►
From YouTube: 2020-07-15 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
A
A
A
A
A
A
A
A
A
A
A
B
B
C
C
D
G
G
G
So
Tigran
actually
review
once
and
his
last
comments
were
basically
because
the
this
PR
is
currently
incomplete.
So
it's
like
it
only
contains
the
configuration
portion.
So
he
asked
me
to
add,
make
it
explicit
in
the
readme
that
this
is
incomplete.
Okay,.
H
G
G
F
I
F
H
G
F
H
F
G
Yeah,
if
possible,
because
I
guess
I
hate
keep
like
just
ping:
hey.
F
J
F
E
So
it
kind
of
seems
that
things
are
getting
backed
up
right
reviews,
yeah,
there's
been
so
maybe
I,
don't
know.
If
there's
anything
for
us
to
discuss
here,
help
maybe
I'll
discuss
it
internally,
but
it
openly
seems
like
we
probably
need
better
some
more
process
like
even
leads
to
be
blowed.
You
know,
make
sure
things
get
reviewed.
I
E
E
B
B
F
F
Okay,
so
approvers
are
a
waste
lane
at
Splunk
Rahul
at
Google
rebels,
not
that
active
and
open
telemetry
anymore
Steve
is
black
James
at
Google
Dimitri
from
spunk
and
the
maintainer
czar
Bogdan,
Paulo
and
Tigran
I.
Wonder
if
we
wanted
to
increase
the
number
of
approvers,
then
for
other
people
who
have
been
pretty
active
in
this
repository.
F
Like
I
know,
James
you
joined
the
the
approvers
list
pretty
recently
James
yep
enough
Niall
has
he's
been
using
the
collector
for
a
lot
of
internal
purposes.
Do
you
know
if
he's
been
making
a
lot
of
contributions
to
the
open-source
repos?
He
was
adapting
it
for
you,
sorry
Niles,
a
Googler
he's
working
on
GK
kubernetes.
E
I
wonder
if
we
should
give
people
more
like
I
wonder
if
we
need
to
like
spend
much
time
reviewing
things
that
are
like
vendor
specific,
so
like
this,
the
x-ray
again
we're
like
you
know
generally,
when
I
review
those
like
a
doll,
I
look
for
more
like.
Is
there
any
like
systemic
issues
with
there
like
something
it
will
like
impact?
Something
else
right,
I,
don't
feel
like
making
sure
it's
gonna
work
right
for
for
their
specific
use
case
right,
they
kind
of
do
better.
E
E
G
Design,
it's
a
big
job,
so
I
guess
the
the
main
thing
I'm
looking
for
in
a
review
from
the
OT
community
or
you
guys
are
actually
more
like
what,
for
example,
whether
I'm
using
the
internal
make
tree
reporting
libraries
correctly
and
less
about
vendor
specific.
Because
I
mean
it's
totally
understandable
that
you
guys
stole
yeah
out
of
the
internet.
F
E
Think,
and
even
like
improved
some
I
mean
the
stuff
that
we
do
bike
check
for
like
week.
Pardon
me
turn
those
tin
like
checklist,
where
it's
just
you
know,
I
mean
the
kind
of
stuff
should
be
like
document
it
somewhere
right
of
like
use
this
of
what
best
practices
for
yeah.
You
know
exporters
processors,
like
you
know,
so
that,
even
though
checks
we
do
do,
we
can
turn
those
into
like
okay,
wait.
We
give
in
streamline
and
make
them
before
explicit
play.
J
F
F
I
A
I
Yeah
I've
been
working
on
issue
1150
and
it's
it's
a
remote
right
exporter
for
the
Prometheus
revolt,
Ramona
exporter
for
the
collector
project.
Coop
now
I've
been
communicating
with
Josh
about
all
my
quick
questions
and
a
lot
of
things
that
I
have
but
I
think
josh
is
Thursday
with
like
doing
ot,
ot,
LP
and
other
stuff.
Yes,
if
there's
someone
that
I
could
like
yeah
ask
my
questions:
do
it
what's
your
thing
just
working
on
the
project,
yeah.
B
A
F
So
yeah
that
the
experts
on
the
general
metrics
spec
inside
of
open
telemetry,
that
this
is
the
spec
that
applies
to
the
SDKs
and
applies
to
the
collector
I,
think
it
probably
Josh
would
be
the
he's
driving
most
of
that
work
and
then
Tyler
and
Bogdan
and
John
are
all
pretty
involved
in
it.
So
any
one
of
them
should
be
able
to
explain
just
general.
F
Sort
of
the
metrics
spec
how
metrics
are
meant
to
work
everything
like
that,
but
let's,
let's
understanding
that
your
issue
now
so
just
quickly,
go
over
it
and
then
see
what
we
should
do
next:
okay,
Prometheus
exporter
yeah.
This
is
issue
number
eleven,
fifty
great,
not
pull
request
one.
Fifty!
Yes,.
J
A
F
I
I
just
oh,
we
just
have
a
couple
questions
about
like
converting
to
specifically
on
like
because
Prometheus
uses
cumulative
metric
types
and
then
and
then
we're
just
wondering
about
how,
like
the
conversion
happen,
perfect.
F
So
norms
for
most
issues
relate
with
a
collector
you're
at
the
right
meeting,
because
this
one
is
actually
related
to
the
metrics
spec
itself.
I
recommend
at
three
o'clock:
Pacific
4
o'clock
Pacific
tomorrow
you
attend
the
open,
Salama
Tree
metrics
sig
meeting,
because
the
treatment
and
aggregation
of
Prometheus
metrics
has
been
a
topic.
That's
come
up.
There
I
suspect
they
already
have
an
answer
to
your
question,
but
all
of
the
right
people
will
be
there
and
they'll
be
happy
to
answer
I'll,
be
there
as
well.
Okay,
great
yeah,
it's
on
it's
on
shared
calendar.
F
K
F
F
F
K
E
F
Yeah
I
don't
stand
agree
like
T
grand
will
have
the
official
answer
because
he's
the
main
person
on
this
but
yeah
we've
got
by
with
you.
I
weekly
releases
as
soon
as
your
thing
is
ready
and
and
approved,
and
merged
it'll
just
go
out
the
next
bi-weekly
release,
so
there's
there's
no
sort
of
like
there's,
not
a
lot
of
benefits,
so
to
speak,
to
being
tied
to
a
milestone.
The
milestones
are
more
for
tracking
things
that
are
required
to
implement
the
next
spec
version.
L
A
question
general
here
with
Google:
this
is
perhaps
just
my
ignorance
but
I
wanted
to
know.
Does
open
telemetry
have
a
data
model
that
accommodates
events
and
so
a
lot
of
times.
What
we'll
do
at
Google
is
we'll
we'll
track
what
are
basically
events
and
then
just
have
them,
have
a
value,
B
track
them
as
a
metric
and
have
a
value
of
like
one.
You
know
yeah,
but
the
truth
is
they're,
not
time
series.
There
are
not
things
that
are
like
oh
server
started
or
server
crash.
L
F
Are
events
in
the
context
of
traces
or
in
some
spans?
Rather,
those
you
know,
have
some
advantages
and
disadvantages,
one
disadvantage
being
if
you're
sampling
like
if
you're
sampling,
your
traces,
you
can't
there's
no
core
events.
Data
model
this
has
come
up
in
the
logging
working
group
is
something
that
we
would
eventually
want
to
do,
and
weather
events
should
be
represented
as
logs
or
weather
events
should
be
their
own
thing.
Yeah.
I
F
L
Yeah,
because
I've
thought
about
this
and
like
that's
some
of
our
data,
so
I
often
the
way
that
it's
modeled
in
other
systems
as
an
event
happens
and
then
an
event
can
have
a
number
of
consequences,
including
fact
that
a
metric
might
be
stored
or
that
a
log
entry
might
be
generated
or
something
else,
and
so
in
the
dip.
At
the
back
end
that
you
send
data
for
for
an
event
doesn't
have
to
be
the
same
as
a
as
time
series.
L
C
This
is
a
Ament
from
AWS.
This
has
come
up
before
where
we
needed
to
deal
with,
alarms
around
kicking
off
alarms
when
certain
metrics
it
specific
threshold
mmm-hmm
and
for
now,
as
a
workaround.
Until
an
inventing
models
built
in
open
telemetry,
we
are
planning
to
use
the
concept
of
labels
that
metrics
today
support
and
we're
just
kind
of
pass.
The
alarms
there
sure
labels
are
just
the
metadata
with
the
metric
object,
yeah.
L
B
L
One
thing
that
I
think
is
an
easy
trap,
but
I
think
it's
a
trap
is
to
say
that
an
event
is
a
log
record.
I
think
like
an
event
is
a
logical
event
and
then
it
can
have
a
number
of
associated
things
like
a
log
record
or
metric
or
Consequences.
If
you
want
to
think
of
it
that
way,
but
I
said
I've
heard
some
people,
especially
those
that
are
in
the
logging
space,
that
just
sort
of
had
logs
on
the
brain
that
were
like.
L
L
E
D
So
what
I
was
planning
on
doing
for
the
initial
implementation
was,
so
we
actually
have
it
with
signal
effects.
We
have
a
specific
event.
We
have
a
specific
event
model
that
you
can
stand
in,
that's
basically
exactly
what
you're
describing,
but
for
all
of
our
monitoring
of
our
actual,
like
servers
and
stuff,
we
use
time
series
for
all
of
that.
We
so
like.
If
a
server
before
I
know
for
servers
up
or
down,
we
have
time
series
that's
either
0
1
or
some
other
thing
that
we
can
track.
D
We
don't
use
events
to
actually
monitor
directly.
We
use
it
more
for
informational
context,
but
anyway
for
actually
sending
events.
The
signal
effects
style
events.
My
initial
plan
was
to
actually
like
to
get
them
from
a
receiver
to
an
exporter
like
kind
of
when
you're
doing
like
a
proxy
type
scenario.
My
initial
plan
is
to
use
is
to
essentially
wrap
them
in
a
specially
crafted
log.
Just
because
of
all
the
three
data
models
right
now
it
exists
in
the
collector.
That's
the
only
one.
That's
really
fits
at
the
closest
I
say.
L
D
E
We
think
there's
gonna
be
sounds
like
every
yeah
there's
a
couple
different,
workarounds
or
packs
that
people
are
going
to
do
to
kind
of
represent
these
dipping
on
like
kind
of
their
specific
needs
sweet.
This
is
that
going
to
cause
any
problems
down
the
road
I
mean
we
kind
of
comment
on
like
defining
a
formal
event
model
or
like,
or
do
we
think
or
is
this
gonna
come
eat?
E
L
I
mean
I,
I,
think
everybody.
If
you
look
carefully,
if
you
have
a
rich
set
of
data,
everybody
has
it
that's
like
it's
just
they're,
not
so
they're,
probably
flowing
back
through
time
series
or
logs
or
some
combination.
So
until
there
is
a
formal
model,
I'm
guessing
people
will
just
continue
to
do
that,
but
I
do
believe.
There
is
a
case
for
it
being
its
own
independent
model.
I.
Think
there's
a
good
case
for
that
myself.
L
Well,
like
I,
said,
if
somebody
is
interested
in
taking
it
up
as
carrying
the
mantle
as
far
as
it
being
a
new
sort
of
class
of
data,
I'd
be
happy
to
partner
on
it,
but
I
probably
don't
have
the
bandwidth
to
drive
it
right.
Now,
though,
eventually
eventually
I
will
come
back
to
it,
but
it's
probably
well
yeah.
B
L
B
L
B
It's
kind
of
the
whole
project,
I
mean
it's
like.
Okay,
we
got
to
you
know:
we've
got
people
go,
we
got
to
get
this
in
there,
we'll
get
this
in
there
and
yeah.
On
the
other
side,
we
got
all
this
work
in
progress
in
yeah
lots:
lots
of
tickets
out
there
that
have
been
sitting
there.
A
long
time
got
to
get.
You
know
we
gotta
get
some
stuff
done.