►
From YouTube: Loki Community Call 2022-01-05
Description
https://docs.google.com/document/d/1MNjiHQxwFukm2J4NJRWyRgRIiK7VpokYyATzJ5ce-O8/edit
Topics discussed
1. Loki 2.7.1
2. New sort and sort_desc functions in LogQL
3. LogQL formatter
4. Datasource updates:
5. Support 'since' query parameter for easier querying using relative time
A
A
C
A
C
C
A
Get
started
yeah,
we
don't
have
much
agenda
also
today,
also
I.
Think
a
lot
of
folks
are
already
in
holidays,
both
internally
so
yeah.
So
one
change
I
think
from
the
last
Community
College
we
released
a
minor
version,
2.71
and
yeah
I
think
there
is
a
link.
If
you
wanna
know
more
detail,
you
can
click
on
this
and
see
what's
happening
there.
A
A
A
A
B
A
B
A
B
So
there's
a
bug
which
causes
us
to
delete
more
data,
then
that
was
requested.
So
we
had
to
change
it
to
filter
only
until
we
fixed
that
bug
and
do
a
new
release,
so
by
default,
Loki
would
do
query
type
filtering
instead
of
doing
any
deletion
on
the
logs
and
I.
Think
next
release
should
have
this
fix.
I
have
the
fix
already,
but
I
need
to
go
through
the
code
again
and
review
the
tests
so
that
we
don't
miss
any
places
where
we
delete
more
data
than
expected.
B
F
E
A
I
think
my
understanding
there
is
yeah
I,
think
the
it
was
causing
some
weird
error
messages
like
a
404
status
code
when
both
this
middleware
was
enabled.
So
even
though
the
delete
is
I,
think
disabled
I
think
this
was
causing
some
error.
I
guess
so
I
think
he
fixed
that
so
that
we
don't
show
the
song
error
message:
yeah,
that's
my
understanding,
maybe
I'm
missing
something
here.
If
anyone.
B
A
C
C
A
Report
any
box
if
anyone
face
any
issues,
I
think
the
the
one
which
also
got
recently
merged.
Is
this
new
feature
I
think
it's
also
from
one
of
our
contributor,
like
one
of
America
I,
think
external
maintainer?
Is
they
added,
like
a
certain
sort,
description,
start
descending.
Sorry
functions
to
lock
here,
which
is
similar
to
procure?
Has
so
yeah
I
think
it's
it's
basically.
A
A
E
A
Those
kind
of
things
yeah
and
also
it
works
nice
with
the
binary
operator,
let's
say
which
one
executes
first
based
on
the
Precedence
and
stuff
so
yeah,
the
the
actual
change
is
basically
like.
We
have
a
new
endpoint
from
query,
so
we
submit
any
lock
your
query
either
as
a
guitar
post
request.
It
should
return
you
the
formatted
one
and
also
we
integrated
with
the
logoxia
line.
So
you
can
do
log,
CLI
font
and
you
can
just
get
to
the
command
actually,
so
it
started
related
into
grafana.
A
B
A
C
E
All
right
so
now
for
context:
I'm
Matthias
form
from
the
Loki
data
source
team
from
the
logs
team.
So
basically,
we've
been
working
on
making
very
simple
for
everyone
for
from
different
perspective,
from
newcomers
for
to
people
who
extensively
use
looking,
and
one
of
the
feedbacks
that
we
received
was
about
processing
errors.
So
if
you
run
a
body
that
is
wrong,
you
used
to
get.
You
know
this
part
server
that,
like
whatever
column
whatever,
so
what
we
did.
We
implemented
quality
validation,
combining
the
query,
parser
the
syntax
parser
with
the
Monaco
editor.
E
So
you
are
able
to
ask
your
right
queries.
You
can
you
know
as
an
as
in
Visual
Studio
code.
You
get
some
feedback
on
where,
where
in
your
queries-
and
it
displays
a
visual
here
so
after
you
write
queries,
one
of
them
is
valid.
You
will
stop
displaying
error.
So
if
you
make
a
mistake
of
your
query
is
incomplete,
you
will
be
able
to
actually
see
it
and
it
has
another
two
cool
things
such
good
features.
One
is
that
you
usually
will
get.
E
The
errors
that
are
displayed
will
point
to
the
same
line
and
column
as
actually
three,
so
that
is
one
you
know
it
will
match
the
depart.
Server
that
you
get
from
the
response.
Second,
is
that
we
support
interpolation,
because
our
syntax
processor
doesn't
understand
grafana
Concepts
like
variables,
but
it
will.
It
will
actually
interpolate
values
variables
and
then
run
it
through
the
parser
to
get
any
errors
and
this
data,
and
if
there
are
otherwise,
it
doesn't
display
false
negatives
and
the
third
one
is.
That
is
also
support.
A
E
A
F
C
C
It's
a
little
painful
to
like
configure
the
correct
addresses
for
compactors
so
like,
instead
of
using
the
ring
right
now,
there's
like
a
fixed
address
for
the
compactor
and
communication
happens
over
HTTP,
so
it's
like
a
little
bit
harder
to
to
track
and
yeah.
So
yeah.
Are
there
other
plans
to
sort
of
short
that
up?
Would
you
be
accepting
patches
to
fix.
B
That
I
can
yeah.
I
can
talk
about
it,
so
we
have
already
added
support
for
grpc
like
talking
to
compactor
overgrpc,
for
internal
communication
and
regarding
your
question
about
not
not
being
able
to
like
having
to
configure
a
static
address
for
compactor
I
think
we
still
have
to
do
that,
because
we
this
there
should
be
only
one
compactor
responsible
for
performing
deletions
and
applying
custom
retention
and
compaction.
B
A
A
This
problem,
like
personally
all
the
time
and
I,
want
to
particularly
using
log
CLI
when
I
want
to
enter
the
absolute
time
for
the
range
queries.
It's
always
hard,
so
I
mean
the
locks
area.
We
had
like
a
small
hack
to
calculate
the
start
and
end
based
on
the
relative
time,
but
what
this
PR
does
is
it
supports
natively
to
the
query
path?
It's
I
mean
the
end
point
itself.
B
A
A
A
E
G
So
we're
trying
to
think
of
the
best
way
to
do
this
and
the
way
that
we
do
this
internally
at
grafana
Labs
is
we
have
design
documents,
and
so
these
documents
describe
the
background
to
the
problem
and
then
a
few
like
pointing
out
the
particular
all
of
the
individual
problems
that
an
individual
is
facing
and
then
what
they
would
like
to
do
to
solve
it.
And
then
internally,
we
discuss
and
come
to
some
consensus
now
with
the
community.
A
A
A
A
F
What
yeah,
maybe
I,
really
like
the
having
a
main
issue
with
the
premix
plane
and
some
design
guidelines
and
stuff
like
that
discussed,
but
quite
often
having
a
live
discussion,
helps
a
lot
so
I,
don't
know
how
you
do
it
internally.
Did
you?
Do
you
often
have
live
discussions
between
developers
from
different
teams
to
discuss
it,
and
if
it's
something
you
do
internally,
it's
something
that
can
be
open
to
external
contributors.
Just
like
this
decisions
like
like,
we
are
like
we're
having
right
now,
but
with
more
development,
oriented
or
design
oriented.
G
So
if
there's
just
a
discussion
between
two
individuals
that
obviously
wouldn't
be
very
easily
searchable,
it
does
happen,
though,
and
sometimes
these
high
bandwidth
discussions
are
really
useful
and
I
think
you
know
this
has
happened
in
the
past
where
individuals
ask
one
another
like
hey.
If
I
wanted
to
implement
this
feature,
how
should
I
go
about
it
and
then
they
talk
a
little
bit
about
the
implementation
detail,
but
no
agreement
is
really
made
internally
without
a
design
document.
G
So,
even
though
there
would
be
that
high
bandwidth
communication
to
figure
out,
you
know
a
few
different
proposals.
Ultimately,
the
design
document
is
the
the
record
of
all
of
those
decisions
that
are
made,
and
we
would
want
something
like
this
for
the
community
too,
because
we
want
to
be
transparent
and
we
want
to
make
sure
that
all
of
the
inputs
and
outputs
from
these
discussions
are
are
captured.
G
So
anybody
can
go
and
review
them
and
see
if
they
agree
or
not
so
yeah
I
I
do
think
that
there
is
a
space
for
dropping
down
to
slack,
for
example,
and
discussing
like
the
finer
details
of
particular
points.
You
know
we're
certainly
not
going
to
discourage
that,
but
the
the
major
points
I
think
should
be
captured
in
the
in
the
issue.
If
we're
going
to
go
that
route
or
in
the
GitHub
discussion
or
the
design
document.
G
G
Absolutely
yeah
and
and
I'll
I'll
say
that
if
anybody,
if
any
external
contributors
want
to
get
involved-
and
you
know
make
make
a
large
contribution-
it's
it's
perfectly
fine
to
reach
out
on
slack
and
say
to
one
of
the
maintainers
hey.
Can
we
have
a
chat
about
this
quickly
jump
on
a
call?
You
know
I
want
to
implement
this
feature
in
this
way.
Can
we
write
the
design
document
together?
That's
totally
cool.
A
Yeah,
just
just
a
note
here,
I
think
when
all
one
was
doing
this
out
of
order
thing,
I
think
I
think
he
made
a
like
a
good
job
of
creating
this
design
dog
and
putting
it
in
both
open
source
repository
and
also
as
a
part
of
a
document
page
so
that,
even
if
you
know
you
can
go
to
Loki
docs
page,
there
will
be
the
same.
Docs
under
that
there'll
be
like
out
of
order,
removal
constraints
or
something
which.
A
A
A
And
see
why
this
is
why
this
decision
has
been
made.
We
should
be
able
to
find
it
easily.
Usually
if
it's
in
the
part
of
Open
Source
repository,
then
that's
the
ideal
place
in
my
opinion,
so
yeah
I
mean
just
I
just
want
to
find
out
this
out
of
what
are
the
sendouts
is
still
part
of
the,
of
course,
repository
yeah.
G
So
maybe
what
we
do
is
we
have
a
process
where
a
contributor
will
open
a
PR
contributing
the
design
dock
as
in
addition
to
the
documentation
right
and
then
we
just
add
comments
to
the
pr
and
basically
say
I.
Don't
know
about
this
point
or
how
would
you
implement
that
or
whatever
and
then
come
to
some
consensus?
And
then
the
document
is
ultimately
completed
and
then
submitted
and
then
now
it's
part
of
the
document,
a
part
of
the
documentation
forever
and
then
the
implementation
can
come
later
and
reference
that
that
document.
C
So
I
think
one
thing
that
would
be
helpful
with
something
like
that
is
guidelines
and
maybe
templates
for
how
you
want
people
to
write
proposals.
What
do
you
expect
how
to
design
docs?
What
sort
of
changes
you
want
to
see
designed
to
explore
some
open
source
projects
have
more
or
less
formality
around
these
kinds
of
things
like
Kafka
has
like
Kips
like
improvement,
process,
stocks
and
I
think
they
have
guidelines
for
forum
when
you
write
one
versus
just
like
sending
a
patch,
so
I
I.
B
G
A
yeah,
that's
a
really
really
nice
idea.
Absolutely
we
can
add
a
template
for
that
yeah
and
I,
like
the
idea
of
of
the
I.
Think
python
also
does
it
as
well.
Php
they've
got
these
proposals
that
have
a
standard
structure
and
they're
agreed
upon,
and
then
some
consensus
is
found
and
then
they
start
the
implementation.
So
I
think
it
would
be
nice
to
not
reinvent
the
wheel.
There.
A
Yeah
I
think
I
totally
agree
with
the
template
as
well.
I
think
we
can
do
I
mean
in
terms
of
facts
and
items,
I
think
we
can
do
multiple
things
like
a
small
things.
I
guess
one
is
definitely
a
template
like.
Whenever
someone
creates
a
pi,
we
can
make
sure
hey.
If
it's
a
big
enough
feature
is
a
ratio
there.
Is
there
a
consensus
there,
something
like
that,
but.
A
Also
has
a
problem
right,
because
people
create
a
p
I
open
a
PR
when
the
branch
is
ready,
so
the
Taiwan
effort-
that's
been
already
being
spent
there.
So
another
good
thing
would
be
like
to
like
what
Anatolia
saying
like
make
it
more
visible.
The
expectation
right,
I
think
one
good
thing
would
be
like
adding
it
in
the
readme
and
linking
into
the
contribution
guide.
Maybe
that's
some
someone
already
checks
in
the
before
contributing
the
open
source
project.
I,
don't
know,
or
also
we
can
I,
don't
know.
A
If,
if,
if
we
have
a
already
issue
opened,
we
can
add
a
tag.
Maybe
that
needs
design,
for
example,
so
that
no
one
anyone
can
pick
stuff
the
issue
they
can
not
just
go
and
implement
the
feature
directly,
but
just
discuss
the
design.
So
the
the
intention
is
here
like
to
somehow
we
should
enable
them
to
discuss
the
design
up
front
right,
so
yeah
I
think
wherever
it's
possible
to
set
the
expectation.
I
think
that's
that's
an
attempts.
In
my
opinion,
that's
the
something
to
start
with
I
guess.
G
Okay,
great
well
I
think
unless
somebody
else
has
something
else
to
add
I
think
we'll
put
together
a
PR
to
make
these
changes
that
cover.
You
was
mentioning
the
change
to
the
contribution
guides
that
read
me
and
then
the
templates,
and
then
we
can
submit
that
to
the
community
and
see
what
everyone
thinks.
C
A
H
A
B
A
G
A
C
D
Like
a
cooldown
exercise
for
the
end
of
the
cultivation,
yeah
I
wanted
to
ask
you
everybody
like.
What's
the
reason
you
are
here
today
and
what
would
you
like
to
learn
or
to
what
would
you
expect
from
this
call
in
general?
So
we
can
make
this
call
even
better.
So
you
know
if
we
don't
want
to
waste
anybody's
time
we
want
to
make
these
calls
very
useful,
so
wondering
if
there's
something
would
you
like
to
to
see
in
the
future
from
these
calls.
A
D
A
F
I
mean
well,
I
can
start
I,
I,
don't
know
who's
part
of
grafana
and
who's,
not
at
the
end.
So
okay,
I'll
I'll
start
so
yeah
I'm,
not
really
new
in
in
Loki
Community
meetings,
because
I
I
used
to
come
at
at
a
global
us-based
meetings
and
but
yeah
I'm
new
for
the
emea
first
time,
I
joined
first
time,
I
see
you
okay,
we.
F
Yeah,
maybe,
and
so
so,
yeah
I
currently
work
at
JN
swam,
which
is
like
we.
We
provide
some
managed
kubernetes
and
we
I'm
part
of
the
observability
team
and
I'm,
mostly
in
charge
of
setting
up
Loki
for
our
customers.
I'm
a
big
fan
of
everything
you're
doing
at
graphene,
I've
been
using
in
different
companies
for
for
a
few
years
now,
I
was
an
early
adopter
of
of
Lucky
Tempo
and
really
helped
me
in
my
different
jobs.
F
So
so
I'm
joining
to
be
informed
of
what's
going
on
of
what
are
the
new
features
and
stuff
like
that
and
yeah.
Sometimes
talking
about
how
I
plan
to
to
to
implement
lookie
in
my
company
and
if
I'm
doing
some
the
right
design
choices
for
for
the
for
for
the
setup,
so
yeah,
like
I
I,
really
like
live
talk,
I
have
way
different
ways
of
to
to
have
some
some
asynchronous
discussions.
F
A
A
I
Iran
I'm
so
proud
of
myself.
My
name
is
Muhammad.
Amin
I
go
by
I
mean
I'm
from
the
login
team.
At
openshift,
Red
Hat
I've
been
working
a
little
bit
with
the
Loki
Community
recently
since
I
joined
Red
Hat
because
we
use
it
we're
gonna,
be
using
it
mainly
as
our
logging
storage
solution,
I'm
going
to
be
replacing
elasticsearch
over
here,
so
I've
been
contributing.
A
few
PRS
and
I
mainly
work
on
the
operator,
but
I
have
also
contributed
to
the
Loki
main
stack.
Thank.
B
I
Ben-
and
actually
this
is
the
first
time
I've
joined
this-
this
reading,
I
didn't
know
it
existed.
I'm
fairly,
new
to
open
source
I've
only
been
consuming
open
source
I've,
never
really
contributed
something
major
to
open
source,
so
everything
is
kind
of
fairly
new
to
me
and
very
I
think
you
guys
know
Barry
from
from
our
team,
his
teammate,
and
he
encouraged
me
to
to
join
these
meetings.
So
for
my
first
meeting,
my
expectation
was
just
to
say
hi
Israel,
to
the
amazing
people
at
grafana,
and
you
know
alive.
I
You
know
chat
instead
of
just
always
writing
text
going
further.
I
think
my
my
expectation
is
is
two
things
it's
to
stay
informed
on
the
on
everything.
That's
coming,
that's
going
to
to
happen
with
with
Loki,
but
also
to
bring
up
topics
that
we
would
like
to
work
on
at
redhead
and
to
see
if
it,
if
it's
aligned
with
the
final
occupation,
so
yeah
it's
to
search
two
main
reasons:
I'm
going
to
be
joining
this
meeting.
A
To
follow
up
so
I
totally
agree
like
we
should
do
a
better
job
as
a
maintenance
like
to
make
visible
what
are
the
tasks
people
can
pick
it
up
and
like
kind
of
like
a
roadmap
and
everything
I
totally
agree.
I'm
curious
like
how
how
how
do
you
pick
up
this
task
anyway,
like?
Is
it
like?
Just
your
personal
interest,
you
just
go
to
analytical
issue,
or
how
did
you
contribute
this
PRS.
I
So
basically,
the
advancement
of
flow
key
features
also
advances
our
solution
at
openshift
logging.
So
up
until
now,
at
least
our
vision
has
always
been
aligned.
So,
for
example,
we
want
this
new
feature.
It
just
happens
that
Griffin
analogy
also
wants
this
new
feature,
and
so
it
just
goes
smoothly,
and
so
for
now
it's
coming
from
Perry
is
our
team
lead,
he's
doing
a
great
job
at
synchronizing
with
the
tikrofana
Loki
team
and
basically
bringing
the
the
features
that
we
both
like
into
our
backlog
at
brighthead?
I
A
C
D
A
C
A
A
A
H
Roger
go
ahead.
Yeah
I
can
say
something
very
short:
I
work
at
kurfana
for
over
a
year,
but
never
really
joined
the
community.
I
also
really
joined
just
to
see
what
this
is
I.
Think
kavi,
you
invited
me
even
a
few
months
ago
and
I
never
showed
up,
but
it
was
really
nice
to
see
see
these
interactions
here
and
I
even
learned
some
things
like
about
lucky
projects
which
I
did
not
know
in
the
future.
H
D
A
Yeah
yeah
feel
free,
like
yeah
I,
think
in
my
opinion,
based
on
what
people
are
saying
here,
I
think
as
a
blocky
or
yourself
itself,
I
think
we
should
maybe
have
more
visibility
on
the
project,
scope
and
the
project
Direction
More,
public
and
transparent
about
it
right
so
yeah.
Given
we
have
a
lot
of
contributions
coming
in
now,
I
think
that's
a
good
way
to
go.
A
I
think
it's
not
new
I!
Think
serial
I,
don't
know
how
many
people
know
here.
He
took
a
lot
of
steps
before
like
last
year
previous
year,
to
have
a
like
a
project
to
set
the
priorities
and
everything
so
that
anyone
not
only
from
grafana
folks
anyone
like
public,
go
and
check.
What's
on
the
list,
what's
the
priority
and
what's
the
kind
like
like
a
direction,
the
celebrity,
we
want
kind
of
thing
and
they
can
start
working
on.
Unfortunately,
we
don't
have.
B
A
A
D
F
F
First
of
all,
disclaimer
I
think
I
have
some
yeah
some
wishes,
but
maybe
it's
already
on
the
on
the
road
map.
Maybe
I,
just
don't
maybe
it's
already
existing
and
I
just
don't
know
where
they
are,
how
to
use
it.
So,
okay,
but
in
my
opinion
there
are
two
things
that
that
I'd
like
to
see,
maybe
just
miss
them.
F
So
when
I
first
talked
about
that
to
to
you
lucky
people,
the
answer
was:
have
one
Central
looking
stance
but
for
various
reasons
our
customers
don't
want
that.
So
so
yeah
foreign
idea
is
about
I
have
a
feeling
I
I.
Think
Loki
is
really
great
for
for
exploration
of
logs
for
debugging
for
live
querying,
but
I'm,
not
so
sure
about
all
the
compliance
stuff
like
when
some
logs
have
to
be
kept
secure
and
you
have
to
make
sure
they're
not
tempered
or
changed
or
anything
and
for
legal
reasons.
F
You
may
want
to
just
push
them
to
a
volume,
to
a
S3
object
that
separated
and
you
want
these
logs
encrypted
and
signed
so
that
if
there
is
any
legal
stuff,
you
can
read
them
and
you
will
be
sure
that
nobody
chanced
them
and
for
these
reasons,
I
we
some.
We
have
some
pipelines
that
will
retrieve
the
logs
and
send
them
to
lookie
and
send
them
to
something
else.
That
will
that
that
will
store
them.
F
B
A
A
C
A
A
F
D
Sorry,
yeah
yeah,
oh
I,
just
want
to
say
like
with
the
longest
photo.
We
just
like
ship
the
logs
to
your
own
bucket.
So
then
you
can
do
whatever
you
want
with
the
logs,
like
you
know,
encrypt
them
decrypt
them.
The
I
think
the
bad
part
is
that
if
you,
if
you
want
to
read
them
via
Loki
again,
then
you
need
like
they
need
to
be
decrypted,
obviously
yeah.
D
F
F
A
A
A
F
Where,
currently
we
are
we
don't,
we
are
not
really
Advanced
to.
Currently
we
have
one
Central
lucky.
We
had
some
issues
with
the
distributed
setup,
but
we
switched
to
simple
scalable
setup
and
to
the
the
the
the
official
handshots
now.
So
it's
really
simpler.
It
helps
a
lot
and
so
no
I
I
don't
have
anything
to
complain
about
it's
really
it's
working.
Well,
we
are
not
at
Big
scale
yet
for
the
moment.
So
no
no.
D
D
A
A
A
B
A
And
EU
folks
that
should
fit
in
their
time
zone
and
yeah.
If
any
of
you
from
U.S
time
zone,
there
is
also
the
later
one.
All
you
need
to
do
is
just
on
the
top
of
the
dock.
There
is
this
calendar
where
you
can
subscribe
and
yeah.
You
will
see
all
the
meetings
there.
Even
if
you
change
in
the
future,
you
will
get
notified
so
yeah.