that
that's
the
way.
A
We
also
prove
that
that
it
that
the
change
is
viable
essentially
right,
yeah,
we
don't,
we
don't
want
there
to
be
like
a
gap,
which
is
why
I'm
really
happy
you're
you're
doing
this.
This
work
right
now,
because
I
was
nervous
that
we
were
going
to
start
marking
things
stable
without
having
a
process
to
actually
like
deal
with
these
schema
things.
E
E
A
Yeah,
there's
also
on
the
collector's
side,
alex
bowden
is
a
maintainer
and
he's
definitely
around
so
you,
you
can
ping
him
about
collector's
stuff,
very,
very
happy
that
we
added
additional
collector
maintainers
before
both
bogdan
and
tigran
went
on
vacation.
At
the
same
time,.
A
B
D
D
Schema
stuff,
I
remember
in
one
of
the
java
meetings,
john
watson
had
some
problem
with
the
schema
versions.
It
was
making
some
difficulties.
Do
you
remember
what
that
was
anarag
like
and
is
there
anything
we
can
do
to
make
schemers
like
easier
for,
like
instrumenters
to
work
with
going
forward.
A
One
thing
and
that
this
is
like
total
bonus
points,
and
it
kind
of
relates
to
this
like
adjacent
issue
of
defining
dashboards
like
open,
telemetry
dashboards
in
some
like
a
coherent,
open
source
way.
There's
this
question
of
of
like
keeping
track
of
what
schema
translations
you
even
need.
A
How
do
you
know
when
you
have
caused
a
schema
bump
that
is
relevant
to
a
dashboard
that
you
have,
and
the
answer
is
like
right
now
you
just
something
breaks
and
you
notice,
and
then
you
you
like
roll
out
a
schema
translation
directive,
but
it
would
be
kind
of
cool
if,
like
that
was
a
thing
we
could
resolve
in
some
way
and
there's
like
been
a
kind
of
related
project
that
people
have
been
like
like
tossing
around,
but
no
one
quite
knows
how
to
start,
which
is
around,
like
you
know,
we're
defining
all
these
semantic
conventions
right
and
we
have
all
this
instrumentation.
A
That's
like
expected
to
produce
stuff.
It
would
be
great
to
have
a
repo
somewhere
that
defined
like
the
default
dashboards
that
you
would
want,
because
that
would
kind
of
like,
like
close
the
loop
like
one
of
the
places
people
get
get
kind
of
like
stuck
is
like
okay,
the
telemetry's
all
open,
but,
like
all
of
my
dashboards,
are
like
locked
up
somewhere
or
you
know,
I'm
trying
I
like
roll
out
a
new
service
or
like
I'm,
trying
to
get
started
quickly
and
it
would
be
kind
of
cool.
D
D
A
A
A
A
D
Exactly
yeah,
maybe
we
need
to
define
our
own
open,
one
that
can
translate
into
from
cool
for
metrics
and
and
other
things.
I
don't
know
if
that's
possible,
but
yeah
be
cool.
It
would
be
really
cool
because
then,
then,
as
those
things
for
those
standard
dashboards
you're
talking
about
all
the
like,
it
really
does
close
the
loop,
because
you
just
provide
a
query
and
then
people
just
dump
the
query
wherever
and
yeah.
Oh
it's
all
there
yeah.
E
E
Like
we
need
like
experiences
group
like
the
group,
that's
looking
at
from
the
back
to
like
the
front
being
like
we
really
want,
like
this
dashboard
to
look
like,
like
the
experience,
is
we're
trying
to
figure
out
what's
wrong
with
this
instance.
What
is
the
crap
that
we
need
on
there
and
then
going
backwards
and
like
for
this
instrumentation,
like
with
this
project,
we
need
these
experiences
like
jvm,
like
gc,
count
and
stuff
like
that,
and
like
take
that
experience
as
like
someone
call
what
else
would
they
need
for
that
or
like
yeah?
E
A
E
Might
be
a
good
starting
place
because
then
yeah.
A
It
totally
totally
that's
like,
like
the
the
closing
of
the
loop
here
is,
is
like
that
bit
you
just
described,
and
it
is
totally
not
part
of
mvp,
which
is
like
why
I'm
not
actively
pursuing
it
right
now,
but
it
is
interesting.
I
have
talked
to
like
numerous
people,
like
even
people
not
associated
with
open
telemetry
ben
siegelman
was
talking
to
somebody
else's
atlassian
like
at
the
like
higher
level.
A
Yeah,
so
it's
just
like
this
is
like
something
that's
like
totally
in
the
in
the
air
that
I
don't
know.
Personally,
I
would
be
like
precise
it's
like
one
of
the
like
several
projects
to
me
that
are
like
the
exciting
work
once
we
get
past
this,
like
basic
skutt
work
of
getting
everything
like
stable,
you
know
and
is,
is
like
sorting
out
that
kind
of
stuff,
because
I
that
that
would
be
that
that's
just
like
when
we're
getting
into
the
realm
of
like
stuff
that
doesn't
like
exist.
A
Yet
you
know
like
no
one's
gone
out
there
and
like
done
a
project
like
that
yet,
and
it
would
be
like
it's
like
the
kind
of
thing
like
I
wish
I
had
you
know
when
I
was
like
rolling
something
out
for
the
first
time
if
there
was
like
a
default
playbook
to
like
start
with,
you
know
like
that,
would
that
would
be
like
really
cool.
A
So
I
don't
know
anyways
food
for
thought.
I've
thought
about
poking
light
step
to
maybe
like
open
source.
Our
query
language,
just
because
I
know
that
we,
like
people,
put
a
lot
of
thought
into
developing
a
query
language
that
does
work
across
all
these
signals.
So
it's
like
it
could
be
like
at
least
like
a
starting
point
where
like
well.
I
know
that
one.
I
know
that
one
works
at
least
so
or
maybe
you
know,
prom
ql
like
is-
is
sufficient
or
something,
but.
A
A
But
it
would
be,
nice
is
what
we're
his
it
would
be
nice
if,
if
you
had
a
thing
like
that,
most
of
these
tools
have
like
an
api
for
defining
dashboards
right,
and
so,
if
you
had
some
some
neutral
description,
then
it
would
be
like
very
easy
to
like
like
write.
A
E
I
know
that's
bit
us
before
internalizing,
like
we've
tried
to
might
write
an
order,
migration
tool
from
one
query
language
to
another
and
the
the
defaults
that
are
unavoidable
is
what
tripped
us
up
like.
I
think
memory
like
going
from
one
vendor
to
another
like
the
default
they
used
was
like
the
roll-ups
used
as
some
roll-up,
whereas
the
new
vendor
just
didn't
use
anything,
and
it's
like.
Oh,
you
got
to
figure
it
out
and
you're
like
oh
okay,.
A
A
E
Yeah
but
yeah,
I
agree
like
the
concept
like
we
don't
strictly
need
to
take
the
like
going
from
one
to
another.
It
could
just
be
like
otlp
exists
in
the
collector
from
going
from
there,
whereas
hotel
otlp
defines
like
the
end
like
the
processing
type,
and
then
the
vendors
just
could
come
together
and
be
like.
We
extend
it
like
this,
and
this
is
what
the
defaults
is.
A
E
A
Yeah-
and
it
would
be
a
great
place
to
get
experts
in
to
also
like
write
up
things
like
what
what
would
be
like
good
default,
alerting
values,
be,
for
example,
if
you're
running
cloud
sql
or
something
you
know
like
what?
What
do
you
want
to
alert
on?
What
kind
of
like
thresholds
would
you
want
to
set?
You
know
it's
it's
of
course
like
hard
to
really
like
give
real
great
default
answers
to
a
lot
of
things,
because
there's
like
a
default
deployment
of
these
things
necessarily,
but
I
don't
know
it
would
be.
A
Yes
right,
there's
like
the
obvious
stuff
which
is
like
if
this
is
happening.
Definitely
bad
and
like
most
of
these
systems
also
have
something
tunable
like
there
are
various
knobs
right,
and
so
there
are
various
metrics
or
observables
that
you
can
emit
that
are
related
to
those
knobs.
A
So
you
know
when
you
know
that
when
something
has
gone
out
of
a
certain
threshold
that
like
these
are
the
knobs,
you
should
look
at
to
tune
it.
I
got
that
one
from
couchspace
actually
started
to
do
some
of
this
stuff.
They
embedded
open
tracing
into
all
of
their
clients
and
then
like
shipped
a
playbook
with
couchspace.
That's
like
so
here's
the
here's,
the
observability
we're
giving
you
and
like
here's
like
what
we
think
you
should
do
with
it.
Yeah
yeah,
I
don't
know.
I
thought
I
thought
it
was
solid.
A
Well,
well,
anyways
food
food
for
thought,
fun
projects,
motivation
to
to
get
through
get
through
the
rest
of
this.
Oh
and
I
completely
forgot
this
is
relevant
to
this
group.
I
don't
know
if
you
heard
but
elastic
elastic,
joined
the
the
logging
sig
and
they
want
to
contribute
the
elastic
common
schema
to
open
telemetry.
A
Compare
it
to
what's
in
open
telemetry
anything,
that's
double
defined,
the
open,
telemetry
side
wins
essentially,
and
they
will
do
the
work
on
their
end
to
like,
basically
do
the
migration
between
their
current
schema
and
whatever
the
new
combined
schema
will
be,
and
once
they've
done.
That
basically
say
you
know.
Ecs
plus
plus,
is
the
open,
telemetry
schema.
A
So
that's
that's
another
place
where,
hopefully
we
will.
I
think
it
matters
more
for
logging,
potentially
than
some
of
the
stuff
we're
defining.
But
if
you
haven't
looked
at
elastic
comments,
yeah.
D
A
Yeah,
so
so,
if
you
haven't
looked
at
elastic
common
schema,
it's
worth
it's
worth,
checking
out!
That's!
I
should
have
mentioned
that
earlier,
but
this
because
they
want
to
merge-
and
I
think
it's
amazing-
that's
like
one
big
step
towards
having
a
real
standard,
because
they're,
I
would
say,
like
elastic
common
schema,
is
like
the
next
biggest
standard
like
thing
in
this
space,
but
they've
already
defined
a
whole
bunch
of
stuff.
So
maybe
there's
there's
stuff
there
that
we
can
use
as
guide
posts
for
some
of
the
work.