►
From YouTube: Monitor:APM Weekly Meeting - 2019-11-06
Description
Discuss issues and 12.6
B
A
B
B
Query
prometheus
yeah,
so
the
decision
I
was
trying
to
follow.
It
seems
like
like
multiple
things
that
are
happening
and
multiple
issues.
So
what
am
I
asking
for
Matt?
Basically,
he
said
either
you
can
assign
some
or
maybe
you
can
go
through
it
and
maybe
I
was
thinking.
Maybe
we
should
close
this
issue
and
open
another
one
with
the
link
to
it
to
explain
exactly
what
is
needed.
B
Originally
I
wanted
to
put
it
in
12.6,
but
I'm,
not
sure
that
we
have
the
right
the
right
strategy
on
how
to
proceed,
because
there
are
some
options
and
then
there
was
another
option
and
I
mean
if
you
read
to
it:
yep
I
don't
want
to
spend
the
time
in
this
meeting
just
going
to
it.
We
can
do
it.
Offline
David
I
saw
that
you
were
also
part
of
this
anyway.
B
B
There
is
another
issue:
I
think
it's
similar,
basically
similar.
Ask:
let's
discuss
this
offline
again.
I,
don't
want
to
wait
to
all
of
it
because
also
follow
the
same
pattern
of
starting
something
and
then
something
else
happen
and
then
suddenly
we
stopped
working
on
it
and
I
don't
have
like
the
context.
B
B
We
have
today
from
cube,
cube
CTL
today,
in
order
to
access
the
logs,
the
user
will
need
to
go
to
the
environment
tab
and
then
he
will
need
to
look
at
the
environment
and
then
click
on
a
specific
pod
and
only
then
he'll
be
able
to
see
the
load
by
the
way
I
was
looking
at
the
telescope
Delta
and
the
amount
of
users
that
are
actually
going
into
that
is
extremely
low.
So
I
suspect
that
one
of
the
reason
it's
very
hard
to
go
into
that
because
you
need
to
do
a
lot
of
clicks.
B
If
you
wish
the
load
and
I
think
it's
a
really
good
functionality.
So,
first
of
all
putting
the
operate.
Putting
a
pod
log
tab
underneath
operation
will
expose
this
functionality
and
ideally
I'm
hoping
it
will
increase
the
amount
of
users.
The
thing
that
we
need
to
discuss-
and
it
is
already
here
so
I
mean
the
discussion-
is
happening
on
the
thread.
I
just
want
to
make
sure
that
everyone
is
well
of
this
discussion.
They
can
came
in
in
the
offline
discussion
as
well.
B
We
can
talk
about
it
in
this
meeting
today
is
that
we
want
to
present
both
option
of
presenting,
looks
from
cube
CTL
directly
from
kubernetes
all
the
logs
that
are
coming
from
elasticsearch
and
basically,
what
we
are
discussing
in
how
to
do
that,
how
to
enable
the
users
to
switch
between
those
two,
because
I
think
it
makes
sense
to
consume
logs
from
the
same
place.
So
that's
the
issue.
B
Whatever
I
forgot
to
add,
maybe
I'll
added
to
the
to
the
agenda
is
an
issue
that
I
open
about
the
12.6
planning
and
I
specifically
wrote
there,
and
it's
on
me,
I
forgot
to
put
it
in
the
agenda.
I
specifically
wrote
that
in
this
coming
coming
new
iteration,
we
will
be
focusing
on
the
logging
part
and
last
Excel.
So
everything
need
to
be
prioritized
accordingly,
because
that
we
want
to
move
logging
into
a
into
viable.
We
are
at
minimal
today
so
want
to
make
it
tribal.
So
everything
about
logs
get
priority
like
higher
priority.
D
B
D
D
A
So
there's
been
a
lot
of
great
comments
in
here:
I
was
gonna,
give
it
another
day
or
so,
and
then
maybe
just
summarize
it
all
and
then
see.
If
we
can
have
some
action
items
that
we
can
take,
there's
specific
things.
We
can
start
with
I'm
just
kind
of
iterate
over
the
process,
but
I
am
super
happy
and
grateful
for
everyone
that
has
taken.
A
You
know
some
time
to
really
think
about
this
and
what's
what's
possible,
causes
and
other
things
that
are
kind
of
slowing
us
down
on
this
I
know
there
I
had
some
ideas
and
but
I
was
glad
to
see
that
other
people
had
shined
in
here,
so
I
think,
let's
not
dig
into
it
too
much
right
more
right.
Now,
if
that's
alright
drudged,
but
maybe
next
weekend
our
meeting,
we
can
kind
of
circle
back.
E
A
way
that
we
could
do
that
by
having
like
a
script
that
we
could
run
against
a
live
cluster
that
had
at
least
seven
days
of
data,
so
that
we
could
mimic
those
preset
and
intervals
that
we
have
and
then
save
that
data
file
somewhere
and
then,
when
you
want
to
load
it
switch
out.
Those
live
Prometheus,
metrics
endpoints
with
the
fake
data.
E
So
my
question
is:
this:
doesn't
really
seem
to
necessarily
address
the
like
intended
users
of
like
application
developers
and
operations.
Teams
like
this
seems
more
directed.
This
solution
seems
more
directed
towards
like
ourselves
as
developers,
so
I'm
wondering
is
that
useful
to
development
is
this?
What
is
this
kind
of
what
this
issue
is
getting
at?
What
do
you
think
is
that
a
viable
and
MBC
yeah.
D
To
me
personally,
I
think
I
think
there
are
two
aspects
to
this
issue.
One
is
to
as
the
epic
sales
to
enhance
metric
dashboards
testing.
So
this
would
be
really
useful
for
us
in
terms
of
having
more
concrete
integration
tests
and
the
second
aspect
to
it
is
helping
developers
like
ourselves
to
easily
load
data
and
quickly
developed
something
so
yeah
I
think
it
is
definitely
useful
in
their
two
perspectives
to
it.
Depending
on
the
solution.
D
B
E
E
Do
we
think
that
that
is
at
least
a
good
first
pass
at
that
we'd?
Have
it
would
only
we'd
only
be
able
to
generate
this
fake
data
off
of
a
deployed
kubernetes,
cluster
and
you'd
want
to
run
it
against?
You
know
one
that
had
been
live,
probably
for
at
least
seven
days
that
you
could
have
metrics
for
seven
days,
yeah,
so
I'm
happy
to
proceed
along
this
path.
If
we
feel
like
this,
this
is
you
know
worthwhile.
B
So
the
end
goal
the
way
the
way
and
see
is
that
we
can
easily
set
up
an
environment
and
quickly
will
be
able
to
see
how
the
data
would
look
like
in
case
everything
is
properly
without
the
need
to
configure
everything
from
from
scratch,
and
if
this
this
is,
this
is
what
we
will
achieve.
I
think
it
is
successful.
F
F
B
I
think
we
can
I
think
we
can
I
mean
they're
out
to
file
at
five
bullet
points
here.
I
think
we
can
break
it
into
five,
meaning
issues
and
say:
okay,
oh
now,
let's
hit
it
love
developer
and
then
we'll
see
how
we
help.
Maybe
in
the
next
iteration
can
you
go
to
demo
environment
if
it
makes
sense,
and
then
we
can
take
it
further.
E
Okay,
yeah:
that's
that
that
sounds
good
to
me
and
I.
Think
that's
what
I
need
so
yeah.
So
if
the
team
feels
like
basically
what
you'd
get
is
somebody
that
has
like
amateur
kubernetes
cluster
that's
been
running
for
a
few
days.
Maybe
you
could
generate
this
one
time
off
of
their
own
and
then
like
they
could
pass
that
to
other
people
on
the
team
so
that
they
could
feed
that
into
theirs
and
without
having
to
set
it
up
themselves.
That's
the
that's!
B
The
part
at
the
West
I
look
at
it
the
priority,
for
that
is
what
what
is
achievable
and
easiest
to
accomplish.
First,
okay,
then
we
can
start
thinking
about
the
next
line,
the
next
one.
Next
one
originally,
when
Kenny
opened
this
issue,
we
didn't
thought
that
you
know
it
will
be
something
that
will
happen
instantly.
G
B
Pink
quality
is
the
test
team
that
need
to
walk
on
it.
I
saw
that
there
were
some
issues
that
appear
on
our
Planning
Board
that
had
like
quality
and
I.
Think
I,
not
this
specific
issue,
but
there
are.
There
were
several
issues
and
I
specifically
think
the
person
that
and
assigned
the
milestone,
so
yeah
I,
think
I,
think
I
think
it's
a
she
I
think
chromium
and
she
told
me
she
told
me
that
let
me
just.
A
H
G
A
B
This
is
what
I
was
I
think,
so
this
is
the
list
of
tasks
that
you
need
to
do
in
order
to
plan
1206
and
I
put
a
comment
that
blogging
remove
it
for
minimum
to
viable
to
happen
by
the
end
of
the
quarter
and
so
I'm.
Basically,
in
the
series,
as
I
mentioned
in
the
beginning
of
the
meeting,
could
be
focusing
on
longing
yeah.
B
B
B
F
F
F
B
F
I
F
Something,
but
if
you
wanted,
if
you
wanted
it
to
be
able
to
like
in
project
one,
if
you
shared
this
between
project,
wanted
project
to
and
then
in
project,
when
you
installed
elasticsearch,
would
you
want
project
two
to
automatically
see
that
elasticsearch
was
being
had
been
been
installed?
I
mean
I
know
this
is
just
prometheus,
but
it's
more
along
the
lines
of
sharing
clusters
versus
sharing
individual
apps
I.
Think
somebody.
F
Because
the
auto
deploy
our
Auto
DevOps
will
automatically
make
an
appropriate
the
named
namespace
for
deploying
that,
so
it
won't
overwrite
anything.
That's
already
out
there
from
an
app
standpoint
so
and
the
the
namespace
for
all
of
this
at
the
medians
and
ingress
and
runner
and
all
those
or
get
lab
managed
apps.
So
they
should
work,
should
just
work.
There
should
be
no
danger
of
so
the
so
runner
would
have
to
be
dealt
with
somewhat
separately
and
from
atheists
prometheus.
B
B
A
K
B
B
F
B
K
Thank
you.
Building
on
this
alerts
button
the
alerts
button
for
some
reason
has
been
moved
way
and
now,
when
the
modal
opens
up,
it
kind
of
opens
up.
Off-Screen
and
I
opened
a
bug,
but
I
think
it's
not
scheduled
until
like
twelve
seven,
but
right
now
the
moral
I
think
it's
actually
off-screen
and
some
of
the
turrets.
So
we
might
want
to
prioritize
fixing
that
bug
since
we're
gonna
be
working
on
which.
K
B
B
B
G
B
K
C
M
B
M
The
reason
we
didn't
make
this
searchable
sorry
I
just
want
to
open
is
because
we
don't
have
a
front-end
component.
Quite
yet
we
have
like
a
bunch
of
selects
and
we're
kind
of
discussing
which
one
we
want
to
like
use
from
now
on.
So
as
long
as
we
have
like
a
response
from
the
maintainer
send,
thank
you
laugh
UI
work
group.
We
can
just
move
ahead
and
use
a
searchable,
select
box
everywhere.
A
M
M
Basically,
there
is
some
missing
work
on
some
of
the
charts
about
a
specific
design
that
Amelia
did
for
this
kind
of
charts,
which
is
like
three
columns
to
display
I
believe
like
most
of
the
charts,
have
them,
but
I
think
the
column
charts
do
not
have
do
not.
So
it's
just
a
matter
of
implementing
the
missing
feature
on
the
column
charts
and
that
should
be
it.
A
B
B
B
A
M
F
I
D
So
the
status
that
you
added
would
be
useful
for
us
and
that's
a
different
issue.
So
if
you
create
an
alert
or
delete
an
alert
and
you
reload
the
page
because
from
week
is
just
figuring
out
about
the
alerts,
I
guess
the
UI
shows
incorrect
state
as
if
there
was
no
cluster
indication
at
all.
So
the
back
and
fix
that
Christy
and
it
would
be
a
source
for
that
this
one
that
we're
looking
at
is
only
UI,
bug
and
I
think
we
only
have
another
issue
open
and
we
had
some
conversations
on
that.
G
Just
a
note
here
there
is
another
potential
issue
so
due
to
the
change
to
psychic
workers,
and
now
the
psychic
workers
will
probably
take
longer
to
complete.
We
might
have
to
increase
the
retry
limit
on
the
prometheus
front
end
because
right
now,
it's
only
three
and
that
counts
for
like
30
seconds.
So
we
probably
need
to
make
it
like
two
minutes.
I.
F
D
A
D
A
I
G
There's
actually
another
related
issue,
so
when
you
normally,
if
you
open
this
page
and
in
six
seconds,
you're,
almost
surely
not
good
in
their
data,
so
you
tend
to
refresh
the
page
now.
The
problem
is:
when
you
open
this
page,
it
makes
queries
to
Prometheus
with
the
start
time
being
your
current
system,
time
down
to
the
second
actually
down
to
the
millisecond.
So
when
you
refresh
the
page,
the
start
time
changes
which
means
that
you're
not
going
to
get
a
cache
hit
on
the
back
end,
which
means
you're,
basically
making
every
request
again.
Oh.
G
F
B
A
K
A
M
So
if
we
solve
the
other
one
because
of
the
timeout
and
then
solve
Reubens
issue,
where
we're
requesting
a
new
time
frame,
so
pretty
much
timing
out
again,
I
said
we
should.
We
should
be
able
to
solve
like
the
original
issue,
so
it
should
be
safe
to
close
at
least
one
just
to
clean
up
and
then
try
to
solve
all
the
issues
altogether.
G
But
so
this
paves
this
matrix.
Dashboard
shows
a
lot
of
data,
so
it
could.
Even
if
they
solve
all
those
issues,
it
could
still
take
a
long
time
to
start
displaying
data.
So
maybe
there's
a
different
problem
here
that
you
know
they
should
show
some
feedback
to
the
user.
But
you
know
it's
happening.
Something
is
going
on
yeah.
C
M
It's
more
like
a
technical
depth
from
what
I
can
see
because
we're
not
using
a
penile
type
component
that
we
have
for
it
like
the
original
dashboard
component.
So
it's
just
a
matter.
Isn't
that
component,
since
that's
the
one
that
like
handles,
which
type
of
graph
is
should
be
rendered
so
should
be
fairly
straightforward.
I
hope
you
know.
Okay,.
A
H
F
A
Yeah
it
doesn't
fit
your
fly
good,
okay,
I
know
that
was
having
those
feature.
Flags
and
getting
them
cleaned
up
is
will
be
very
helpful.
So
we
don't
have
all
this
code
testing
for
the
other
thing,
which
makes
everything
a
little
more
complicated.
So
that's
good
all
right.
Here's
next
one
error,
handling,
QL.
A
D
C
B
C
B
B
B
A
Okay,
good
all
right,
so
that's
all
the
issues
were
out
of
time.
I
took
a
few
notes:
I'll
try
to
update
some
of
the
issues
on
the
agenda
as
we
all
go
through
them
and
I
want
to
do
it
real
time.
So,
yeah
thanks
everyone
and
we
will
talk
shortly-
have
a
good
rest
of
the
day.
However,
much
is
left
for
you,
bye.