►
From YouTube: 2022-09-16 meeting
Description
Instrumentation: Messaging
B
C
A
Problem:
okay,
who
does
anybody
else
want
to
share?
Should
I
share.
A
A
Yes,
so
I've
trashed
most
of
the
recent
issues,
one
I
was
not
really
able
to
quickly
determine
what
to
do
with
it
was
this
one.
Basically,
it
looks
like
the
request
is
to
have
a
processor
some
way
of
turning
unix
timestamps
into
timestamps
that
include
a
time
zone
as
part
of
the
timestamp.
D
So
I
know
that
the
transform
processor
will
let
you
screw
with
any
field
and
so
timestamp's
one
of
them,
but
like
it,
doesn't
provide
any
sort
of
guardrails
like
you
can
basically
just
set
the
value
to
anything
you
want
and
right
now,
I'm
pretty
sure
you
could.
You
would
just
have
to
set
with
like
a
a
specific
value.
There's
no
function
for
like
doing
something
smart
with
that,
but
we
could
add
that.
D
E
E
As
a
timestamp
string,
you
know
not
not
a
unix
time
if
you
receive
like
a
timestamp
string
with
a
time
zone
in
it
and
whatever
then
I
suppose
it
can
did.
Is
this
even
in
the
data
model,
though,.
D
And
so
maybe
maybe
the
heck
receiver
like
explicitly
expects
a
string
like
timestamp
to
be
in
the
string
and
then
the
heck
receiver
is
doing
some
calculation
to
convert
that
into
the
otp
data
model.
And
then
it's
not
doing
it.
The
way
that
the
customer
wants.
E
B
E
Yeah,
so
maybe
it's
just
the
maybe
this
is
just
a
bug
and
a
splunk
receiver
that
it's
not
respecting
the
time
zone.
I
don't
know
enough
about
the
format.
I
guess
it's
not
clear
to
me
what
he's
even
asking
for
here,
because
if
it's
not
something
to
do
specifically
with
the
receiver,
then
I
don't.
E
D
E
D
E
E
Not
necessarily
the
location
or
something
like
that,
you
know
it
could
easily
be
something
like
a
lunch
ordering
service
or
something
right.
That's
displaying
a
time
zone
ordered
at
11
30
and
it
matters
what
the
time
zone
is,
but
maybe
they
don't
store
the
exact
location,
information
yeah.
I
don't
know
I'm
kind
of
spitballing
a
little
bit
I
like,
I
said
I
think
that
would
be
a
a
bigger
discussion
for
now.
I
think
we
shove
it
out
to
splunk
and
let
it
go.
B
D
And
then
I'm
curious
actually.
A
E
D
D
Attention
yeah,
I
think
we
need.
We
definitely
need
some
clarification
on
it
because,
again
like
I,
don't
really
see
why
we
should
ever
be
propagating
a
time
like
an
actual
time
stamp
as
like
a
string,
but
it's
whatever,
if
that,
hopefully
demetri
can
follow
up
internally
or
or
in
the
issue
and
figure
out
what
they're
asking
for.
A
Cool
okay,
so
I
was
thinking
next,
we
could
go
through
some
older
issues.
I've
been
trying
to
go
through
most
of
the
issue,
backlog
and
triage
it,
but
I
find
that
the
further
I
go
back,
the
less
context.
I
have
there's
just
a
lot
of
stuff:
that's
changed!
A
B
D
I'm
curious
alex:
do
you
know
by
the
way,
how
long
our
github
stale
action
like
waits
before
it
closes
things.
A
B
A
D
A
E
A
It
doesn't
appear,
that's
the
case.
No,
I
think
applying
a
stale
bot
and
even
doing
something
very
far-reaching,
like
60
or
90
or
180
days,
I
think,
would
be
a
good
start.
C
With
the
latest
issue
yet
opened
there,
I
think
you
can
move
this
to
the
open
telemetry
operator,
and
this
is
caused
as
far
as
understood
by
the
sidecar,
which
gets
injected
from
istio.
It
sets
some
ip
table
rules
so
that
the
collector
is
not
able
to
communicate.
C
And
yeah,
I
have
similar
issue
currently
in
jaeger
and
you
could
ping
me
there
and
I
can
have
a
more
detailed
look.
B
A
I
think
this
one.
I
think
this
was
reverted
at
one
point:
do
you
guys
recall
the
where
this
ended
up.
A
D
D
Not
super
familiar
with
prometheus
does
prometheus.
Do
delta,
metrics.
D
Actually,
it
looks
like
the
one
of
the
prometheus
maintainers
replied.
What
did
he
say
down
there.
E
It
may
for
the
remote
writer,
but
when
you're
scraping,
the
concept
of
a
delta
metric
is
a
little
bit
weird.
A
This
issue
needs
to
get
moved.
Does
anybody
have
right
access
to
the
open,
telemetry
go
build
tools,
repo.
A
D
D
For
the
comp
for
the
existing
comp
labels
evan,
I
those
were
an
original
solution
before
we
like
started
getting
specific
with
the
processor
exporter
or
whatever.
If
you
find
one,
that's
got
comp
on
it
and
it's
not
like
supposed
to
be
a
company.
D
You
can
add
the
correct
label
for
whichever
component
it's
talking
about
or
if
it's
multiple
you
could
add,
multiple,
because
I
think
there's
some
there's
some
com,
comp
labels
that
are
supposed
to
be
for
company
and
there's
some
comp
labels.
That's
like
this
is
a
prometheus
thing,
or
this
is
a
whatever
thing
I
think
prometheus
is
the
one
I
see
the
most.
D
D
Extension,
prometheus
kubernetes,
although
maybe
would
be
vendor.
A
Okay,
so
we
want
just
comp
as
in
company
not
as
in
component.
B
B
E
Specific
company
names
and
vendor
names
may
be
a
mistake
anyways.
If
a
company
is
interested
in
a
particularly
in
a
particular
component,
they
can
easily
follow
that
component
or
you
know,
between
the
the
code
owners
and
like
notification,
filtering
and
stuff,
you
should
be
able
to
filter
down
to
the
ones
that
you
care
about.
I
don't
know
if
we
want
to
necessarily
you
know,
maintain
a
list
of
these
companies
care
about
these
components.
E
There's
nowhere
else
that
really
like,
I
guess,
some
of
the
filtering
stuff,
if
you're
trying
to
get
a
particular
vendor
behavior,
but
for
the
most
part
the
otlp
is
the
only
exporter.
That's
like
cross-vendor.
A
Rob
yeah,
probably,
I
think,
maybe
if
you
have
like
managed
solutions
for
certain
things,
but
no
that's
fair.
I
don't
know
we
can.
We
can
probably
discuss
that
on
an
issue
so
alex
this
one.
Technically,
I
guess
is
this
issue
is,
I
suppose,
more
around
using
the
change
log
generation
tool
inside
of
the
contrib
repo
for
releases,
so
this
one
should
probably
stay
in
contrib.
A
B
Okay,
we're
good
I'm
just
going
through
the
backlog
and
finding
any
issues
that
don't
have
any
labels
and
adding
them
as
need
triage
without
actually
looking
at
the
issue.
So
because
you
see
a
bunch
of
old
issues
that
don't
aren't
relevant
anymore,
it's
because
I'm
not
actually
reading
them
right
now,.
D
It's
pretty
crazy
in
our
repository
how
like
once
an
issue
gets
like
kind
of
old,
like
maybe
even
like
two
months
old,
like
it's
just
like
out
of
mind
and
it's
pretty
hard
to
like
come
back
and
like
actually
deal
with
that
problem
anymore,
like
when
people
open
issues
that
they're
not
like
instantly
like
people
like
component
owners,
the
code
owners,
don't
instantly
say
yeah.
We
can
do
that
like
they
just
kind
of
like
disappear
into
the
board.
E
I
mean
that's
kind
of
unofficial
priority
priority
triage
process
right
if,
if
nobody.
E
It
and
says:
yeah,
that's
a
great
idea,
I'm
going
to
implement
it,
then
maybe
it's
never
going
to
get
implemented.
I
guess
that's
kind
of
the
motivation
behind
the
stale
bar
right
the
priority
later
like
once
you
tackle
all
the
low-hanging
fruit.
Maybe
somebody
reopens
the
issue
later
and
says:
hey.
Can
we
work
on
this
now,
but
you're
never
going
to
dig
into
like
three-year-old
issues
to
look
for
a
feature
to
implement.
B
A
Yeah,
no,
I
definitely
think
that
a
stale
bot
would
be
would
be
good
to
at
least
at
least
label
them
for
review,
or
something
there's
probably
should
be
a
point
where
we
look
at
closing
things,
but
we
could
probably
push
that
out
a
little
bit.
D
E
D
I
think
they
ended
up
making
their
own,
we
recommended
because
they
needed
speed,
that
they
go
make
their
own
component.
Essentially,
because
we
weren't
willing
to
like
reverse
the
cumulative
delta
processor,
they
could
contribute
the
reverse
if
they
wanted,
but.
D
It
would
be
great
if
we
could
just
close
this
issue,
but
you
could
go
put
like
a
really
low
priority
on
it
right
now
and
then
it's
it's
a
new
component
essentially
or
I
think
it's
a
new
component
because
either
way
it's
not
the
cumulative
to
delta
processor
and
it's
a
staple
thing.
So
we
can't
do
it
in
the
transform
processor
either
so
yeah.
That's
where
that
one's.
D
Yeah,
though,
it's
not
technically
the
right
component
request
like
thing,
which
is
why
I
would
love.