►
From YouTube: 2021 07 26 APAC Sharding Group Sync
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
B
C
I'm
kind
of
thinking
like
about
like
some
some
way
for
us
like
to
to.
I
don't
know
certain
styles
of
like
what
we
are
doing
or
or
something
I
don't
know.
If
this
is
like
strongly
necessary,
would
it
be
helpful
or
not
of
like
describing
the
just
like
in
one?
Second,
that's
what
I
did
and
describing
in
the
next
sentence.
What
is
my
plan
for
the
next
week.
A
A
C
I
don't
know
if
table
maybe
just
like
each
week,
we
start
like
with
just
everyone
describing
what
they
finished
and
like
what
you
plan
to
do
this
week,
like
one
sentence
at
most
so
kind
of
have
like
the
the
overall
like
perspective
from
on,
like
who,
on
what
everyone
is
working
on
and
like
what
is
like
the
goal,
because,
like
I'm
kind
of
like
thinking
about
from
that
perspective,
that,
like
that
we,
we
don't
really
overlap
with
what
we
are
doing
yet,
because
we
still
have
a
lot
of
plenty
of
other
stuff.
A
My
understanding
is,
we
have
like
so
many
places
where
we're
doing
that
already
and
introducing
another
one
like
I
was
saying
to
craig.
If
we
create
an
issue
and
we
want
to
work
on
it,
we
have
to
assign
a
milestone
to
it,
but
then
we
also
have
to
assign
the
sharding
active
label
to
it.
Otherwise
it
doesn't
appear
on
our
board.
A
We
need
to
assign
it
to
ourselves,
then
we
need
to
assign
it
to
an
epic,
and
then
we
need
to
update
the
epic
with
a
due
date
and
we
need
to
there's
a
single
source
of
truth
table
written
in
one
of
the
epics
that
we
need
to
update
with
the
latest
on
what
we're
doing.
We
get
asked
by
geekbot
every
day,
what
we're
working
on
there's
much
more
than
that
that
it's
not
actually
coming
to
the
top
of
my
head
right
now
but
like
we
just
have
so
much
status,
update
stuff
to
do.
Okay,.
B
B
A
Okay
next,
then,
my
goal
is
working
on
performance
metrics
stuff
for
this
week,
which
is
extra
fields
in
our
logs
that
explain,
digby,
count
and
duration
broken
down
by
database
and
additionally
prometheus.
A
There
are
the
discussion
around
denormalizing
project
id
for
a
couple
of
fields
and
the
security,
the
denominating
project
id
on
the
security
scans
table,
which
actually
solves
like
quite
a
few
of
the
queries.
I've
spoken
to
shubach
and
he's
actually
working
on
this.
Now
that's
his
top
priority
and
then
the
second
thing
is
denormalizing.
The
pipeline
id
on
that
same
thing,
which
solves
several
other
queries,
and
then
there
is
a
separate
discussion
going
on
with
the
secure
and
the
verify
team
about
having
a
denormalized
table.
A
That
says
which
projects
have
used,
which
ci
features
in
the
past
for
doing
joins,
because
both
of
those
teams
have
a
feature
where
they're
like.
I
want
to
go
to
a
dashboard,
and
I
want
you
to
show
me
only
the
list
of
projects
that
have
used
code
coverage
or
only
the
list
of
projects
that
have
security
reports
on
them,
and
that
is
a
tricky
query
to
write
and
it's
actually
super
inefficient.
A
The
query
they
have
today
because
it
joins
to
a
massive
table
to
see
if
the
product
exists
in
there
through
a
massive
table
and
yeah,
we
could
actually
denormalize
this
concept
of
store.
Whether
or
not
a
project
has
used
a
feature,
just
one
time
write
that
to
a
table,
and
now
it's.
C
A
Exactly
yeah,
that's
the
query
and
it's
and
it's
such
a
weird
query
to
write
like
I
want
to
see
the
list
of
projects
that
have
code
coverage
and,
like
you,
join
through
ci
tables
and
then
artifacts,
which
is
two
huge
tables
to
just
to
see.
If
the
project
exists
in
there
and
once
it
exists,
that's
permanently
cacheable
detail,
so
we
may
as
well
just
write
it
to
a
table.
A
Yeah,
that's
that's
those
that
one
is
not
started
yet
by
either
verify
or
secure.
As
far
as
I
can
see,
then
there
was
another
new
one
that
you
created
last
week
that
I
just
got
to
in
triage
today
described
the
problem
and
the
solution.
A
There
is
there
is,
there
was
about
like
12,
different
queries
or
something
that
you
and
tom
have
identified.
Some
of
it's
just
been
removing
code,
but
there
is
now
about
maybe
four
classes
of
problem
that
are
quite
distinct.
Okay,
one
is
simply
denormalizing
fields,
so
we
just
join
through
less
things.
The
other
one
is
the
usage
one
which
I
just
described,
and
then
the
secure
one
that
tong
found
today
was
a
kind
of.
A
B
I'm
playing
to
get
the
poc
amount
to
green
almost
there.
I
think
there's
one
last
class
of
problems
which
is
hanging.
I
think,
there's
a
trade
that
locked
somewhere
a
few
feature
specs.
So
hopefully
that's
all
the
same
issues.
If
not,
then
we
have
six
different
features
that,
after
that,
I
think,
there's
about
43
solutions
that
don't
have
open
issues.
So
I
don't
know
what
I'll
do
maybe
open
an
issue
to
open
issues
and
then
I'll
look
into
shipping
schemas.
C
There
is
like
the
this
problem
of
the
two-phase
commit,
and
I'm
kind
of
like
wondering:
if
should
I
invest
time
into
this
now
or
wait
for
adam
to
go
back
because
he
started
working
on
that?
This
is
my
decision
point
and
second,
I
actually
need
to.
As
I
mentioned,
I
need
to
think
about
like
this
fundamental
stuff
of
handling
many
databases
and
like
how
we're
gonna
want
to
solve
stickiness,
because
this
also
impacts.
C
C
B
B
Yeah
or
like
migration,
so
I
reverted
that.
B
Okay,
so
it
doesn't
make
me
issue,
but
basically
we
need
to
ban
visa
because
there's
a
database
thing
called
reset
all
and
it
resets
all
parameters,
so
postgres
parameters-
I
don't
know
what
you
call
them
to
default
value,
so
the
search
part
went
back
to
user
public,
but
rails
doesn't
know
about
it.
Rails,
actually,
caches
schema
search
parts,
so
it
doesn't
know
about
it.
So
that's
why
it
happened
and
it
and
probably
only
happens
on
db
minor.
B
C
B
B
A
I
had
one
thing
I
ran
into
today
that
I
was
thinking
good
discussion
point
about
disabled
joins
that
a
lot
of
the
disabled
joins
belong
are
part
of,
has
many
queries
that
might
be
unused.
A
C
I
I
think,
like
my
opinion
on
that,
is
like
we
actually
have
two
patterns
in
the
code.
One
is
using
has
manual
second
pattern.
Is
people
use
some,
sometimes
a
word
clause
and
they,
if
the
implement
additional
methods?
For
that
I
at
least
my
preference
is
like
using
what
race
offers
so
like
has
many
has
one,
but
sometimes
if
you
look
for,
for
example,
some
some
queries
like
for
the
runners,
but
not
only
elsewhere,
people
just
implement
that
as
a
class
methods
doing
some
very
complex
queries.
C
So
we
don't
maybe
use
this
manual
as
magnets,
but
we
also,
we
are
not
very
strict
about
the
access
patterns,
how
you
need
to
access
these
cross
relations,
because
sometimes
we
actually
manually
write
queries
where
we
put
all
the
filters
into
vehicles.
C
C
So,
I
think
about
the
relations-
and
this
is
another
problem
that
we
need
to
solve
like
the
foreign
foreign
keys,
because
replacement
lot
of
these
has
managers
not
with
the
disabled
joint.
We
require
dependent
configuration
on
the
relation,
so
I
I
mean
like
at
least
from
my
perspective
like
it
would
be
better
to
actually
have
the
reflection
on
on
the
relationships
that
you
have
defined
in
the
model.
C
But
I
I
don't
know
I
actually
need
to
kind
of
scan
the
foreign
keys
and
see
if
the
removed
foreign
keys
has
associated
relationship
on
the
model
and
whether
it's
actually
properly
configured
because,
like
I
think,
each
foreign
key
that
we
would
remove,
would
have
to
have
these
the
three
properly
configured
of
the
relationships
right
on
the
right
to
actually
remove
objects.
C
A
It
gives
us
guarantees
that
all
of
the
project
id
columns
have
integrity
in
that
they
refer
to
a
project
that
still
exists
for
the
most
part.
Assuming
we
keep
those
two
tables
in
sync,
but
it
doesn't
solve
the
other
foreign
keys.
There's
a
lot
of
things
that
foreign
key
reference
pipelines
and
ci
builds
particularly
from
secure
stuff.
C
Yes,
correct
it,
it
doesn't
solve,
like
all
others,
but
actually,
like
I
kind
of
think
that
you
at
least
have
like
the
clear
orphans
being
covered.
But
I
think,
like
one
of
the
things
that
I
kind
of
learned
when
discussing
different
people
is
like
that
at
the
current
scale
for
the
very
big
project,
the
foreign
keys
actually
are
becoming
a
problem
from
actually
executing
them
as
a
delete
statement,
because
it
actually
was
very
wide
system
log
to
to
like
to
delete
data.
So
maybe
even
trying
to
do
it.
C
This
way
it
may
be
like
the
wrong.
Maybe
we
should
kind
of
really
try
to
rewrite
our
the
reason
to
be
more
sequential
that
you,
our
first
progress
relations
to
remove
data
outside
of
the
transaction,
and
then
you
kind
of
actually
remove
like
the
the
project
to
actually
have
better
control
about
what
is
deleted.
C
Sorry,
I
I
have
a
lot
of
random
thoughts
about
the
foreign
keys.
I
I
need
to
write
them
down.
Yes,
probably
because,
because,
like
I
I
think
like
in
any
in
any
case,
we
need
to
kind
of
figure
out
some
kind
of
I
I
don't
know
topological
tree
of
deletion,
to
figure
out
that
we
actually
cover
all
bases
when
we
delete
data,
because
in
general,
if
we
use
dependent
destroy
so
if
we
use
dependent,
we
we
really
want
to
use,
delete
all,
but
it
has
all
downsides
of
not
calling
callbacks.
A
Yeah
orphan
data
seems
to
be
like
for
a
lot
of
categories.
Orphan
data
is
just
extra
wasted.
Storage.
There
is
weird
categories
where
you
can
find
things
by
id
without
including
a
project
in
the
path,
but
they
are
pretty
rare
and
permissions,
and
things
are
basically
going
to
make
it
impossible
to
ever
get
any
of
this
data
out
of
the
database.
C
A
C
A
Yeah
that
kind
of
makes
me
think
like
we
should
not
really
be
encouraging
on
dependent
delete
ever
but
yeah.
B
A
C
Yes,
actually,
like
I
used
to
do
many
things,
but
I'm
kind
of
thinking
that
we
need
to
think
about
that,
because
we're
gonna
have
like
adam
paris
like
working
with
us,
so
they
could
be
like
they
cover
some
of
these
things
later.
C
If
we
figure
out
how
we
want
to
approach
that
so,
at
least
in
my
head,
we
are
still
in
this
mode
of
like
creating
issues
and
charging
them
like
in
some
kind
of,
like
I
don't
know
like
the
direction,
but
we
don't
know
yet
exactly
how
to
implement
that,
and
only
then
I
can
start
like
charging
these
issues
when
we
start
working
on
them.
C
So
too
many
things
too
many
problems
to
solve,
but
I'm
I'm
actually
super
happy
that,
like
we
got
this
messaging
from
the
eric
that
we
expect
to
engage
other
teams
and
we
need
like
their
very
quick
assistance,
because
it
kind
of,
I
think
it
still
like
it's
some
of
our
capacity
on
doing
stuff
because,
like
you,
for
example,
need
to
help
the
secure
and
we're
gonna
need
to
help.
But
we
actually
have
more
time
to
think
about
this
fundamental
stuff.
C
Instead
of
fixing
these
issues
that
are
usually
very
domain
specific
and
it's
really
hard
to
understand
like
what
is
the
best
way
to
solve
them.
So
I
think
this
is
pretty
good
outcome
of
the
last
week
that
we
have
like
this
clear,
like
expectation
that
others
will
actually
do
that,
and
we
can
help
them
out.
But
we
can
continue
focusing
on
discussing
foreign
keys.