►
From YouTube: 2021 08 04 Sharding Group Sync2
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).
C
C
A
They
do,
let's
start,
there's
nothing
in
the
agenda.
Does
anyone
have
anything.
B
So,
to
give
an
update,
we're
making
progress,
I'm
just
setting
up
now
j
meter.
I
have
a
problem
yesterday
with
the
srv
team
that
simply
cannot
reprovision
the
boxes.
There
is
some
bugging
something
that
they
added
regarding
versions
of
ubuntu,
because
since
they
are
attempting
to
migrate,
ubuntu,
20
or
18,
but
in
production
we
have
16
and
I'm
keeping
the
environment
as
production.
B
C
Not
sure
what
I
thought
I
know
the
one
that
you
reviewed
all
of
these
weeks.
D
I
I
my
day
started
today
with
ten
merge
requests
assigned
to
me.
Several
of
them
were
yours.
I
think-
and
I
have
just
finished
them
before
this
call
finished
everything
aside
to
me,
but
yeah.
Thank
you
for
putting
all
of
that
together
and
actually
making
some
of
our
abstract
conversations
more
concrete.
C
C
So
I'm
still
like
kind
of
in
a
place
that
like
we,
are
not
breaking
anyone.
So
let's
try
something,
definitely
much
closer
to
like
what
we
discussed
with
with
a
tongue
on
on
another
issue
about
the
singular
structure
because,
like
I
think
like
at
least
in
my
perspective
like
I
have
like
much
better
perspective
on
how
to
do
some
of
these
stuff.
C
Like
this,
like
cross
noise,
I
have
still
like
pretty
blurry
like
how
to
do
the
structure
sql
stuff,
but
I'm
kind
of
also
anticipating
that,
like
if
we
can't
match
these
things
earlier,
we
will
be
one
step
closer
to
figure
out
what
is
like
the
best
way,
and
hopefully
like
at
least
with
these
like
crossties.
C
This
could
be
like
a
good
building
block
for
the
work.
That
adam
is
right
now
doing
on
the
two
pc
because,
like
we
may
not
yet
enable
that
everywhere,
but
like
we
can
have
some
bits
initially,
rather
to
kind
of
like
refine
them
once
we
start
leverage
them
in
much
wider
and
like
also
like
this
is
the
way
that
would
be
the
disabled
ones.
How
to
handle
that,
like.
We
could
actually
like
figure
out
how
yeah
nice.
D
That
that
disabled
joins,
I
I
realized
my
comments,
saying
it's
not
useful,
yet
it's
actually
wrong.
We
can
let
people
use
that
when
they
want
to
submit
a
merge
request
to
fix
a
cross
join,
they
can
add
their
prevent,
prevent
cross-joints
to
their
test
and
prove
that
that
thing,
no
longer
cross-joins
and
they'll
be
able
to
see
a
failure
end
up
in
the
green
test.
So
I
think
it
is
good.
C
D
C
C
It's
tricky
as
well.
For
example,
a
factory
for
the
runners
executes
some
policy
tracks
on
the
creation
that
okay
exhaust
has
manufactured
on
the
object
creation,
so
yeah,
it's
not
so
straightforward,
I
think,
but
like
we
are
one
step
closer
with
that
and
for
for
some
cases
it's
exactly
gonna
be
as
you're
describing
that
we
could
actually
use
that.
C
D
On
the
specific
circumstances,
but
we
might
execute
joints,
you
don't
always
need
data
like
a
lot
of
the
joins.
Don't
require
you
to
actually
use
the
factories
to
generate
all
of
the
stuff
that
would
be
returned
before
you
can
see
a
joint
query
that
fails
imagining
like
vulnerabilities
and
other
things
like
that.
You
can
maybe
get
a
minimal
amount
of
data
without
creating
those
joints
without
the
factory
stuff.
D
A
Yes,
yeah,
I
just,
I
guess,
happy
to
go
to
workshop
quick
workshop
too.
If
you
need
to
reset
your
database,
but
the
short
answer
is
dropbox
yeah
database.
C
Yes,
I
I
was
thinking
the
people
who
would
have
these
structuresql
marriage.
I
would
then
really
the
next
step
after
that
would
be
try
to
enable
some
kind
of
testing
of
like
two
databases
on
the
on
the
ci
to
run
the
town
on
site
in
some
circumstances,
because
this
would
be,
I
think,
a
natural
step
to
start
exhaust,
like
instance,
variable
because
I
noticed
that
we
merged
the
application
record,
and
we
also
have
this
intermediate
ci
database,
something
stuff.
C
A
I
think
that's
one
more
thing
we
need
to
fix,
which
is
or
disabled,
which
is
partitions.
It's
not
working.
C
Yeah,
the
partitions
are
broken,
but
fortunately
ci
database
doesn't
use
partitioning
today.
C
D
C
C
So
so
this
will
be.
I
I
think
this
can
be
like
a
variety,
because,
like
one
thing
that
exactly
what
is
he
was
saying
like,
I
noticed
like
some
comments
from
some
verified
folks,
how
I
can
validate
that.
So
this
could
be
pretty
handy
for
giving
people
away
from
50
dates.
D
It
would
be
good
to.
I
mean
I'm
happy
to
take
that
action
too.
If
you
don't
get
to
that
I'll
do
that
tomorrow,
because
I
can
add
that
tool
to
the
docs,
because
we
have
docs
about
removing
joins,
and
now
we
can
extend
those
dots
with
how
to
validate
that
you've
actually
removed
the
join
by
using
this
test
technique.
C
D
C
A
C
Yes,
it's
exactly
what
we
talk
on
the
monday.
It's
another
way
to
be
disabled
joints,
but
with
plucking,
but
it
already.
It
also
has
like
the
same.
Actually
it's
pretty
worse
than
black
than
disabled,
because
disabled
joints
always
have
some
kind
of
some
sort
of
very
neat.
But
here
we
are
plugging
in
like
in
these
kind
of
places,
and
this
is
what
I
also
like
writes
in
the
comments
that
like
we
need
to
understand
the
quantity
of
the
data
sets
and
somehow
figure
out
a
way
to
safely
roll
out
it.
C
C
Is
really
like
this
way
of
solving
will
be
a
pattern
for
a
lot
of
these
changes.
If
it
is
all
of
these
patterns
will
have
the
same
implications
about
the
performance.
If
it
is,
we
need
to
kind
of.
I
don't
know
these
people
on
how
to
properly
measure
that
and
how
to
propel
you
all
of
that,
because,
like
the
trend
in
this
form,
I
think
it's
unsafe
to
be
merged
because,
like
we
don't
know
the
quantities
of
this
data
set,
we
don't
know
how
many
answers
you're
gonna
look.
C
We
don't
check
any
sqls,
and
things
like
that,
so
I
I
think
like
if
you
are
doing
this
kind
of
changes
music
to
be
more
pro.
So
then
my
question
is
like
how
we
could
kind
of
create
a
template
or
like
provide
the
guidance,
maybe
do
some
q
and
a
session
with
people
yeah
on
your
work?
How
to
prepare
you
like
sort
these
things
to
ensure
that
we
don't
close
the
farm
to
the
performance.
D
Yeah,
I
agree
this
one
is
looking
like
it's
going
to
be
in
the
category
of
the
easy
option
of
pluck
is
not
a
good
idea.
I
I
don't
know
it
doesn't
look
right
to
me,
though,
I'm
kind
of
confused
because
you're
building
a
list
of
runner,
ids
and
you're
plucking
those
into
memory
and
then
you're
using
them
in
a
where
query
for
a
runner,
but
that's
the
same
table.
C
Yes,
like
like
there
is
cross
line
between
runner
and
space
and
groups.
Here.
Definitely
the
product
id
is
solved,
but
anyway,
like
okay,
the
crosstalk
validation
will
actually
discover
that
system
properly
solved.
But
I
think
this
is
so
it's
like
the
bigger
the
bigger
problem,
which
is
like
the
plug.
D
D
D
C
So
it's
still
like
you,
don't
know
what
is
on
path
of
the
group
id
parameter.
D
C
Anyway,
but
I
think
it
kind
of
shows
like
we
need
to
provide
people
a
way
to
ensure
that
it's
properly
tested.
Yes,
yes,
because
they
will
simply
get
the
solution
wrong.
A
Switch
back
and
forth
between
schema
and
schematics
right.
The
problem
is
that
not
every
gitlab
instance
can
switch
to
using
scheme
schema.
That's
like
search
files
to
the
new
search
part
immediately,
so
the
street.
Strictly
speaking,
we
have
to
wait
for
every
single
gilap
instance
to
auto
roll
or
set
schema
search
parking
down
the
other
way
before
we
can
even
have
these
schemas
in
the
db
structure.
A
C
A
C
I'm
kind
of
thinking
that
the
like
kind
of
tipping
point
in
the
discussion
was
that,
like
we,
we
can
still
support
like
two
forms
of
the
conflicts,
because
I
I
think
they
were
simply
worried
that
like
if
they
switched,
we
they're
gonna
break,
but
we
still
like,
I
think,
for
one,
maybe
at
most
two
releases
can
support
like
two
ways
of
config.
Yes,
yes,
it
is
compatible,
which
is,
which
is,
I
think,
like
the
the
main,
the
most
important
aspect
that
like
it's,
that
application
simply
works.
We
like
to
change.
C
Anything
like
you
need
some,
I
don't
know,
unblock
block
or
like
some
additional
info
on
anything
that
would
unblock
you.
B
My
main
blocker
at
the
moment
is
the
sr
team
because,
like
while
I
was
provisioning
two
weeks
ago
or
one
week
ago,
I
cannot
do
it
now
because
they
broke
something
fixing
something
else,
because,
unfortunately,
our
provisioning
system
here
doesn't
have
any
testing
levels.
So
it's
a
bit
of
black
box
some,
let's
break
things
after
this.
I
think
I
will
have
the
emitter
running
soon
and
I
think
we
can
start
to
do
some
tests
or
plan
how
we
are
breaking
or
failover
strategy
and
see
more.
Why
we
need
to
tune
things
sounds
good.
D
Yes,
I
don't
think
I
know
anything.
No,
I
think
what
you're
doing
is
good.
It
sounds
like
the
the
biggest
kind
of
concern
we
have
upcoming,
which
you're
aware
of
already
is
just
figuring
out
how
structure
and
migrations
are
going
to
be
handled.
I
mean
you,
I
think
your
current
tactic
may
be
wait
and
see
a
little
bit,
but
I
think
those
are
the
biggest
challenges
and
unknowns
for
the
whole
team
that
may
factor
into
certain
decisions.
We're
making
but
they're
also
the
hard
questions
that
we've
gone
back
and
forth
and
many
times
well.
B
What
I
see
here
is
like
we
will
have
all
the
trafficking
on
clusters.
We
have
no
others
how
we
will
do
our
strategy.
We
will
split
the
traffic
in
between
two
pg
bounces,
as
we
spoke
briefly
normal
in
ci
traffic.
We
can,
I
think,
like
the
main
points,
how
we
will
stop
the
traffic
we
can
block
by
the
what
goes
to
one
pg
bouncer
and
leave
the
other
stuff.
D
I
was
referring
to
something
camille
asked
me
a
second
ago,
and
I
wasn't
responding
to
your
to
your
comments,
but
but
anyway
that
the
yeah,
I
I
think
the
power
block
traffic
thing
is
probably
the
next
thing
we
are
yeah.
It
is
going
to
be
relevant
to
how
we
run
the
migration
too,
which
is
what
you're
talking
about,
and
I'm
curious
to
know
if
we
end
up
making
decisions
on
that
too.
But
that
wasn't
what
I
was
specifically
referring
to
a
second
ago.
That's
right.
C
Yes,
I
I
think
that,
on
you
are
correct,
like
we
have
this
structure
migration
stuff
like
to
figure
out,
but
I
think
we're
gonna
be
pretty
much
fine.
We
are
like,
then
we
know
10
more
things
than
we
know
like
in
the
past
discussions
about
how
to
deal
with
this
stuff.
So,
but
it's
still
funny.
C
Have
a
good
day
everyone,
so
I
say
thank
you
and
good
luck.
Keep
swinging
towards
the
shore
on
this
hard,
complex
project.