►
From YouTube: Grafana UX Feedback Call 2021-03-03
Description
Discussed in this session: improvement ideas for Grafana query editors, the user flow of editing panels, and helping users the right way in the Grafana Cloud signup flow. If you have any feedback after watching, we're happy to read your comments!
A
Happy
to
see
everyone
here
today,
yeah,
we
have
three
topics
on
the
agenda.
I
think
they're
also
all
safe
for
youtube
yay.
So
that
means
we're
gonna
be
sharing
all
of
them
with
the
internet,
hello,
internet.
So
the
first
topic
is
quick
fixes
for
grafana,
query
editors
and
I
will
be
demoing
those
and
we'll
be
showing
you
what
kind
of
quick
fixes
I
came
up
with,
and
then
we
can
discuss
whether
they
make
any
sense.
A
I
also
have
a
panel
edit
user
flow
prepared,
where
I
still
have
a
lot
of
knowledge,
gaps
and
understanding
even
like
how
people
use
the
panel
edit.
So
I'm
curious
about
all
of
your
input
and
maybe,
if
the
youtube
audience
out,
there
also
has
input
on
how
they
edit
panels,
then
feel
free
to
add
comments,
we're
happy
to
read
them
and
then
teddy
wants
to
tell
us
about
api,
key
interaction
and
we'll
show
a
flow
for
that
and
yeah,
because
our
agenda
is
quite
full,
I
would
say:
let's
get
quite
right
into
it.
A
One
thing
before
we
start.
However,
I
hope
that
most
of
you
already
know
we
have
a
note
stock
for
this
call.
I
just
shared
it
in
the
chat.
Everyone
is
a
note
taker,
so
please
help
us
take
notes
and
record
the
feedback
that
everyone
is
giving
so
that
nothing
goes
forgotten
after
this
call,
and
also
we
want
to
assign
to
note-taking
champions,
to
encourage
everyone
else
to
take
notes-
and
I
cannot
do
it
today,
because
I'm
presenting
any
volunteers
who
want
to
make
sure
that
the
notes
are
being
taken.
A
C
A
A
Okay,
yeah,
for
some
reason.
No
now
I
loaded
great.
Thank
you
sorry
for
for
the
delay
all
right,
so
the
problem
that
I
wanted
to
address
with
these
query
editor
quick
fixes
is
about
certain
enterprise.
Plugin,
query
editors,
so
one
of
them
is
the
clout
watch,
query
editor
and
we
have
a
similar
problem
in
the
google
cloud
monitoring.
A
So
basically,
what
you
can
see
here
is
that
first,
you
pick
a
query
mode
and
then
these
metrics
here
region,
namespace
and
metric
name-
they
are
all
linked.
So
if
I
pick
one
oh
great,
this
is
not
working
but
theoretically
like
there
is
a
dependency
between
picking
this
one
and
then
the
next
one.
I
think
it
shows
better
in
the
google
cloud
thing
yeah
now
you
can
see
it
like.
A
I
have
to
select
the
project
first
and
then
I
can
select
service
and
then
the
metric
gets
pre-selected,
but
I
can
select
another
one,
but
if
I
pick
a
different
service
here,
then
I
sync
this
switch
to
something
else.
So
like
it's
hierarchical,
but
in
order
to
select
like
the
deepest
level
of
the
tree,
I
have
to
pick
all
the
others
first
and,
as
you
can
see,
this
like
vertically
arranged
drop
down.
A
Collection
just
takes
up
a
lot
of
vertical
space
and
the
query
edition
gets
a
bit
big
and
clunky,
and
so
I
was
approached
by
some
plug-in
developers
to
make
this
better
and
then
also
what
they
talked
about.
Was
that
sometimes
like?
If
I
wanted
to
group
these,
then
they
would
kind
of
belong
to
the
same
category.
A
So
if
you
would
put
like
project
and
then
next
to
it
service
and
then
the
selector,
the
developers
weren't
quite
happy
with
that,
and
they
wondered
whether
we
could
have
sort
of
like
a
a
super
label
that
would
label
like
all
of
these.
As
a
group
together
and
the
current
query,
editor
components
don't
do
that.
We
only
have
one
type
of
labels
and
we
don't
have
the
option
to
make
label
groups.
A
A
So,
basically,
for
the
field
picker
in
the,
for
example,
in
the
I
think
this
is
a
google
view
there.
They
only
have
one
drop
down
instead
of
like
multiple
levels,
and
if
you
click
it,
then
you
get
sort
of
like
a
modal
view
where
you
can
filter
your
entire
database
and
you
can
see
how
resources
and
scopes
are
grouped
and
you
can
filter
them
and
search
directly.
And
then
you
can
pick
one
element
here.
There
are
some
things
that
are
not
so
great
about
this
interface.
A
For
example,
you
have
checkboxes
here
which
don't
really
make
any
sense,
because
currently
you
cannot
select
multiple
of
these
things
to
monitor
as
your
metric,
you
can
only
stick
one.
So,
theoretically,
we
don't
need
checkboxes,
but
I
did
try
to
steal
this
kind
of
interface
because
I
think
for
all
relational
data
sources
having
the
option
to
explore
your
data
more
deeply
and
see
what
is
actually
in
your
database
could
be
helpful,
especially
as
I
know
that
for
some
enterprise
data
sources,
the
drop
downs
that
we
currently
have
they
sort
of
break.
A
Then
you
don't
even
get
all
the
results
so
having
some
sort
of
modal
view
like
this
one,
where
you
can
filter
and
search
through
your
metrics,
even
if
you
have
a
lot
of
them,
could
be
a
better
option
and
because
this
is
kind
of
a
table
based
view
with
filters,
I
was
inspired
by
the
dashboard
search
that
we
already
have
because,
like
the
filters
on
top
here
are
quite
similar
and
then
we
also
have
the
list
with
folders
and
hierarchy,
and
then
we
also
have
list
items
with
check
boxes.
A
So
I
thought
oh
hey,
that's
quite
similar.
Maybe
I
can.
I
can
reuse
that
and
then
I
also
wanted
to
make
it
similar
to
the
table
list
that
we
already
have,
because
it's
a
bit
more
space
efficient
because
like
these,
they
have
a
lot
of
padding,
and
you
can't
see
so
many
at
once.
A
So
I
try
to
throw
all
of
that
together
into
a
query,
editor
design.
So
basically,
instead
of
having
multiple
drop
downs,
I
thought
like
we
just
call
it
scope
and
we
have
one
drop
down
and
then,
if
you
click
on
it,
you
could
get
a
model
where
you
are
able
to
pick
a
scope
and
have
a
search
box
and
options
to
sort
and
filter
these.
A
And
the
thing
is
that
the
data
that
we
had
in
the
grafana
query
editor
looked
a
bit
different
and
was
a
bit
differently
arranged
than
the
data
that
was
in
this
screen
here.
So
I'm
not
entirely
sure
whether
just
copying
this
and
using
these
categories
or
like
metadata,
makes
perfect
sense.
We'll
still
have
to
discuss
that,
but
we
would
be
able
to
maybe
show
additional
data
like
locations
and
resource
types
and
then
maybe
also
add
a
description
like
in
the
in
the
grafana
query:
editor.
A
A
I
just
played
around
with
multiple
things,
so
I
try
to
find
a
query
editor
that
has
a
problem
with
the
label
group,
and
I
think
this
is
one
like
that
could
also
just
be
a
general
visual
bug
here,
because
this
just
looks
really
off,
but
that
was
how
it
looked
in
my
browser,
so
I
used
this
one
and
tried
to
arrange
it
better
and
thought.
A
Okay,
how
can
I
make
this
aggregate
functions
thing
a
group,
so
this
was
one
idea
that
I
had
to
just
make
a
double
column
label
and
in
this
case
it
works
well,
but
I
thought
okay,
what
if
you
just
have
one
line
and
not
two
and
then
here
I
think,
having
the
super
label
and
then
the
regular
label
look.
The
same
might
be
a
bit
confusing
like
that's,
not
the
most
clear,
so
I
thought.
A
Okay,
could
we
do
like
a
group
area
behind
all
the
labels,
but
then
we
already
have
so
many
different
background
colors,
and
this
could
get
tricky
to
pick
the
right
background
color.
So
I
experimented
like
this
would
be
the
same
background
color
as
the
labels
and
that
looks
kind
of
off.
If
the
labels
don't
have
a
background
color
anymore,
so
this
would
be
sort
of
a
semi
translucent
version.
A
Maybe
this
could
work,
but
that
also
depe
depends
on
the
new
seeming
and
then
I
also
tried
what,
if
we
just
drop
the
label
background
color
completely
for
the
super
labels
and
only
have
labeled
background
colors
for
labels,
and
I
think
that
is
quite
interesting
because
it's
so
clean.
But
then
I
wasn't
sure
like
are
the
color
and
the
rose
still
clear
enough
like?
Is
it
so
easy
enough
to
read
it
like
horizontally?
So
I
also
tried
a
version
with
just
line
dividers,
which
I
think
also
works.
A
Okay,
but
then
we
have
sort
of
like
a
little
bit
of
like
this
line
so
close
to
these
lines.
I
think
that's
a
bit
a
bit
hard
to
look
at
as
well,
and
then
I
also
tried
using
borders
for
the
labels
so
that
we
can
still
use
the
same
background
color.
But
that's
a
bit
weird,
because
we
don't
really
use
borders
for
elements
like
this
anywhere
else
in
grafana,
so
yeah,
two
things
that
you
can
give
me
a
little
bit
of
feedback
on
now.
It's
your
turn.
E
F
Yeah
for
sure
I
I'd
like
to
circle
back
to
that
pop
out
that
the
modal
that
you
had
and
that
kind
of
blew
my
mind
in
in
good
ways,
because
you
kind
of
completely
changed.
As
far
as
I
know,
there's
nothing
else
that
we've
got
that
does
this
and
it
seems
like
a
really
potential
great
way
both
to
handle
varied.
F
Data
sources,
and
also
maybe
this
could
be
a
step
towards
how
we
handle
wizards
in
the
future,
because
I
know
jess
you've
played
with
how
to
make
grafana
easy
enough
that
your
mom
could
could
use
it.
So
maybe
you
know
this
sort
of
space
could
you
know
like
like
have
an
alternate
version
for
other
data
sources.
That
is
a
an
even
simpler
version.
That's
a
wizard
with
more
guidance,
but
anyway,
really
really
cool
way
to
to
think
outside
the
box.
Here.
A
I
myself
have
the
problem
that
I
try
to
query
stuff,
but
usually
I
don't
know
the
data
at
all,
and
then
these
drop
downs
really
make
it
hard
for
me
to
understand
what
is
actually
in
the
database
and
being
able
to
have
all
the
groups
visible
at
once
and
being
able
to
see
how
they
are
nested
and
how
they
are
named
could
be
helpful
for
people
who
don't
know
the
data
so
well
in
exploring
their
stuff
before
they
query
something.
A
So
I
think
that
could
be
interesting,
especially
for
less
advanced
users
and
then
for
more
advanced
users.
They
have
the
upside
that,
instead
of
having
to
click
through
several
layers
of
hierarchy
to
get
where
they
need
to
get,
they
can
just
directly
filter
for
whatever
they're
looking
for,
because
they
might
already
know
the
metric
name.
G
F
Chatting
about
this
daniel
or
other
or
dave,
cal
or
other
people
who
regu,
who
do
actual
queries
on
the
regular.
What
what's
your
take
on
this?
Would
this
be
useful
to
you?
What
jumps
out.
D
Because
I'm
involved
in
this,
so
the
the
example
that
just
showed
there
before
that's
from
the
azure
portal,
so
I
think
josh
is
the
one
who's
suggesting
that
we
should.
We
should
start
using
that
and
you
didn't
actually
show
the
azure
one,
but
that's
actually
far
worse
than
the
two
examples
you
showed
where
it's
not
two
or
three
drop
downs.
It's
like
ten.
I.
D
In
a
row,
you'd
fill
out
to
get
one
metric.
They
have
like
so
many
categories
they
have
to,
and
you
have
to
choose
every
single
category
in
the
right
order
to
be
able
to
show
one
cpu
metric.
So
it's
the
usability
of
that
particular
that
one
is
really
really
painful.
D
I
think
we
also
have
a
similar
problem
for
sql
databases
where,
if
you're
in
grifon
you
just
you
can't
there's
no
easy
way
to
see
like
what
tables
exist
and
what
columns
do
they
have.
So
it's
slightly.
D
Than
this
here,
you're
looking
for
maybe
for
a
a
a
metric
and
there
may
be
it's
rows
and
columns.
So
it's
a
slightly
different
metaphor,
but
you
have
a
similar
problem
so.
A
A
When
I
was
designing
this,
I
was
remembering
the
conversations
we
had
with
the
adx
contacts
from
microsoft,
who
also
wanted
to
get
a
better
overview
of
their
data.
So
I'm
hoping
to
I
mean
I
just
stole
the
terminology
from
this
one
solution,
but
I'm
hoping
to
make
this
more
general,
so
we
can
use
it
for
all
kinds
of
relational
databases,
because
I
think
the
more
tables
you
have
the
more
helpful
this
becomes.
H
So
the
thing
that
the
that
I'm
a
little
bit
confused
in
is
like
the
hierarchy
here
so
because
these
apis,
you
have
to
kind
of
select
one
and
then
the
others.
They
are
very
close.
The
information
is
kind
of
you
have
you
have
to
select
one
thing
first
and
the
other
and
the
other,
and
I'm
not
sure
how
that's
going
to
work
in
this
design.
H
B
Yeah
yeah,
I
think-
maybe
I
might
talk
to
that.
I
I
think
from
like
that
kind
of
technical
feasibility.
I
think,
is
some
questions
we
do
have
around
what
apis
we
could
use.
However,
though,
I
think
this
kind
of
does,
even
even
in
there
might
be
some
new
apis,
we
can
use
that.
Allow
us
to
query
the
data
to
build
this
interface
better,
but
I
think,
even
with
the
existing
kind
of
hierarchical,
hierarchical,
hierarchical
apis,
this
is
actually
still
kind
of
like.
B
H
So
you
can
see
the
top
level
scope
this,
the
second
level
in
separate
sort
of
tables
almost
like
or
separate
columns
next
to
each
other
and
that's
kind
of
how
influx
db.
For
example,
if
you
go
to,
if
you
look
at
their
ui
for
if
you
select
measurement
and
labels,
and
so
it's
kind
of
gets
yeah,
because
this
looks
too
flat
for
actually
showing
you
a
hierarchy.
A
A
Issues
with
the
hover
but
yeah
good
point-
I
we
are
out
of
time,
but
ookas
is
raising
his
hands
so
last
feedback
from
mukash.
I
Sorry
so
I
I,
I
think
it
surfaced
much
more
information
which
is,
which
is
great
and
it
improves
understanding.
The
thing
that
I
am
a
bit
afraid
is
that
we
are
breaking
two
conventions
here.
The
one
is
that
with
query
editors,
you
can
always
get
that
like
a
command
lined
flow
of
working
as
a
with
keyboard,
so
we
go
step
by
step
and
the
second
part
is
here:
we
don't
have
checkboxes.
I
So
the
analogy
from
the
previous
examples
here
is
lost
because
we
are
checking
by
clicking
on
a
row
which
usually
means
in
our
terms
like
going
into
specific
object
or
specific
element
after
after
clicking,
and
that
approach
definitely
needs
to
be
tested
with
those
enterprises
that
sometimes
load
like
thousands
of
rows,
because
with
some
query
editors,
we
hit
the
limits
of
our
current
solutions.
So
if
someone
would
need
to
render
here
like
80
000
elements,
probably
it
would
take
forever.
I
So
only
those
three
considerations,
but
it
is
now
I
understand,
what's
what's
all
of
those
things
that
we
are
picking
from,
so
definitely
improvement
on
that
part.
A
A
I
wanted
to
do
a
content
audit
for
that.
So
the
first
thing
I
tried
for
this
effort
was
to
create
a
user
flow,
and
while
I
created
that
I
noticed
that
I
still
was
confused
about
many
parts
of
the
panel
edit.
So
I
just
want
to
show
you
what
I
currently
understand
and
then
maybe
you
can
enlighten
me
and
also
we
can
discuss
like
which
different
options
different
types
of
users
might
want
to
take
and
whether
there
is
even
like
one
global
user
flow
or
whether
there's
multiple
options.
A
So
basically
the
question
is
like:
are
you
creating
a
panel
from
scratch
and
then
there
is
options
for
yes
or
no,
and
if
yes,
then
probably
you
want
to
start
by
querying
from
querying
some
data,
because
you
don't
have
a
query
yet,
if
you
would
not
create
a
panel
from
scratch
but
copy
a
panel,
then
you
might
already
have
a
query
but
yeah.
Usually,
if
you
want
to
create
a
new
panel,
then
sometimes
it's
like
a
pre-filled
query,
but
when
I
did
that
I
got
basically
an
empty
graphite.
Query
editor,
so
I
knew
okay.
A
I
first
have
to
pick
my
data
source
in
order
to
query
the
right
things.
Then
I
do
want
to
create
a
query
here
like
in
this
case
it's
still
empty.
So
I
have
to
do
that
in
some
cases
you
do
have
like
a
pre-filled
query
here
and
you
already
get
a
graph,
but
sometimes
not,
and
then,
if
you
have
a
proper
working
query,
then
the
visualization
appears
in
this
version
like
when
I
tried
it.
A
I
just
got
the
note
data
feedback
here
and
I
thought
maybe
that's
not
so
ideal,
because
especially
for
and
not
familiar
user
with
this
panel,
edit
flow
users
might
not
know
why
there's
no
data
like
what
does
this
mean?
Oh
it's
just
because
the
query
is
empty,
but
is
it
obvious
to
people
that
the
query
is
empty?
I'm
not
sure.
A
So
that's
something
that
we
could
discuss
and
then
after
the
visualization
appears
either
you
can
go
directly
to
tweak
it
or
you
can
add
a
panel
title
in
the
panel
settings
and
you
can
tweak
the
time
range
if
you
notice.
Oh,
this
is
not
the
specific
data
that
I
want
to
see.
You
could
also
tweak
the
query,
options
or
add
more
queries,
but
yes,
then
I
thought
the
querying
some
data
would
sort
of
be
one
bigger
step
and
then
afterwards,
the
entire
part
of
oh.
A
I
have
a
visualization
now
and
I
want
to
make
it
better,
so
it
serves
me
better.
It
would
be
kind
of
like
a
second
step
in
the
process
of
tweaking
the
visualization,
so
you
would
probably
want
to
pick
a
type
and
then
there
are
different
things
you
can
adjust
like
the
legend,
the
units.
What
is
this
displayed
and
how?
A
But
here
I'm
not
sure
whether
all
these
terms
are
equally
clear
to
users
like
what
options
hide
in
these
accordions
and
how
would
they
know
which
option
to
look
for
in
which
accordion
and
that's
something
that
we
are
already
aware
of,
that
it's
tricky.
That's
why
torkle
is
currently
experimenting
with
how
to
throw
a
search
into
this
panel
settings
pane.
A
But
I
think,
even
if
we
have
a
search,
we
still
have
the
issue
that
maybe
users
aren't
familiar
with
the
terminology
like
they
wouldn't
know
to
look
for
the
legend
name,
so
they
wouldn't
know
how
to
find
it
so
worst
case.
Is
they
try
to
search
for
it,
and
then
they
don't
find
it,
and
then
they
click
through
all
these
accordions
trying
to
find
whatever
they're
looking
for.
A
But
honestly,
I
don't
even
know
like
are
these
more
advanced
options
like,
for
example,
adding
links
having
repeat
options
having
time
regions
tweaking
your
variables?
I
think
these
would
sort
of
be
additional,
because
not
everyone
needs
those.
You
might
need
them
as
a
more
advanced
user
and
then
there's
things
that
I
don't
even
really
know.
Are
they
advanced
settings
or
are
they
something
else
like
field
settings
overrides
and
transforms?
A
A
That,
I
think,
is
that
there
is
a
lot
of
potential
there
to
make
that
clearer
and
better,
but
I
think
we
should
start
with
the
general
structure
and
how
we
arrange
these
options
here
before
we
can
go
into
detail
on
how
we
label
stuff
and
how
we
make
it
searchable
and
yeah.
Then
there's
also
the
difference
between
starting
from
scratch
and
starting
with
the
query
or
tweaking
an
existing
query.
A
A
F
You
chose
the
graph
panel
which
well
it
is
true
that
it
is
currently
our
kind
of
default
panel.
It
is
also
old
and
going
away.
It
is
also
our
only
the
only
panel
that
has
not
had
someone
go
through
it
and
fix
it
up
the
way
that
we
have
the
stat
panel
and
bar
gauge
and
everything
else.
It
is
going
to
be
replaced
by
the
time
series
panel.
I
would
suggest
doing
this
again
with
the
time
series
panel
and
see
if
there
is
it.
F
If
that's
a
little
bit
easier.
However,
most
of
what
you
got
from
most
of
your
analysis
is
still
going
to
be
totally
true
with
the
time
series
panel.
They
just
got
rid
of.
They
took
stuff
like
series
override
and
a
bunch
of
that
and
put
it
in
places
that
make
more
sense
like
field
options
and
overrides
so,
but
yeah
you're,
absolutely
right.
What's
really
weird
with
some
of
the
things,
especially
the
order
is
query.
Options,
for
example,
is
probably
something
that
you
would
and
torkel
and
daniel.
F
Tell
me
if
I'm
totally
wrong
here,
query
options
is
something
that
you
would
probably
fiddle
with
after
you
get
get
results
back
from
your
query,
because
you
use
the
query
options
to
limit
results
and
and
data
points
and
kind
of
make
the
results
more
manageable.
That
query
options
is
on
top,
so
yeah,
the
the
order
of
things
is
totally
not
intuitive.
D
So
yeah,
I
think,
we've
mixed
it
up
with
some
with
transformations
and
other
sort
of
data
options.
D
H
Terms
of
the
transformations
was
part
of
the
panel
options
yeah,
but
yeah
I
mean
this
is
one
of
the
reasons
that
where
I
can,
where
I
proposed
unifying
the
two,
the
field
tabs
and
the
panel
and
the
settings
tabs
yeah
and
and
in
that
process
we
need
to
really
can
also
go
through
the
the
categorization
and
the
order
of
the
of
these
kind
of
collapsible
sections.
H
Because
that's
going
to
be
also
critical,
I
think
for
to
have
and
and
it's
super
hard
to
sort
of,
but
because
we
have
so
many
standard
options
like
min
and
max
and
thresholds.
Where
should
they
be?
Should
they
be
always
in
the
same
position
in
the
side
menu?
H
They
might
be
diff,
sort
of
more
or
less
important
for
different
visualizations,
but
yeah
I
mean
and
but
but
in
this
change
that
I
proposed
a
couple
of
weeks
ago
was
to
get
links
and
repeat
options
and
and
and
so
into
the
panel
settings.
So
so
you
have
the
sort
of
panel
settings
which
has
the
so
it
shouldn't
be
as
cold
settings.
There
should
be
maybe
sort
of
title
and
advanced.
H
F
All
right
terkel,
I
could
not
keep
up
with
that,
so
I
hope
you
can
enter
your
and
I
I
think
that
what
you're
saying
makes
sense
but
to
but
I
think
it
also
is
more
granular
than
the
problem
that
jess
is
really
pointing
out
here
is
that
we
don't
really
have
an
obvious
user
flow
in
in
the
panel
editor.
So
if
you're
an
you,
know
new
or
or
heck,
even
an
intermediate
easier
you
go
in
and
you
look
at
the
panel
there's
no
obvious
trigger
to
okay.
F
F
So
maybe
we
should
think
about
how,
as
we're
fiddling
with
the
panel
editor,
how
how
or
should
we
add
some
some
guidance
there
to
guide
the
user
where
to
go
first
next
and
and
what
a
logical
flow
would
be-
and
I
know
it's
hard
because
there
are
so.
B
F
Different
use
cases,
but
maybe
kind
of
general
guidance,
at
least.
A
Yeah,
that's
why
I
was
trying
to
find
these
sort
of
general
groups
of
tasks
that
people
might
want
to
do
so
that
we
have
something
to
start
from
in
terms
of
like
categories
or
steps,
but
I
can
also
see
that
lukas
is
raising
his
end
again
so
lukas.
Please.
I
Yeah,
I
definitely
represent
the
user,
who
doesn't
know
how
to
deal
with
query
editor
because
that's
all
magic
for
me,
but
the
thing
that
always
bugs
me
when
I
actually
open
it
is
that
I
catch
myself
thinking
why.
But
I
I
have
idea
of
how
I
want
to
see
things,
but
actually
that
type
of
visualization
is
only
a
option
within
all
the
other
stuff
and
the
I
in
my
head,
the
type
of
visualization
impacts,
also
the
query,
or
at
least
like
what
kinds
of
data,
how
aggregated
or
what
type
of
like
information.
I
If,
if
that's
not
a
time
series
database
what
I
need.
So
this
is
like
this:
the
outcome
that
I
expect
or
the
output
I
expect
like
influences.
How?
I
think
of
what
steps
should
I
perform
to
get
that
visualization
in
place
and
that
visualization
is
only
like
under
settings
and
usually
usually
folded.
So
this
is
something
that
always
bugs
me
a
bit
not
much
of.
A
No,
I
perfectly
agree
with
you
there
and
we
have
noticed
this
issue
and
turkey
has
been
working
on
ways
to
get
the
visualization
picker
on
a
global
level
kind
of
out
of
the
settings
pane.
But
I
still
wanted
to
understand
so
at
which
step
in
the
process
do
users
usually
pick
the
visualization?
A
Exactly
I
also
looked
at
competitors
and
I
looked
at
kibana
and
talked
to
geo,
our
elastic
expert
a
bit
about
it
and
there
they
focus
very
much
on
selecting
the
visualization
first
and
then
picking
your
query
and
doing
other
stuff
and
at
least
geo
who
is
used
to
grafana,
obviously
he's
biased,
but
still
he
was
like
yeah.
A
It's
fine
that
I
can
pick
the
visualization
first,
but
then
sometimes
I
query
something,
and
then
I
want
to
change
the
visualization
after
all,
and
then
I
can't
because
in
kibana
it's
not
as
easy,
so
I
think
there
is
probably
like
a
bit
of
both
where
you
might
want
to
select
it
first
but
then
later
on,
you
notice.
Oh
no,
it
doesn't
look
the
way
I
want
and
then
you
want
to
change
it.
J
Yeah,
I'm
I'm
also
trying
to
like,
or
what
I
think
could
be
worth
pursuing,
is
the
sort
of
starting
with
something
non-empty
and
reasonable,
and
then
the
user
refines
more
and
more
right.
So
this
would
mean
having
having
a
panel
with
random
data,
saying
that
this
is
random
data
and
then
you
like,
and
then
you
have
the
option
of
first
either.
Okay.
Well,
I
want
to
change
the
styles
or
I
want
to
tweak
the.
J
I
want
to
tweak
the
query
because
I
want
to
I
want
to
plug
in
my
data
right,
so
this
might
also
satisfy
the
case
where
someone
is
just
keeping
the
tires.
They
don't
even
have
a
data
source
yet,
but
at
least
they
can
play
around
with
the
graph
panel
already
and
they
can
just
try
a
couple
of
visualizations
without
having
to
do
anything
else.
A
Good
point:
I
actually
like
it
that
in
some
data
sources
we
have
the
pre-populated
panels
where
you
already
get
a
graph,
basically
out
of
the
box.
That
always
like
reassures
me
as
a
more
novice
user,
because
I
know
everything
is
working
and
like
now
I
just
have
to
tweet
the
query
yeah.
So
I
will
consider
that
charcoal.
H
Yeah,
that's
something
I've
been
thinking
about
is
related
to
transform
specifically,
but
I
I
think
it'll
relate
to
this
as
well,
so
I
think
grafana
can
help
the
user
a
lot
more
in
terms
of
kind
of
informing
you
that
the
the
data,
your
query
returns,
is
not
good
for
this
visualization
or-
and
maybe
it's
automatically
switched
to
another
data
source.
Visualization
aurius
should
suggest
sort
of
switch
to
this,
so
the
data
you
your
you
have
now
cannot
be.
H
We
have
this
as
only
one
look
for
the
graph
where
you
can
sort
of
switch
to
the
table
when
the
graph
identifies
that
the
data
cannot
be
visualized,
but
I
think
that
that
could
be
improved
and
it
could
also,
I
think,
be
investing
proof
of
transformations
where
you
can.
Maybe
you
can
suggest
transformations
or
or
yeah
have
some
kind
of
interplay
there,
where
grafana
can
guide
the
user
more
or
describe?
Why
we're
not
showing
anything
or
something.
F
A
A
To-Do
list
all
right,
we're
almost
out
of
time,
so
josh
was
raising
his
hand
before
a
tall
girl
josh.
Is
that
still
relevant.
B
B
You
know
I
myself
have
really
had
struggled
with
formulating
the
right
queries
to
kind
of
get
that
end
result
that
I
want
to
understanding
like
the
like
grouping
and
aggregate
functions,
and
things
like
that.
I'm
like
okay,
I
want.
I
know
I
want
to
see
how
many
of
this
is
happening
per
day
but,
like
I,
don't
quite
understand
how
to
get
there
like
those
those
functions.
A
Okay,
yes,
great
points,
I
struggle
with
the
same
stuff,
so
I
can
absolutely
relate
all
right.
So
thank
you
so
much.
I
will
stop
sharing
now
and
head
over
to
teddy.
K
Yes,
I'm
ready,
I'm
looking
for
feedback
on
a
very
specific
ui
interaction
that
I'm
struggling
with.
For
some
reason,
I
feel
like
I'm
just
having
a
brain
freeze
and
I
started
working
on
a
million
permutations
last
night
and
decided
it
would
be
better
to
just
get
some
feedback
in
the
feedback
session.
K
I'm
going
to
give
you
some
background
on
what
I'm
talking
about.
So
what
I'm
working
on
is
a
way
to
make.
Oh
sorry,
the
this
part
of
the
onboarding
wizard
in
grafana
cloud,
easier
to
use.
So
right
now
we
have
all
of
these
agent
integrations
where
the
user
can
install
an
agent
on
their
computer.
K
We
have
these
sort
of
like
plugins,
where
user
can
visualize
existing
data
and
then
this
section
is
meant
to
provide
the
user
instructions
on
how
to
get
their
existing
metrics
logs
and
traces,
although
I'm
not
sure
it
applies
to
traces
just
yet
into
grifon
cloud,
and
so,
if
you
click
on
prometheus
loki
or
graphite,
I'm
going
to
click
on
prometheus
here
and
I
go
next.
K
You
get
a
quick
overview
sentence
and
then,
when
you
click
configure
prometheus,
we
actually
take
the
user
to
their
grafana.com
portal
with
some
instructions,
and
so
what
we
would
like
to
do
is
actually
bake
these
instructions
into
the
wizard,
and
this
will
solve
a
couple
of
problems.
First,
the
user
won't
have
to
flip
back
and
forth
between
grafana.com
and
grafana
cloud.
So
that'll
just
be
like
a
more
seamless
experience.
K
We
would
also
like
to
automate
some
of
this
process,
so
essentially
we
give
the
user
some
instructions
with
these
code
snippets
that
they
have
to
run
on
their
command
line
or
add
to
their
prometheus
configuration
in
the
case
of
prometheus
and
they
have
to
enter
their
own
information,
and
so
we'd
like
to
automate
this
in
the
cloud
in
grafana
cloud
and
then
the
third
thing
that
this
will
solve
is
also
that
there
will
be
sort
of
like
a
next
step
where
they
can
see
specific
information
related
to
this,
and
so
we
want
to
make
all
of
this
better.
K
There
is
the
specific
interaction
that
I
am
working
on
to
generate
the
api
key
and
somehow
put
it
into
this
snippet
and
I'm
struggling
with
how
to
make
this
as
seamless
as
possible.
So
that's
what
I'm
looking
for
feedback
on
specifically,
so
I'm
going
to
jump
over
to
figma
and
the
actual
mockups
that
I'm
working
on.
K
So
this
is
the
new
updated
version
of
the
easy
start
integrations
wizard
that
awana
has
been
working
on,
and
it's
very
beautiful
and
you'll
notice
that
there
is
a
sticky
footer,
which
is
something
that
a
lot
of
people
have
been
requesting.
So
that's
a
huge
improvement,
but
essentially
the
user
would
click
prometheus
locuar
graphite
saying
I
have
my.
K
I
have
a
local
prometheus
instance
and
I
want
to
get
that
data
into
graphonicloud,
so
I'd
click
on
prometheus
and
then
I
come
here
and
I
can
select
whether
I
want
to
send
metrics
from
a
single
instance
from
multiple
instances
and
from
multiple
high
availability
prometheus
instances
and
based
on
that.
The
instructions
will
change
a
bit,
but
the
interaction
with
the
api
key
will
be
the
same
for
all
of
them.
So
I'm
just
going
to
show
one
use
case
and
I
have
a
couple
of
options
for
how
the
user
can
do
this.
K
So
one
option
is
that
this
is
a
an
actual
step.
It's
like
an
explicit
step
to
grant
permissions,
and
essentially
the
user
would
name
the
api
key
and
then
click
generate
and
then
once
they
do,
that
the
ap
they'll
get
a
message
saying
the
air
api
key
has
been
generated
and
placed
into
the
code
snippet
below,
and
then
you
can
see
that
it's
automatically
been
placed
here.
K
I
also
created
a
version
where
the
user
has
to
explicitly
copy
the
api
key
and
then
paste
it
themselves
into
this.
This
snippet-
and
I
think
this
is
an
extra
step
and
it's
a
little
bit
more
cumbersome.
However,
I
have
thought
about
the
fact
that
this
might
be
confusing
that,
like
there's,
no
there's
almost
like
no
closure,
you've
generated
an
api
key
and
you
might
not
understand
that
it's
been
placed
here.
So
that's
something
I've
been
working
through
is
like
what
makes
more
sense.
K
I
also
created
a
flow
where
there's
essentially
like
a
modal
that
comes
up
so
the
user
would
click
generate
new
api
key
and
then
they
have
to
name
it
and
choose
the
metrics
publisher,
name
or
role,
and
in
order
for
this
to
work,
I
think
that
the
role
has
to
be
metrics,
publisher,
and
so
something
else
I
was
thinking
about
is
like.
Should
we
even
make
that
an
option
for
the
user
to
select
the
role
or
should
that
just
happen?
On
the
back
end?
The
api
key
is
created
with
the
metrics
publisher
role.
K
They
would
click
api
key
and
then
they
get
a
message
saying
again.
Your
api
key
has
been
generated
and
pasted
into
the
code
snippet
they
close
it,
and
then
it's
already
here,
so
those
are
sort
of
like
some
of
the
things
I'm
thinking
about
is
like.
Should
we
explicitly
ask
for
the
role
or
should
we
make
that
happen
on
the
back
end,
and
should
we
automatically
paste
it
into
this
snippet,
or
should
we
force
the
user
to
actually
like
paste
it
manually
themselves.
E
I
still
think
what
you
could
do
is
to
show
fourth
step,
which
is
enable
promote
use
remote
right
after
user
has
generated
the
api
key,
so
keep
the
text
as
you
give
feedback
to
the
user
that
he
has
been
generated,
and
then
the
interface
can
shift,
focus
and
scroll
a
bit
down
where
that
is
happening
or
combine
the
message
that
the
key
is
generated.
And
now
you
can
copy
paste
to
clipboard.
F
E
F
So
I
like
your
first
mug
up
the
one
that
automatically
puts
the
key
in
when
they
and
gives
you
a
message.
However,
I
think
that
you
should
tell
them
at
this
point
that
the
api
key
with
the
the
metrics
publisher
or
whatever
permissions
has
been
generated,
because
even
though
they
might
not
care
about
that
right
now,
as
a
beginning
user,
they
will
probably
care
about
it
later,
as
they
learn
more
and
learn
that
api
keys
can
be
associated
with
different
permissions.
F
F
Yeah
and
if
it
turns
out
that
we
that
they
might
want
different
permissions,
then
you
could
also
add
a
field
beside
that
I
mean,
if
metrics
publisher
is
the
only
correct
one
for
the
api
key.
You
know
that's
totally
for
right
now,
then
you
don't
need
to
give
them
a
selection
here,
but
if,
at
a
later.
B
F
F
So
I
like
this
because
there's
there's
room
to
iterate
more
later
as
well,
and
it's
it
makes
it.
It
both
makes
this
simple
for
new
people,
and
it
also
gives
room
to
expand
later
on.
Okay,.
I
I
just
wanted
to
let
you
know
that
for
the
grafana
enterprise
metrics,
where
the
access
policies
are
created
and
tokens
are
generated,
we
went
with
levy
and
the
rest
of
the
team,
the
backend
team,
through
the
api
key
generation
process.
I
think,
like
patrick,
improved,
that
a
bit
as
well,
so
there
is
already
a
flow
happening.
I
I
I
It's
for
the
grafana
enterprise
metrics,
so
I
think
I
think
like
patrick,
is
not
on
the
call,
but
I
think
if
you
ask
patrick
he
will
just
point
you
to
the
specific
and
current
figma
file.
I
would
do
that,
but
I
just
don't
know
right
now,
which
one
is
is
the
current
one?
So
don't
want
to
misguide
you
on
that.
G
A
I
have
a
little
bit
of
feedback.
I
think
in
general,
this
one
page,
clear
steps
directions
are
really
great
and
I
love
that
you
are
giving
lots
of
user
feedback
so
whenever
they
interact
with
the
page
in
some
way,
it's
clear
that
okay,
something
has
happened.
My
interactions
have
triggered
something.
So
that's
very
nice.
The
one
thing
that
I
would
like
to
remark
and
that
also
ties
in
a
little
bit
with
owana's
suggestion
about
the
progressive
disclosure.
A
You
should
consider
these
designs
of
yours
within
the
constrained
viewport
context,
because
if
you
have
multiple
steps
and
then
the
button
is
sort
of
at
the
bottom
of
the
screen
and
the
user
clicks
there
and
then
something
appears
under
it
like
would
it
be
obvious
to
the
user
that
they
still
have
to
keep
scrolling
to
get
to
the
next
step
and
stuff
like
that,
so
yeah,
just
something
to
consider.
L
Yeah,
I
left
most
of
my
notes
in
in
the
notes
my
thoughts,
but
I
would
you
talked
about
that
thing
needing
to
move
between
the
product
and
the
terminal.
L
F
Yeah,
I
appreciate
that
because
I
remember
when
I
tried
setting
up
the
prometheus
stuff
on
my
machine.
This
was
before
the
wizard
or
any
of
that
I
just
was
very
confused.
It
was
very
jarring
to
have
to
go
from
the
portal
which
looks
completely
and
totally
different
from
everything
else,
grafana
ish
to
back
back
to
grafana
so
yeah.
I
like
the
idea
of
keeping
the
the
beginning
users
well
away
from
from
that,
because
it's
speaking
from
experience,
those
instructions
are
kind
of
intimidating.
K
While
I
have
all
of
you
on
the
call
I'd
like
to
get
feedback
on
one
more
thing,
so
I'm
I've
I've
been
working
on
slows
for
prometheus,
loki
and
graphite,
but
prometheus
seems
to
be
the
most
complicated,
because
if
you
have
multiple
instances
or
multiple
high
availability
instances,
there
are
sort
of
like
optional
values
that
you
can
add
to
the
code
snippet.
So
it's
very
simple:
if
you
have
one,
you
literally
copy
and
paste
the
snippet
and
add
it
to
your
on
local
prometheus
instance.
K
B
K
Which
would
essentially
make
it
easier
to
query
specific
data
from
each
of
those
instances
from
grafana
cloud,
so
you
could
query
prometheus
one
or
prometheus
o2
and
then,
in
the
case
of
high
availability
instances.
Gotham
was
explaining
this
to
me
yesterday,
but
essentially
this
would
allow
you
to
reference
one
prometheus
instance.
K
When
another
is
down,
they
both
have
the
same
data,
and
we
make
it
work
that
they're
not
we're
not
storing
both
copies
of
the
data
in
grafana
cloud
and
the
user
has
access
to
it
at
all
times,
even
if
one
like
a
node
is
down,
and
so
for
that
reason
we
give
users
instructions
to
basically
like
right.
Now
I
have
written
instructions,
optionally,
edit,
the
values
for
cluster
and
replica
labels,
or
in
the
case
of
just
the
multiple
instances.
K
It's
just
the
origin
label
and
basically
like
the
intention
here,
is
to
tell
the
user
that
they
can
add
this
to
each
of
the
snippets
for
each
of
their
instances.
But
they'll
have
to
edit
it
themselves,
and
I
was
wondering
if
anyone
had
any
opinions
on
like
how
to
make
this
interaction
easier
or
more
clear
that
this
is
sort
of
like
a
manual
optional
thing
that
the
user
could
do
if
they
have
multiple
instances.
I
Is
the
is
it
like
because,
when
I'm
reading
that
I
am
not
understanding,
but
I
don't
know
what
is
all
that
text
about,
of
course,
but
I
don't
know
instantly
whether
that
optionally
introduces
that
optional
window
as
well
as
well
or
is
it
a
comment
that
is
that
you
can?
I
can
optionally
do
something
with
that
with
that
code,
so
I
think
like.
K
Yeah
so
my
understanding,
this
is
an
interesting
thing
because,
while
it's
optional,
it's
also
like
suggested,
and
so
I
got
some
feedback
that
even
if
this
is
here
like.
K
If,
for
example,
I
gave
these
or
just
this
code
snippet
and
then
gave
them
instructions
afterward
to
add
this
part
to
it,
that's
fine,
but
if
this
is
included,
it
doesn't
negatively
affect
that
instance
like
there's,
no
negative
repercussions,
so
we
might
as
well
just
include
it
in
the
code
snippet
and
then,
if
they
want
to
do
things
with
it,
they
can
like
change
them
for
each
prometheus
instance.
J
So
I
would,
I
would
probably
try
to
figure
out
how
common
each
of
these
cases
are
and
then
maybe
optimize
for
if
there
is
sort
of
a
90
case
here
then
kind
of
optimize
this
whole
workflow.
For
that
one
and
then
try
to
figure
out
how,
where
you
can
hide
the
advanced
stuff
and
then
also
put
all
the
complexity
there
and
essentially
allow
the
user
to
freely
edit
this
thing
and
put
all
the
burden
on
them.
J
E
J
I
guess
I
guess
my
argument
here
is
this:
these
these
things
could
be
simple
to
someone
who
is
managing
a
high
availability
instance
right,
so
they
they
don't
know
what
to
do
right.
On
the
one
hand,
you
have
someone
who
you
actively
trying
to
help
here
and
on
the
other
hand,
if
someone
managed
to
install
like
a
high
building
high
availability,
metric
sort
of
prometheus
setup,
then
I'm
pretty
sure
they
can
handle
this
bit
of
confidence.
Here
yeah
I
mean
yeah
that.