►
From YouTube: Gitlab 12.5 Kickoff - Monitor::APM
Description
Review issues for 12.5
A
B
No,
so
this
is
for
the
keep
of
discussion
for
12.5
I
think
we
should
like
make
make
it
like
something.
The
way
that
the
weekly
meeting
for
a
p.m.
before
the
kickoff
meeting
should
involve
the
kick
of
discussion
for
that
iteration
I
think
this
is
put
into
the
agenda
like
once
in
a
month.
I
think
it
makes
sense
instead
of
having
like
a
dedicated
meeting
for
that,
we
should
leverage
this
meeting
thoughts
or
comments.
B
B
B
B
B
So
thermos
is
one:
it's
an
open
source
project
project
I
think
we're
using
panels
as
well.
I
mean
infrastructure,
team
and
kotas.
Cortex
is
almost
similar,
but
it's
hosted
solution
for
per
meters.
So
it's
a
federal
aid
highway,
lability
hi
available
for
me
to
use
as
a
service-
and
this
is
in
in
a
nutshell.
B
So
the
purpose
of
this
investigation
is
to
find
the
suitable
database
that
can
store
matrix,
can
query
the
matrix
form
store
the
metals
from
primitives,
and
while
we
query
the
data
form
poetess
printers
can
go
to
that
database.
Query
this
data
and
fold
it
to
us.
There
are
several
solutions
that
I'm,
aware
of
and
I
will
list.
The
hill
until
for
mobile
is
a
talk
that
I
so
in
in
cube
con
and
what
they
actually
discuss
this
exact.
So
this
is
one
approach
we
can
take.
So
this
is
the
investigation
and
I
would
like
to.
B
And
the
second
one
is
the
infrastructure
dashboard
and
the
idea
here
is.
We
would
like
to
introduce
some
sort
of
a
birds
of
an
eye
visualization
across
your
underlying
infrastructure.
Basically,
a
question
of
pods
nodes
containers
that,
on
top
of
them
the
workloads,
we
don't
have
a
similar
solution
and
I,
don't
think
there
is
one
for
graph
on.
Even
there
are
many
solutions
that
have
been
provided
by
the
other
of
gendell's
metal
windows
and
I.
B
B
B
This
is
to
store
logs
to
visualize
it
within.
We
didn't
get
lab,
you
are
and
we
may
end
up
by
the
way.
This
is
an
update
that
I
will
put
in
the
issue.
We
may
end
up
shipping
it
with
cabana,
but
so
I
mentioned
without
cabana,
but
and
maybe
if
we
see
that
with
cabana,
we
are
getting
more
value,
maybe
the
first
second
iteration
we
will
enable
keyboard
users.
B
B
Next,
one
is
also
an
investigation,
but
ideally
I
would
like
to
make
sure
it
will
also
be
documented
and
we
even
maybe
publish
a
blog
walk
to
it,
to
see
how
it
goes,
but
basically
I
want
to
simulate
ourself
as
a
user.
That
would
like
to
monitor
any
type
of
application.
Let's
say
is
deploying
MongoDB
you
want,
and
the
user
would
like
to
see
the
metrics
on
this.
So
what
are
the
steps
that
you
need
to
do?
Let's
make
sure
that
they
are
documented
I
foresee
that
this
issue
will
expand,
eventually
create
multiple
issues.
B
B
B
B
B
B
C
B
E
Are
they
you're
right
Ruben?
They
are
different.
Multiple
metrics
just
means
we
can
add
multiple
metrics
that
we're
getting
alerts
for
and
the
computer
one
is
just
expanding
the
language
we
have
to
set
up
those
alerts.
So
computed
is
like,
like
you
see
there,
my
so
we're
capacity
is
greater
than
95
percent,
whereas
multiple
metrics
means
you
could
have
requests
over
capacity
is
greater
than
95
percent
requests
over
capacities
greater
than
ninety
seven
percent
requests
over
capacity.
You
know
you
can
take
the
same
thing
and
just
alert
it.
Multiple
different
threshold.
B
E
B
B
My
clothes
basically
makes
up
logs
available
from
the
operation
tab,
and
today
we
don't
have
it
easy
way
to
go
into
logs
user.
We
need
to
go
to
environment
and
then
click
on
the
pod
and
you
see
the
logs
and
we
want.
You
first
create
a
section
for
pod
load
and
what
we
discussed
in
the
last
meeting
is
that
those
logs
will
be
the
one
that
will
come
from
elasticsearch.
B
B
So
I
didn't
custom
eight
weeks
for
good
luck,
managed
for
me
to
scope
and
I've
also
seen
a
couple
of
issues
around
inconsistent
behavior
when
we
have
for
me
to
set
the
instance
level
and
how
it
affect
the
group
and
the
project
level.
So
apparently
it
is
impossible
to
add
custom
metric
to
a
to
manage
parameters
at
the
group
level
and,
as
I
mentioned,
maybe
the
following
one.
The
following
issue
will
see
similar
behavior
I.
Think.
C
C
E
D
C
B
B
Not
show
up
correctly,
this
is
what
the
issue
mentioning
the
thing
that
I
found
very
interesting
is
the
fact
that
Josh
thought
that
solving
this
issue
will
fix
the
problem,
and
you
can
see
that
seven
month
after
world
there
was
another
user
that
complain
about
this
issues
were
so.
We
need
to
look
into
that,
but
I
understand
why
maybe
she
was
not
fixed
and
we
have
two
users
that
complain
about
that.
B
So
here
another
one
about
Auto
deploy
committees
can
only
use
one
project
and
there
is
also
like
two
issues
that
I
think
are
the
same.
So
this
is
one
when
you
install
vomitus
and
kubernetes.
It
ends
up
in
this
namespace
and
integration,
so
mostly
in
Abel,
but
other
projects
no
longer
install
meters,
but.
C
Not
if
I
mean,
if
it's
at
a
group
level
or
a
project
level,
they
can
use
it
and
the
the
namespace
the
application
name
space
is
like
group
or
project
name
ID.
You
know
production
or
staging
or
whatever,
for
the
for
the
different
environments.
So
the
namespaces
for
the
applications
are
unique,
but
it
is
certainly
a
case
with
instance,
level
of
level
kubernetes
installed
from
atheists
that
they
can
do.
Multiple
applications
can
take
advantage
of
the
same
prometheus.
E
C
E
C
B
C
In
a
group
or
whatever
it
several
times,
I've
wanted
to
be
able
to
oh
I've
got
this
other
project.
Can
I
tell
this
project
to
slurp
up
all
the
you
know,
the
kubernetes
information
that
I've
got
over
some
other
project.
You
know
I
always
make
my
kubernetes
clusters
of
the
instance
level,
which
may
not
be
available
for
many
customers,
so
it
would
be
nice
to
be
able
to
copy
cluster
from
another
project.
So.
E
C
And
it's
an
advanced
topic
to
get
rid
of
the
helm,
tiller
and
stuff.
That's
inside
the
that's
deployed
inside
the
kubernetes,
you
have
to
have
some
serious
kubernetes
knowledge.
I
would
love
to
have
a
copy
cluster.
You
know
feature
to
copy
a
cluster
from
project
one
to
project
two
like
customer,
here's
one
today.
That
would
be
really
awesome,
but
that
would
mean
that
when,
like,
if
I,
if
I
install
from
atheists
on
project
one,
the
project
two
would
need
to,
it
would
need
to
magically
show
up
for
project
two
as
well.
C
E
E
I
think
I
think
probably
at
this
point
we're
at
like
the
feature
exploration
level
like
how
do
we
ease
this
pain
point
for
this
specific
issue,
though
I
think
the
answer
that
we
currently
have
is
to
set
up
Prometheus
at
the
group
level
or
at
the
instance
level
if
they
want
to
use
it
for
multiple
projects.
But
as
David
was
saying
you
know
if
we
want
to
ease
this
user
experience,
I
guess
what
I'm
saying
is
we
do
support
using
auto
deploy
Prometheus
for
more
than
one
project.
It's
just
there's
some
pain
points
involved.
E
If
you
already
started
locked
into
one
project
getting
that
you
know
to
use
other
places,
so
we
could
we
could
do
some
exploration
like
hey.
Can
we
make
this
easier
for
our
users,
because
this
is
happening
where
they
change
their
mind
midstream,
but
actually
I'm,
not
even
sure
that
this
particular
user
knows
that
hey
we
do
support
this.
You
can
just
gonna
have
to
set
it
up
at
a
group
level.
This.
E
E
So
for
this
issue,
we
could
probably
just
comment
on
that
spawn
a
new
issue
off
of
this
to
explore,
making
it
easier
and
then
kind
of
schedule
that
issue
whenever
we
want,
but
it
would
take
some
as
David
was
saying
it
sounds
like
it'll
take
some
exploration
as
to
what
the
best
way
to
do.
That
is,
and
you
know
probably
just
do
a
spike
issue
on
this.
For
now.
B
A
B
C
B
B
E
E
I,
don't
know
that
it's
the
same
thing,
though
I
think
this
is
more
about
like
auto,
like
we
know
that
we
have
a
kubernetes
cluster
installed,
but
we
should
be
able
to
auto
detect
the
presence
of
the
I
guess.
This
would
be
the
pre-existing
presence
of
Prometheus
in
that
cluster,
so
this
would
be
more
for,
like
a
custom,
managed
Prometheus,
maybe
well.
A
A
C
B
C
E
C
F
C
We
wanted
to
use
them,
as
you
know,
assuming
that
the
permission
that's
correct
between
the
two
projects
had
the
same
have
similar
permissions,
but
it
gets
to
be.
It
gets
to
be
complicated
if
you
know
if,
if
one
user
has
access
to
a
subset
of
projects,
that
is
intersected
by
the
set
of
projects
that
another
user
would
have
access
to,
but
they
kind
of
intersect
on
one
or
two
projects,
and
that
means
that
I
could
deploy
apps.
C
It
would
mean
that
and
I'm
not
sure
about
it,
but
if
there's
an
intersection
of
our
other
projects
that
we
can
have,
access
to
and
I
have
and
I
add
one
to
the
the
intersected
project
and
to
a
not
intersected
project,
then
the
person
that
would
have
if
they
took
over
my
project
I
think
we
just
think
about
the
permissions
associated
with
it.
So.
C
C
D
E
E
B
Like
a
spike
issue
from
this
one
and
set
it
up
to
them,
maybe
this
duration
or
next
one,
and
this
one
is
something
that
we
need
to
also
create
some
sort
of
a
spy
issue
but
see
if,
while
we
are
working
on
this
one,
maybe
we
can
ease
the
way.
We
are
thinking
about
that
when
an
owl
had
a
chest
with
Josh
to
better
understand
new
stencil
didn't
have
time
to
walk
talk
to
him.
Bob.
B
C
C
E
B
B
D
Many
of
the
backend
API
is
use
caching,
so
the
thing
is
any
pair
that
uses
caching
when
it
gets
in
a
request.
It
checks
whether
this
particular
request.
The
result
of
this
particular
request
has
already
been
cached.
If
it's
not
cached,
it
returns
to
zero.
Four
two
zero,
two
old
API
is
returned
to
zero
for
I.
Think
the
newer
ones
will
return
to
zero
two,
so
the
front
end
is
supposed
to
retry
until
you
get
a
200
response
now
here.
D
E
Yeah,
so
the
front
end
is
I,
don't
know,
I
mean
we
need
to
look
into
make
sure
the
front.
It
is
actually
retrying
in
the
first
place,
but
even
if
the
front
end
is
retrying,
it's
a
significant
potential,
significant
problem,
because
the
browser's
just
gonna
say
oh
I,
know
the
response
to
this.
This
is
cash.
Then
just
you're
gonna
sit
there
retrying
until
it
decides
that
cash
is
over
and
then
you
know.
D
E
D
E
E
F
D
B
E
May
be
an
epic
worth
creating,
and
then
we
can
gather
these
little
issues
where
we're
actually
doing
it.
I
think
it's
helpful
to
have
them
broken
down
into
their
specific
places
like
where
we're
doing
any,
where
we're
not
so
maybe
yeah,
maybe
an
epic
for
addressing
this,
making
sure
we're
retrying
on
to
zero
fours
and
on
to
zero
twos,
and
then
add
this
one
to
that
epic
and
then,
as
we
find
them,
we
can
just
keep
adding
them
to
those
to
that.
Epic,
okay,.
B
E
B
B
Okay,
and
apparently
there
is
one
metric
from
parameters
that
is
behind
feature
flag
on
comm
I.
Think
it
from
the
discussion
doesn't
really
nobody
understand
why
nobody
responded.
Why
so
I
think
we
need
to
move
the
future,
and
this
wasn't
asked
from
Miguel
to
bomb
the
version
of
prometheus.
We
ship
with
gitlab.
E
I
think,
actually,
that
previous
one
is
already
being
taken
care
of
by
this
one,
this
one
or
no
I'm.
Sorry,
the
one
after
this,
the
bum
prometheus
version,
I
think
this
is
already
being
handled
in
order
commented
on
this
I
think
if
you
scroll
down,
but
this
may
already
be
addressed
in
a
separate
issue
up
just
a
little
more
yeah
yeah
there.
It
is
so
he's
already
working
on
one
on
those
two
issues
which
should
address
this.
E
C
B
E
B
B
D
F
B
F
F
B
Okay,
this
one,
the
lodge-
you
had
a
comment
on
that.
So,
basically,
when
there
is
no
data
on
the
chart,
you
have
to
still
you
click
on
it.
It
doesn't
do
anything
so
the
lodge
your
comment
is
that
let's
simply
remove
all
the
headers
with
no
data
so
that
we
populate
it.
So
only
when
the
data
is
in,
we
can
show
this
one.
My
comment
here
is:
we
need
to
make
sure
history
is
also
relevant.
B
B
E
Used
to
also
do
it
that
way
where
we
just
wouldn't
show
it.
The
reason
we
changed
was
because
we're
allowing
custom
dashboards.
So
if
a
user
goes
in
and
adds
like
some
custom
metrics
for
you
know
and
nginx
ingress
or
something
and
then
there's
no
data,
it
would
be
an
odd
user
experience
or
they
may
think
something's
wrong,
because
they're
creating
a
group
and
then
nothing's
showing
up
and
there's
just
no
feedback
at
all.
E
As
to
why
nothing's,
showing
up
there
I
think
what
we
agreed
we
were
going
to
do
was
in
these
cases
we
would
start
with
them
collapsed
and
then,
when
you
click
it,
if
there
really
is
nothing
configured,
we
would
just
have
an
empty
state
that
would
link
them
to
okay.
You
need
to
configure
nginx
ingress
in
order
to
see
stuff
here
or
you
need
to
configure.
You
know
whatever
it
is
so
we
can
have.
E
D
F
Because
I
feel
like
maybe
it
leaves
this
a
one
line
message
in
that
products
of
the
section
saying
that
as
you're
seeing
an
empty
section
of
them,
because
there's
nothing
configured
would
be
helpful
them
removing
the
arrow
all
together
and
you're,
just
seeing
the
gray
section!
Well,
that's
not
clickable
doesn't
open,
doesn't
have
narrow,
but.
B
B
A
A
Definitely
a
few
of
those
we
had
some
good
discussion
on
so
I'll
I
tried
to
take
some
notes,
so
I'll
try
to
update
the
issues
today.
If
I
have,
if
I
can't
and
then
you
know
some
other
people
mentioned,
they
would
add
some
context
to
so
the
more
details
we
can
put
in
each
of
these
issues,
the
better
yeah.
B
B
Yeah
yeah,
so
first
thanks
for
labeling
those
issues
and
to
better
prioritize
where
to
put
them
I
think
that
we
should
take
also
address
suggestion
about
weights
a
on
the
issues
and
I.
Don't
think
it
needs
to
be
like
to
this
book.
We
need
to
put
weights,
at
least
for
me,
okay,
at
least
for
me.
If
you
will
be
able
to
put
weights
when
we
are
moving
sliding
issues
from
one
generation
to
another,
it
will
be
very,
very
helpful,
so
not
on
all
issues.
B
E
Not
I
think
Kenny
was
a
little
concern
that
we
have
like
a
sliding
weight
scale
that
goes
down
but
I,
don't
know,
I
mean
we
can
use
weights,
however,
we're
not
using
them
at
all
right
now,
so
maybe
even
if
weights
just
represent
whatever
the
current
milestone
is
and
then
it's
okay
to
change
it.
You
know
if
it
does
have
to
get
scheduled
to
the
next
one,
then
we
know
I
mean
it
is
weird
right:
it's
like
an
issue
waited
ten
or
twelve
or
something
I.
B
I'll
tell
you
what
what
what
will
help
me
so
when
I
say
that
to
schedule
now,
I
can
look
to
the
issues.
Okay,
someone
walk
on
that
issue
or
someone
maybe
didn't
walk
on
that
issue
and
if
I
say
that
nobody
walk
on
that
issue,
there
is
no
comments.
Nothing
and
I
think.
Maybe
this
one
is
not
as
important
for
this
milestone:
I
push
it
to
the
next
one
or
maybe
to
the
back.
That's
fine,
but
then
there
are
some
issues
where
people
are
actually
putting
some
effort
and
say:
okay.
B
B
He
walk
already
on
that
week
and
like
50%
of
the
work
is
done,
maybe
say
we
say
one
day,
but
he
didn't
work
on
that
either
so
putting
weights
and
if
we
can
like
say
internally,
zero
to
ten
return
is,
let's
say:
Kang
is
didn't
put
any
at
all
to
that.
550
percent
of
the
walk
is
already
done.
You
know
three,
two
one
and
this
way
how
to
like
things
if
I
see.
A
B
A
B
A
A
E
B
E
Yeah,
probably
especially
because
we
need
to
have
somewhere
to
point
if
some
have
a
community
member
is
like
what
this
issue
is.
Not
you
know
five
wait
like
well,
we
use
them.
This
is
how
we
use
them.
This
is
how
which
is
I,
don't
know.
Well,
we
may
get
pushback
on
that,
but
let's
start,
let's
just
try
it
and
we'll
adapt
as
we
go.
A
D
D
A
A
A
E
A
Yeah
we'll
plan
on
doing
this
once
the
milestone
at
least
and
maybe
kind
of
going
through
towards
EE.
Maybe
every
couple
weeks
we
can
kind
of
alternate
like
here's
go
through
the
whole
board
again
and
here's
what's
left
kind
of
thing.
So
we
kind
of
did
that
last
week
as
well
during
our
weekly
meeting.
So
I
think
that.