►
From YouTube: Fluent Community Meeting June 24th 2021
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Fantastic
okay,
so
just
for
the
recording,
hey
everyone!
This
is
the
first
fluent
community
meeting
in
2021.
A
B
Yeah
hi,
I'm
benjamin
from
coop
sphere,
conspiracy,
team,
I'm
a
couple
sphere
architecture
and
did
the
good
sphere
observability
team
and
we
have
logan
juan
here.
He
is
a
previously
coop
sphere
member
and
he
is
in
holland,
europe
and
now
so
hey.
I
is
the
first
of
flynn
bit
operator
authors,
so
we
invite
him
here,
logan,
you
say
hi
to
everyone.
B
Hello,
greetings
from
netherlands,
yeah
greetings
and
we
have
freeman
here.
He
is
the
open
source
project
manager
of
codesphere,
yeah,
so
I'll
started
to
introducing
the
flume
bit
operator
now
right
share
my
sql,
absolutely
okay,.
B
So,
let's
start
it
why
we
create
a
film
bit
operator.
So,
first
of
all,
I
think
we
allow
flambeat
because
it's
lightweighted
and
costs
less
resource
resources
like
cpu
and
memory,
and
you
know
we
are.
We
are
working
on
the
sphere
container
platform,
so
we
need
to
manipulate
different
bait
configs,
where
incuba
incorporated
style,
where
api
likes
crd
and
something
like
that.
B
So
initially,
actually
we
looked
around
and
find
the
login
operator.
The
login
operator
is
good,
but
there
is
one
point
and
I
that
permit
that
stop
us
using
it,
because
that
is
every
log
that
collected
by
the
fluent
film
beta
has
to
be
sent
to
the
fluently
and
then
forward
to
the
final
destination.
B
So
imagine
if
you
have
a
cluster
with
hundreds,
hundreds
of
nodes,
the
fluent
funding
actually
itself
could
be
become
a
single
point
of
fair
and
could
become
a
bottleneck
so
from
the
flambeater
actually
can
send
logs
to
the
final
destination
itself.
So,
while
not
using
flynn
basis
of
ability
to
send
logs
to
the
final
destination
directly
needn't
to
use
a
fluency
as
a
proxy
yeah,
it
is
friendly
as
as,
basically
it
is
useful
for
in
some
scenarios
it
should
be
optional,
but
should
not
be
a
monetary
option.
B
So
we
decided
to
create
our
own
flame
bait
operator
by
ourselves,
and
then
we
came
across
one
problem.
As
the
issue
shows,
the
film
bit
cannot
dynamically
reload
the
configs.
Whenever
it
changes,
you
have
to
restart
the
front,
build
process
or
the
front
build
port
to
make
it
happen
to
reload
the
config
change
whenever
you
change
the
crd
so
to
solve
that,
we
we
create
a
mechanism
like
this.
We
create
a
process
called
flint
bit
watcher.
B
It
will
watch
for
the
flying
bit,
configs
in
the
front,
build
port
and
whenever,
whenever
this
config
file
changes,
the
from
bit
washer
will
restart
the
frame
bit
process
yeah.
So
the
this
way
different
process,
we
will
load
the
new
config
files,
so
we
still
resolve
it
this
way
and
then
we
we
can
create
our
crd
crds
by
our
ones.
We
have
input,
filter
and
output
crd.
B
Actually,
we
have
a
password
crd
as
well,
and
we
have
a
flambet
configs
rd
you
which
use
the
label
selector
to
select
which
input,
filter
or
output
value
to
be
used,
and
we
have
from
base
crd
to
manage
the
lifecycle
of
the
entire
front
bit
demon
side
yeah.
This
is
the
architecture,
and
this
is
familiar
to
flambeat
users.
I
think
one.
B
There
is
one
thing
to
note:
we
have
add
some
labels
to
the
crds.
You
know
you
can
enable
the
es
output
and
this
will
go
the
kafka
of
forward
output,
so
it's
more
flexible
to
use
and
we
finally
integrate
philippe
operator
in
two
cool
sphere
and
cooksphere.
Now
is
using
from
data
operator
to
manage
the
entire
login
stack.
You
know
the
code
severe
event.
B
Server
will
call
the
front
base
crd
api
and,
from
bit
operator,
will
config
the
frame
bit
config
files
and
manage
the
input,
filter
and
output
to
lock
a
yes
kafka
or
flint
fund
b
yeah.
B
So
this
is
how
it
works,
and
I
have
a
demo
later
to
show
to
show
an
actual
demo
and
I
have
future
plan.
You
know
recently,
we
think
differently
is,
should
be
a
optional
option
should
be
optional.
I
should
not
be
mandatory.
B
You
know
log
message
needed
to
go
through
flindy
before
reaching
the
final
things,
but
still
sometimes
it's
necessary
to
such
as
distributing
logs
to
different
things
for
different
users
is
more
powerful
than
film
based.
So
we
are
considering
ifrandi
crds
yeah.
We
have
a
proposal
already.
B
This
is
our
future
plan
and
we
can
show
you
a
demo
here.
This
is
a
cool
sphere
container
platform.
B
A
B
And
anik
plug-in,
actually
we
on
on
the
coop
sphere,
ui.
We
have,
we
support,
yes,
kafka
and
friendly,
and
actually
you
can
add
a
I'll
show
you
them
this
is
you
can
add
a
kawka
here
right
and
actually
this
is
the
so
many
destinations
could
be
spotted.
We
added
lucky
last
couple
of
months.
I
think
so
the
input
that
we
support
systemd
and
to
yeah
my
passwords,
the
filters
yeah.
B
They
are
all
spotted
yeah
and
we
have
initially
we're
talking
about
the
flambeat
watcher.
It's
a
it's
source
code
actually
previously
is
in
the
the
flambe
water.
The
source
code
was
put
in
a
customized
flame
bit
repository,
but
I
think
it
is
some
something
awkward
because
this
is
a
c
product
and
we
add
some
gold
files
here
and
so
it's
awkward.
B
So
we
decided
to
archive
this
modification
and
move
it
to
the
film
bit
operator
repo
and
put
it
here
so
this
process
is
this
is
written
by
logan
yeah.
It's
just
to
will
restart
the
film
bit
whenever
it's
stop
it
and
it
will
watch
the
film
bit
config
and
whenever
the
flambe
config
changes
it
will
kill
the
flambeat
process.
B
So
these
two
work
together
to
to
implement
the
interior
of
limbit
water
logic
to
reload
the
config
dynamically
yeah.
So
this
is
the
flambeat
water
and
we
have
a
darker
valve
here.
B
Okay,
we
have
a
docker
file
here.
This
dockerfield
will
just
build
the
phone
with
watcher
and
copy
it
to
the
official
film
bit
container
image
and
replace
the
envy
point.
So
we
make
our
own
customized
flambeat
image.
Actually
so
from
operator
is
just
using
this
image
to
to
do
the
entire
logic
yeah.
B
So
I
think
we
have
something
some
yamas
to
show.
This
is
the
the
input
yamo
input
tail
and
we
have
some
customers
who
want
to
collect
the
kubilat
kubrat
logs
and
docker
logs.
We
have
also
this
kind
of
support.
This
is
to
click
collect
the
container
locks.
B
B
Kunitis
filter
we
have
make
some
modification
to
this
to
the
original
date
collapsed,
remove
some
unnecessary
date
and
the
final
details
looks
like.
I
think
the
final
date
is
look
like
this
yeah,
the
the
data
collected
by
the
film
bit
operator.
Yeah,
it's
yeah.
It's
a
remove
some
unnecessary
blocks.
Yes,
it's
I
think
it's
all
my
sharing!
So
guys
do
you
have
any
questions.
B
C
Oh,
that,
that's
great,
I
think
operator
looks
pretty
great.
Actually
I
don't
have
too
much
experience
with
operators,
but
we're
pretty
happy
to
try
to
you
know,
find
the
ways
to
incorporate
all
these
required
changes
like
the
config
watcher
or
find
a
a
smart
way
to
restart
the
process
yeah.
Maybe
we
can
do
some
kind
of
hack
for
now
right,
but
maybe
the
ideal
solution
might
take
more
time
yeah.
But
yes,.
A
A
So
I
think
this
goes
a
little
bit
into
the
the
next
topic
which
is
like
bringing
the
operator
into
like
the
fluent
organization
and
and
if
there's
interest
in
donating
it,
I
think
there's
probably
a
set
of
requirements
that
we'd
all
want
to
to
have
as
well
as
just
being
part
of
the
community.
Saying
hey
like
we
should
use.
You
know
the
community-based
doctor
images
or
we
might
want
to
say.
A
Okay,
we
need
these
two
three
features
to
be
native
and
we
don't
have
to
have
some
some
gopis,
but
I
mean
I,
I
think
that
would
it
would
be
really
cool,
in
my
opinion,
to
to
work
towards
that,
and
then
you
know
even
as
a
as
a
goal
as
a
as
someone
who
always
tries
to
set
the
goal.
It's
like
fluent
con
north
america.
A
Can
we
go
out
and
say,
here's
this
official
operator
that
you
know
the
communities
work
together
to
put
together
cube
spheres
donated
a
big
piece
of
it
and
then
others
have
contributed
on
top.
D
A
And
and
and
so
I
know
if
and
you
had
the
you
had
the
topic
anything
specific
there
that
I
want
to
cover
about
it.
C
Yes,
just
a
little
bit,
I
just
want
to
mention
this
project
from
bit
operator
from
the
beginning
from
the
operator
is
style
started,
and
it
is
totally
a
community
driven
project.
You
know
we
have.
We
have
the
core
maintenance
from
four
companies,
including
digital
ocean
cubesphere,
marshawn
financial
company
and,
as
I
remember,
high
pace,
which
is
a
australia
company.
C
So
I
think
this
project,
you
know
it
doesn't
have
too
many
contributors,
but
we
believe
that
if
we,
if
we
can
contribute
this
project
to
the
upstream
and
I
I
I
believe
that
it
will
gain
more
attention
and
contributors
from
the
front
community,
so
I
think
it
is
it.
It
will
help
us
to
to
grow
this
community
and
to
help
this
project
have
a
great
and
active
community
in
after
this
migration.
C
So
we
we
totally
look
forward
to
this
migration
and
move
forward.
This
contribution-
yeah,
that's
great,
and-
and
I
have
one
question
because-
and
actually
this
is
my
experience
right
when
we
donate
a
fluency
and
with
an
airflow
bit
once
the
first
step
right
is
all
the
alignments
the
company
says
yeah.
We
want
to
donate
this
right,
so
we
get
it
as
part
of
the
fluent
organization.
We
do
all
the
handling
with
the
cncf,
but
the
next
step,
which
is
more
harder,
is
about
the
maintenance
of
the
project.
C
So
because
the
important
part
is
that
companies
I
mean
your
the
companies
that
employs
you
will
give
you
the
time
to
continue
maintaining
the
project,
because
we
got
some
experience
in
the
past
that
we
got
something
donated
and
it's
there
right.
But
the
thing
is
that
it's
a
project
class
that
you
will
be
continue
working
as
maintainers
in
the
project.
So
I
think
that
that
is
the
important
piece
and
I
think
I
don't
have
any
objections.
If
we
have
the
project
plus
maintainers.
B
Yeah
I
want
to
see
we
definitely
will
continue
to
work
on
this
product
yeah,
because
we
actually
in
coop
sphere,
we
use
it
as
a
default
logging
agent.
B
So
we
have
to
maintain
ourselves
and
we
are
looking
forward
to
find
some
beautiful,
more
elegant
solutions
for
the
dynamic
reloading
problem
and
I
think,
once
it's
ready,
we
will
remove
the
front
base
washer
and
use
the
official
frame
with
darker
directly
with
a
reload
interface
like
front
d
and
cause
some
rest
api
and
just
reload
config,
and
it
will
be
great.
I
mean
we
can
yeah
yeah.
We
are
looking
forward
to
the
interface
yeah.
A
One
1.8
is,
is
the
big
thing
that
I
think
that's
we're
finishing
up,
has
multi
new
multi-line
features
multi-format
and
the
prometheus
metrics
for
medium
metric
collection
like
node
exporter,
type
functionality
which
fingers
crossed
have
rc
out
soon
and
then
once
that's
done
yeah.
I
think
I
think
in
general,
there's
definitely
things
that
we
all
feel
is
hey.
This
would
be
good
to
have
so
that
it
feels
like
this
belongs
as
part
of
the
the
organization.
We
probably
need
to
write
those
things
down.
A
Hey,
we
believe,
like
we
need
config
reload.
We
need
these
three
four
features,
I'm
just
making
something
up
and
then
probably
try
to
work
together
at
it.
As
as
a
team
of
like
okay,
hey,
we
can
take
this
on.
We
can
take
this
on,
but
I
I
think
yeah,
maybe
the
best
way
to
do
that
is
just
through
like
a
github
issue.
Okay,
let's
get
a
flintbit
operator
in
maybe
we
move
the
repo
over
mark
it
as
like
hey.
A
This
is
not
yet
ready
for
prime
time
in
as
a
as
a
fluent
project.
If
you
want
go
to
the
the
current
place
and
then
these
are
the
things
that
we're
actively
working
on
and
continue
to
just
sync
up
yeah
every
every
few
weeks
or
so
I
don't
know
any
any
thoughts
on
that
folks.
E
I
mean
it
sounds
reasonable
to
me
I'd
like
to
understand
what
the
scope
of
the
operator
is
as
well,
because
it
says
a
little
bit
operator,
but
obviously
it
does
more
than
for
a
bit
and
where,
when
you
draw
in
the
line,
you
know
you
touch
prometheus
operator,
how
does
it
interact
with
the
other
operators
out
there
as
well?
If
we're
deploying
loki
we're
going
with
a
graphama
stack,
are
we
also
deploying
prometheus?
E
Do
we
configure
that
whole
stack
together?
Do
the
operators
have
to
interact
crds
as
well
having
done
or
doing
an
operator
upgrading
crds
absolute
nightmare,
particularly
when
loads
of
people
use
helm
and
the
helm?
Just
doesn't
do
crd
upgrades,
so
there's
there's
those
kind
of
things
to
consider
as
well.
If
we're
going
with
the
operator
what's
happening
to
the
helm
charts,
are
they
the
same?
Are
they
yeah
how's
how's
that
going
to
be
supported
as
well?
So
I
think
there's
a
few
scoping
things,
but
it
does
sound
good.
E
One
area
I
was
interested
in
as
well
is
currently
the
operator.
Presumably
the
coin
demon
set.
Have
we
looked
at
anything
like
doing
a
sidecar
injection
or
something
like
istio,
where
I've
got
some
some
containers?
I
just
want
to
pull
on.
I
don't
want
a
demon
set,
I'm
not
allowed
to
run
a
demon,
set
those
kind
of
questions
as
well.
That
might
be
interesting
for
my
my
use
case
as
well
and
I'll
I'll
say
as
well.
E
B
Yeah
there's
many
many
things
to
yeah,
just
as
you
mentioned,
yeah.
B
Yeah,
actually,
we
use
promise
use
operator
as
well
and
I
think
we
love
the
way
it
manages
the
workload
so
the
lab
cycle.
So,
as
you
mentioned,
the
helm
doesn't
do
the
operating
part
so
operator
has.
His
places
has
his
to
shine.
Yeah.
E
E
People
just
use
the
helm
chart,
but
don't
upgrade
the
crds
and
then
complain
that
the
operator
is
broken
and
it's
it's
really
hard
to
to
debug
it,
and
it's
really
hard
to
get
people
to
read
documentation
to
do
it
the
right
way
as
well.
So
I
just
need
to
be
clear
what
what's
doing
what
and
if
we're
going
with
a
helm,
chart
the
helm
charts
just
for
fluid
bits.
It's
not
for
maybe
installing
the
operator
or
something
like
that
as
well.
B
I
think
a
fluent
operator,
as
silver
mentioned,
is
also
one
name.
We
thought
we
already
thought
about.
You
know
after
you,
after
we
add
the
fluent
crd,
so
you
can
call
it
a
fluent
operator
in
style
yeah.
So
I
think
I
agree
with
silvas
yeah.
B
We
can
remove
rename
it
to
your
fluent
operator
later.
Sometimes
maybe.
C
Yeah,
I
think
that
yeah
when
we
choose
the
name
that
needs
to
be
the
final
name,
the
fluent
organization
when
things
land
there
changing
names,
is
a
bit
hard
right
and
I
think
that
in
general,
and
so
the
way
that
I
look
at
the
operator
is
pretty
much
the
simplistic
weight
that
I
look
at
as
a
package
right.
C
So
some
people
prefer
chef,
ansible,
other
rpm
packages,
it's
a
deployment
model
right.
So
at
the
end
we
have
to
support
demon
set.
We
have
to
support
hand,
charts
for
an
operator
right
and-
and
I
think
that
the
flexibility
for
the
user
is
what
gets
more
benefits
for
everybody
right.
So
yeah,
I'm
pretty
happy
of
having
an
operator
right.
That
is
one
of
the
missing
pieces
in
the
ecosystem,
from
a
project
itself
and
yeah.
B
D
E
And
you'll
be
fine,
but
yeah.
It's
I
mean
yeah,
because
I
see
like
on
the
slack
channels
and
stuff
a
lot
of
the
problems
people
are
seeing
are
with
the
downstream
connections
as
well,
and
it's
those
kind
of
connections
that
that
would
be
useful.
E
I
mean
I've
looked
at
like
the
grafana
helm,
charts
for
their
for
their
stacks,
and
that's
quite
nice
to
you
know
just
have
a
proof
of
concept.
You
can
just
go,
install
the
stack
everything's
talking
together.
I
don't
need
to
figure
out
why
my
logs
aren't
showing
up
in
grafana
or
something
like
that,
so
that
that
would
be
a
really
useful
use
case.
For
me,.
A
A
You
know
the
community
calls
will
be
on
a
monthly
cadence
and
I'm
I'm
happy
for
us
to
give
like
updates,
but
I
would
actually
think
this
might
be
better
handled
outside
almost
like
a
kind
of
a
working
group
and
we
could
probably
meet
more
frequently
if
needed,
or
you
know,
try
to
prioritize
for
async
communication.
A
I
think
that
it's
very
hard
to
get
a
time
zone
that
works
well
with
with
everyone,
and
I
believe
some
of
you
are
joining
at
midnight.
I
might
check
the
time
zone
so
if,
if
everyone
agrees
with
that,
I
I'm
happy
to
go
ahead
and
create
some
of
those
assets
and.
B
A
B
A
Okay,
great
okay,
I'm
gonna
add
that
as
an
action
for
myself.
F
F
We
actually
have
some
custom
plugins
output,
plugins
that
we
are
looking
to
open
source
at
some
point
in
the
future,
so
one
of
them
is
actually
websocket
plugin.
So
it
was
welcome
to
see
that
to
add
it
to
the
core
and
yeah,
so
we've
been
exploring
so
we're
using
fluid
in
production,
we've
been
exploring
to
use
operator
for
some
picture
work
and
most
of
our
usage
is
kind
of
programmatically.
F
So
some
of
the
work
that
drook
has
done
is
add,
like
the
gold
client
to
the
operator
and
we've
added
a
syslog
output
and
and
some
other
stuff
so
yeah
from
like.
I
guess
you
know
kind
of
this
sustainability
angle.
You
know
we
are
looking
to
utilize.
The
operator
for
production,
picture,
work,
workloads
and
stuff
like
that.
Yeah.
B
And
we
thank
you
for
the
contributions.
Thank
you
for
the
contribution
to
the
flimbit
operator.
Thank
you
and
your
colleague.
I
think
I
I
forgot
his
name.
Yeah.
Sorry,
yeah.
Yes,
yeah,
okay,
you
are
very
professional
in
the
code
yeah.
I
think
thank
you.
A
Awesome
yeah
no
that'll
be
that'll,
be
great
to
to
highlight
so
yeah
when,
when
you
contribute
to
to
the
project-
and
you
have
a
blog
or
something
webinar,
I
think
actually
that's
how
fema
and
I
got
in
contact
was
through.
You
know
some
contribution
of
a
potential
blog.
Maybe
some
cncf
blogs
yeah.
Let
us
know
that's
that's
the
way
we
can
highlight.
Awesome
work
that
community's
done
yeah
pat
just
got
a
blog
out
a
few
days
ago.
So
yeah
sounds
great
cool.
Okay.
A
E
Yeah,
I
think
at
the
moment
it's
just
to
kind
of
you
look
at
git
and
then
go.
Oh,
maybe
I'll
go
ask
ping
one
of
you,
two
and
quite
odd.
There's
a
lot
of
the
same
questions
as
well.
So
yeah
I
mean
I'm
not
saying
something
that
oh
in,
like
you
know
july
first,
this
is
what
you'll
get,
but
just
the
general
idea
would
be
would
be
helpful
of.
What's
coming
up.
You.
A
Know
we
can
maybe
brainstorm.
I
think
that's
a
great
topic
made
for
the
next
community
call
and
I
have
some
ideas
about
like
github
projects,
but
we
could
also
look
at
you
know
other
tools
as
well
I'll,
really
open
the
suggestions
there.
What
folks
have
found
works
well,
it
could
even
be
just
a
google
doc
right,
google
excel
sheet
that
everyone
can
see
but
yeah.
Let's,
let's
brainstorm
on
that.