►
From YouTube: 2021-08-25 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
Okay,
let's
get
started.
I
think
we
have
a
question
vineeth
did
you
want
to
start
first.
B
I'm
not
sure
if
anything's
here
this
was
a
question
that
got
added
at
the
tail
end
of
our
discussion.
Last
week
we
didn't
have
time
so
it
looks
like
it
was
moved
to
the
top
of
this.
A
B
But
I've
commented
on
the
issue:
that's
linked
there
as
well,
describing
some
of
the
work
that
we've
done
as
part
of
the
target
allocation
project.
To
enable
this
to
happen.
So
the
we
had
the
configuration
function
that
is
used
by
the
prometheus
operator
to
generate
prometheus
config
from
those
crds
exposed
so
that
it
could
be
reused
by
the
open
telemetry
operator
and
the
target
allocation
project
added
a
prometheus
config
rewrite
step.
B
So
we
have
a
point
at
which
it
would
be
logical
to
rewrite
the
the
prometheus
configuration
with
new
configuration
from
observed,
pretty
prometheus
crds
such
as
service
monitors
and
pod
monitors.
We
just
need
to
wire
the
the
things
together.
That's
a
as
a
final
step.
B
And
I
think
trusty
has
commented
on
there
as
well
that
he's
in
favor
of
that
just
some
questions
about
exactly
the
mechanics
of
making
it
happen.
A
Okay,
I
mean
that
that
makes
sense,
it's
good
to
be
easier
to
find
the
information
on
the
issues.
So
that's
thank
you
for
doing
that.
B
B
We
are
at
the
point
now,
where
we're
starting
to
remove
components
from
the
core
repository
and
hope
to
finish
the
cleanup
of
that
today
or
tomorrow,
to
make
a
new
release
of
the
the
quorum
contrib
by
tuesday
of
next
week
at
the
earliest
or
sorry
at
the
latest.
Anyone
who
had
a
pr
against
any
of
the
prometheus
components
in
the
core
repository
will
need
to
move
that
pr
to
the
contrib
repository
and
target
the
components
there.
Instead,
it
should
be
fairly
straightforward
because
we've
just
picked
them
up
and
moved
them.
B
A
Anthony,
do
you
need
any
folks
to
test
in
terms
of
once
we
have
a
release,
I
mean,
is
it
I
mean
we'll
have
to
refile
the
prs
that
are
expanding
right
now
on
core
right?
B
Yeah
so
so
prs
that
are
against
core
will
need
to
be
refiled.
I
just
put
a
link
to
the
open,
telemetry
collector's
releases
repo
that
gerasi
has
where
he
has
a
release.
That's
built
from
the
current
state
of
corn
contrib
with
the
core
components
pulled
out
of
contrib.
B
So
if
people
wanted
to
give
that
a
test,
there's
a
release
that
could
be
tested
to
validate
that
all
of
the
things
are
working,
we've
run
it
through
the
unit
test
and
integration
tests
and,
and
things
appear
to
work,
but
as
always
more
eyes.
The
merrier.
A
Yeah,
that's
definitely
good
to
know
vishwa
and
grace
would
like
to
get
your
help
on
brothers,
so
just
give
it
a
spin
and
give
us
a
feedback.
If
you
see
anything
broken.
D
A
Okay
also
wanted
to
kind
of
grace.
I
know
your
pr
probably
will
end
up.
Okay,
you've
already
filed
it
on
good
trip,
so
I
think,
as
soon
as
we
can
get
all
these
tests
passing
and
everything
you
know
built,
then
definitely
things
will
start
moving
we're
expecting
early
next
week
for.
A
Alex
did
you
have
an
update
on
your
any
of
your
prs
or
work.
C
Yeah,
nothing
major
jurassie
was
reviewing
it
and
he
just
told
me
to
rebase
it,
and
I
did
that
this
morning.
A
Okay,
okay,
so
just
this
is
three
five.
Four.
E
One
very
interesting
thing
is,
it
doesn't
seem
like
contributes
whether
the
github
actions
or
whatever
they
are,
are
building.
I
see
yeah
yeah
in
there.
There
are
all
these
failures
like
go
mod
download
but
yeah.
We
already
have
some
content
added
to
the
go
mode
file,
so
yeah
there's
that
going
on
and
then
you
know
we're
in
the
prometheus
work
groups,
I'm
assuming
we're
talking
about
prometheus
yeah.
E
Yeah
yeah,
so
in
in
regards
to
issues
going
on,
we
still
have.
We
still
haven't
yet
finished
closed
out
that
issue
of
remote,
sorry
of
stillness
markers.
E
We
haven't
yet
finished
that
out
and
then
oh
yeah,
like
yeah
of
stainless
markers,
and
then
we
have
one
in-flight
thing
of
bringing
out
the
test.
Bed
like
decking
it
out
to
add
remote,
remote
right
compliance
tests
so
that
they
run
in
every
single
commit
push.
A
Okay,
but
that's
a
good
call
out
on
the
github
actions
being
broken,
we'll
take
a
look
at
it
because
that's
that
needs.
F
E
A
I'm
just
going
to
bring
up
the
work
group
issues
and
we
have
pretty
much
completed
all
the
items
on
the
fee
on
phase
one,
and
there
were
a
couple
of
tracker
items
that
I
moved.
Let
me
just
share
my
screen.
A
Yep,
okay,
cool
yeah,
all
right,
so
I
think
that
we
have
about
16
issues
that
we've
been
tracking
for
phase
two
which
are
open
now.
Some
of
them
are
tracker
issues,
so
we
can
ignore
them,
primarily
they're
the
requirements
for
the
prometheus
receiver
tests.
I
think
that
that
was
something
visual.
You
were
going
to
help
us
in
defining
some
requirements
in
terms
of
what,
overall,
we
need
to
do
in
order
to
you
know
kind
of
add
those
tests
and.
F
D
G
D
A
I
see
that's
fine,
I
mean
again.
I
think
that
if
we
can
even
capture
some
of
the
requirements
that
you're
seeing
which
are
common
with
the
compliance
tests,
then
we
can
actually,
you
know,
figure
out
what
makes
sense.
Maybe
we
can
also
get
some
insight
from
julian
or
brian
on.
You
know
some
of
the
standard
discovery
mechanisms
and
the
I
mean
what
the
functionality
of
the
receiver
that
may
already
exist
in
prometheus.
Do
you
guys
have
any
insight
into
that?
H
Yes,
I
don't
think
we
have
end-to-end
test
for
so
we
have
the
unit
test,
but
we
don't
have
like
acceptance
for
the
service
queries.
I
see.
A
B
What
he
was
saying
was
that
it's
kind
of
hard
to
test
end
to
end
the
service
discovery,
but
if
one
of
them
is
working,
they
should
all
work,
which
is
a
proposition
that
I
think
I
agree
with
as
well.
G
Yeah,
that
was
that's
what
I
was
saying
all
right.
Sorry,
audio
settings
were
weird.
A
Okay
cool
so
again,
I
think
that's
a
good
good
area
to
kind
of
complete
out,
because
I
do
think
we
need
a
comprehensive
testing.
You
know
set
of
set
of
tests.
Then
we
have
this.
This
number
60
is
work
in
progress,
so
that
so
that's
what
alex
was
reporting
on
we're
working
through
and
working
on
the
last
pr
for
that
and
just
got
some
comments
from
jurassic.
A
I
think
that
this
ensure
parity
of
the
prometheus
receiver
with
prometheus
server.
I
think
this
is
a
design
task.
Will.
A
Crepe
and
point
counts
out
of
water
samples.
E
Yes,
yeah:
some
of
these
do
exist.
It's
the
the
last
two
or
three
that
do
not
exist.
E
Although,
actually
now
that
I
think
about
it,
we
do
have
the
some
of
these
tests
in
code,
but
what
I
essentially
wanted
to
do
was
set
up
some
sort
of
environment
where
we
have
well.
We
have
you
know
like
a
source
that
produces
data
yeah
and
we
have
one
endpoint,
which
is
the
receiver
and
the
other
being
prometheus
server,
okay,
yep
and
then
yeah.
E
So
you
know
do
like
a
parity
check
because,
for
example,
you
know
the
compliance
tests
pass
and
say:
hey,
you
know,
what
are
they
called
stillness
markers
are
properly.
You
know,
produced
they're
properly
produced.
All
all
is
well,
however,
when
they're
like
targets
that
go
in
and
out
between
scrapes
yeah,
that
could
that
kind
of
condition
isn't
tested
for,
but
we
could
trivially
get
that
if
we
modified
our
test
setup,
we
could,
you
know
quickly,
rule
out
that
edge
case.
A
Okay,
that's
yep,
that's
a
good
point.
I
mean,
I
think,
that's
a
good
test
actually
versus.
B
B
Sending
a
stale
marker
and
does
it
look
correct,
which
is
is
happening,
the
remote
right
exporter
does
that
properly
the
receiver
is
just
not
properly
generating
them.
F
A
A
I'm
not
assigning
it
yet
because
you
know
these
are
all
tests.
These
are
all
tests,
sell
and
re-categorize.
Clearly
that
these
are
a
bunch
of
tests
that
we
need
to
work
on
all
right.
Then
we
have
the
gauge
histogram
and
the
prometheus
receiver.
That's
a
data
model
work.
I
think
that
was
completed.
Actually
josh
you're
here.
Do
you
know
if
josh
sureth
completed
it
josh
was
here?
Did
we
lose
him?
Okay?
E
Nah,
I
don't
think
it's
implemented.
I've
been
translating
a
whole
bunch
of
that
code.
E
C
Have
we.
H
E
Have
cumulative
histograms
we
don't
yet
have
gauge
histograms.
F
A
Open
issues
for
prometheus
components
and
collector
remote
right
seem
to
have
multiple
of
these
that
remove
the
ones
which
are
duplicates.
This
was
something
anthony
that
you
had.
B
Yes,
alex
probably,
if
I'm
wrong
the
phase,
three
pr
has
been
merged
correct.
B
What
was
that?
The
the
third
phase,
pr
that
that
handled
the
prometheus,
discover
configuration,
interception
and
rewriting
that
that
was
merged?
The
one
we're
waiting
on
is
just
the
the
second
part,
with
the
the
allocation
service
right.
C
Sorry,
my
my
internet
was
not
working.
I
couldn't
hear
that
last
bit.
B
No,
actually,
it
looks
like
it
hasn't
been
merged,
so
this
is
still
outstanding.
Jurassic
said:
he's
gonna
review
it
once
the
allocator
services
is
reviewed,
so
the
work
here
is
largely
done.
We
just
need
to
get
through
the
review.
Okay,
what's
the
pr,
I
will
make
it
one.
Second.
A
A
Okay,
let's
move
on
quickly
outline
prometheus
pipeline
and
use
cases.
I
think
we
have
been
doing
these.
We
actually
have
a
pretty
good
list.
I
should
just
document
it
and
list
it
out
here
document
labeling
guidelines
and
recommendations.
This
was,
I
think,
something
josh
was
working
on.
A
Okay,
we'll
need
to
follow
up
on
this
publish
reference,
kubernetes
deployment,
yml
files
and
with
prometheus
configuration
examples.
This
also
looks
like
documentation.
A
B
This
all
together
just
needs
to
be
moved
into
the
contribution.
Now,
since
we
moved
the
component,
it
is
this.
A
A
B
A
B
Yeah,
so
that's
just
waiting
on
those
two
prs
in
the
operator,
repo
for
the
target
allocator
and
the
config
rewrite.
This
is
basically
a
driving.
A
B
We're
sort
of
thinking
about
it
and
there's
an
issue
in
the
collector
repository
that
I
suppose,
probably
needs
to
be
moved
to
the
contrib
repository
now.
That's
assigned
to
me
that'll
be
yeah,
something
I
need
to
start
digging
into.
E
B
E
You
know
the
translation
between
you
know,
like
I
mentioned
before,
p
data
doesn't
use
structs,
it
uses
some
getters
and
setters.
So
just
translating
between
data
types
is
very,
very
robust.
You
get
like
one,
you
get
like
100
lines.
Just
to
add.
You
know,
like
five
data
points
into
yeah,
it's
it's
very,
very
robust.
E
No,
unless
not
really
yeah
exactly
unless
p
data.
E
B
I
think,
what's
been
done
on
the
tree
status
side
is
adding
separate
translator
packages
for
like
jaeger
and
zip
kid
data
that
handle
the
conversion
from
one
to
the
other,
so
that
it
takes
it
out
of
the
the
receiver
and
exporter
components,
but
it
still
needs
to
be
done
somewhere.
F
A
Okay,
I
think
that
at
least
this
is
the
list
that
we
have
here.
What
I'll
do
is
I'll
take
a
look
at
all
the
issues
that
are
in
in
collector
right
now,
which
may
actually
some
of
them
will
probably
get
moved
to
the
token
trip,
given
our
prometheus
components
are
all
there
now
and
I
think
one
of
the
areas
that
emmanuel
you
were
mentioning,
which
I
think
we
should
call
out
as
an
issue
here.
Also,
if
you
can
file
one
is
the
that
we
should
deprecate
the
pull
exporter.
A
There
was
a
discussion
we
had
had
earlier
in
one
of
those
work
group
meetings
there
we
wanted
to
kind
of
at
least
promote
the
idea
of
using
a
remote
right
exporter
and
the
prometheus
pull
exporter
through
the
collector
right
and
not
maintain
language
implementations
in
every
sdk
library.
Oh.
E
Okay,
I
see
yeah
so
the
tricky,
the
very
very
tricky
thing
about
that
is
now
now
kind
kind
of
reflecting
on.
Essentially
what
I
was
saying
was
that,
right
now
you
know
we
have
zipkin
jaeger
and
prometheus,
and
it's
like
a
mandate
for
every
single
language
to
implement
yeah
exporters
for
these.
But
we
should
you
know
we
should
potentially
just
be
pushing
that
everyone
just
use
the
open,
telemetry
collector.
A
B
E
Hey
anthony,
not
truly,
if
I
mean
they
would
basically
replace
the
endpoints
with
the
collector,
but
I
mean
that
that
might
ask
them
to
hey,
add
something
else.
But
if
you
look
at
the
strain
that
provides
to
like
language
implementations,
it's
you
know,
the
the
the
return
on
investment
is
basically
non-existent
because
doing
like
a
prometheus
translation.
Is
it's
not
that
trivial
yeah?
A
F
A
Okay,
I
mean
that's
fine,
I
I'll
also
file
it
discussion
needed
in
specsing
yeah,
so
I'll
open
it
up,
and
I
don't
think
should
we
just
do
it
on
the
spec
I'll.
Do
it
there
too
we'll
find
more
detailed
issue
and
spec
people.
A
A
A
Okay,
I
think
that's
all
I
had
on
my
end,
just
wanted
to
kind
of
walk
through
the
backlog.
So
thanks
guys
thanks
thanks
everyone
for
walking
through
that
anything
else
that
we
have
on
our
notes.
I
don't
think
so
for
anybody
who's
attending
brian.
Did
you
guys
add
your
names
to
this.
D
Okay,
one
thing
I
just
want
to
tell
you
man
on
the
on
the
stainless
marker,
so
you
manual,
I
see
the
the
the
stainless
markers
are
sending
stainless.
D
I
mean
we
are
sending
stainless
marker
for
up
metric,
which
seems
incorrect,
even
though
the
target
is
up
and
running
and
they
are
going
as
nan
and
some
checks
are
actually
failing.
I
see
there
is
a
check
to
to
make
sure
that
up
is
either
zero
or
one.
I
think.
Now
now
it's
nan,
it's
actually
failing
just
fyi
when
you're
going
to
take
a
look
at
the
student's
marker.
E
G
Okay,
yeah,
it
keeps
on
resetting
my
volume
level.
I
don't
know
why.
Okay,
that
should
be
okay,
so
I
know
for
normal
scrapes,
yes,
but
if
a
target
goes
away,
that's
possibly
the
case
for
a
stale
marker.
I
honestly
have
to
check
the
code.
It's
been
a
few
years.
H
D
I
A
E
You
know
what
viscera
said
as
a
suggestion.
I
thought
would
be
correct,
but
the
prometheus
team
has
said
no.
Naturally,
we
should
still
be
sending
stillness
markets
even
for
up.
So
perhaps
we
should
tag
him
in
that
issue
and
then
vishwa
posts.
His
feedback,
then
I'll.
You
know
I'll
incorporate
it
into
the
pr.
D
H
D
But
in
our
case
we
are
actually
sending
for
all
the
targets
all
the
time
as
an
ad.
So
that's
I
will
tag
me
in
that
issue
and
I
will
post
details
in.
E
There
so
just
to
let
you
know,
please
don't
open
a
new
one,
just
chime
into
the
old
one.
H
But
something
that
promoters
is
doing
and
that
you
might
also
maybe
do
is
that
when
we
insert
the
stem
markers
for
the
the
series
like
the
up
matrix,
we
don't
do
it
directly
when
the
target
goes
away.
But
we
wait
one
or
two
script:
script
intervals
before
doing
that,
so
that
if
the
target
is
created
by
the
service
curry
in
some
way,
then
we
don't
have
the
stainless
marker
in
between.
H
G
Yeah,
it
just
depends
basically
on
the
tsd.
Information
will
reject
this.
If,
because
you
know
it's
append
only
so
we
depend
on
that
for
the
work
internally
and
then
it
isn't
ultimately
saying
the
remote
right,
because
it's
not
accepted
by
the
tsdb.
G
A
E
Except
I
don't
know
if
the
issue
exists
anymore,
it
used
to
be
in
the
collector
as
three
seven
three
three.
A
It
could
be
closed.
I
don't
think
it
was
closed.
B
It
should
still
be
there
if,
if
it
got
moved
going,
there
should
follow
it
to
the
new
location
as
well.