►
From YouTube: 2021 03 09 Database Team Weekly
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
It
is
march
9th-
and
this
is
the
database
team
weekly
meeting.
I
know
they
just
opened
up
the
issues.
It
looks
like
the
only
new
one.
Is
this
first
one
here,
the
sql
queries
running
in
kubernetes
falling
back
to
primary
no
secondaries
available.
A
A
Outside
of
what's
going
on
today,
is
there
anything
that's
top
of
mind
for
you,
jose
that
we
should
be
looking
at.
A
B
A
A
Actually,
I
only
did
the
last
eight
reports
to
see
which
queries
and
I
figured.
I
could
look
at
the
actual
issues
to
see
which
queries
these
ids
relate
to
figure
out
how
frequently
they're
happening
by
total
count,
appearances,
which
is
one
measurement
and
total
time
appearances,
which
is
the
other.
Let
me
resort
to
make
sure
that
we
are
paying
attention
to
the
ones
that
happen
most
frequently.
So
there
were
eight
reports
that
I
collated
here,
so
this
query,
id
happens
has
been
in
every
single
report
that
was
counted
here.
A
So
we
should
probably
make
sure
that
we
have
an
issue
for
investigation
for
this
query
id
and
make
sure
that
if
we
can't
fix
it,
there's
a
team
assigned.
So
that's
how
far
I've
gotten
so
far.
I
was
going
to
pull
in
the
next
couple
reports
and
make
sure
that
we
have.
We
used
that
query
template
that
we
created
to
make
sure
we
have
investigations
on
these
issues.
A
I
I
got
into
a
rhythm
eventually,
where
I
could
figure
out
where
the
query
was,
and
just
repeat
it
over
and
over
and
not
to
pick
on
anyone
but
we're
those
reports
are
inconsistent.
They
all
don't
always
pull
in
the
same
window
and
they're
not
always
consistently
on
the
same
date,
but
we're
seeing
some
interesting
data
there,
and
I
think
one
or
two
of
the
queries
that
are
called
out
in
those
tables
already
have
issues
for
them.
A
But
there
are
quite
a
few
that
still
do
not
and
based
on
our
current
workload
and
priorities.
I
would
prefer
to
identify
them
and
I'll
need
help
from
you
all
to
read
the
marginalia
comments
to
figure
out
where
the
originating
endpoint
is,
if
it's
not
obvious
to
me,
so
that
we
can
then
send
it
to
a
team
and
say
hey
you're,
causing
some
traffic
problems
here.
Can
you
prioritize
this
and
fabian
and
giannis
can
lend
some
product
management
pressure
too
to
say
hey.
A
You
should
cut
out
some
time
in
the
next
milestone
or
two
to
address
this
issue,
so
I
will
I'll
ping
you
all
on
issues
once
they're
actually
created
and
we
we
have
at
least
a
template
issue
around
the
query
and
then
we
can
hopefully
find
the
owner
for
them.
C
There's
one
thing:
I
wanted
to
point
out
with
respect
to
those
seeing
those
query:
ids,
a
good
part
of
that
information
is
sort
of
transient
and
we're,
basically,
I
think,
we're
we're.
Currently
we
we
do
want
to
reset
pg,
start
statements
on
production
and
that's
when
we
lose
all
that
information.
It
does
come
back.
So
the
query
id
is
basically
a
fingerprint
of
the
normalized
query,
so
you
can
expect
that
if
we
still
have
those
queries,
you
can
expect
to
see
them
in
pg
set
statements
after
a
reset.
C
After
a
while,
so
you're
still
able
to
translate
from
query
id
to
query,
even
though
we
reset,
but
we
basically
lose
the
rest
of
the
statistics,
so
it
does
make
sense
to
create
those
issues
early
enough.
Let's
say
when
we
make
those
observations
and
capture
those
details,
also
capture
those
statistics
on
the
issue,
because
you
know
the
some
of
these
statistics
they
go
away
after
a
while
right.
B
D
Yeah,
no,
I
mean
it
so
if
they're
not
anymore
there,
so
that
means
that
whatever
we
have
done
for
a
milestone
or
two
yeah.
B
A
All
right,
we've
already
talked
about
number
two,
a
bit
number
three
andreas
premier,
key
migration.
C
C
C
Mostly
mostly
a
question
about
or
to
to
your
pet,
how
can
we
help
us
with
with
the
topic.
F
The
third
one
is
adding,
then
updating
the
migration
helpers
that
existed
for
the
backfill
and
then
initializing
the
conversion,
and
so
I'm
not
sure
if
we
can
easily
make
that
a
parallel
process,
because
the
really
the
migration
is
the
final
one
that
needs
to
go
and
all
the
other
changes
need
to
go
in.
First,
basically,.
F
F
Not
using
the
helpers
and
just
you
know,
write
the
code
to
do
it
sort
of
well,
then
that
wouldn't
work,
because
we
are
calling
that
from
within
the
other
migration
helper.
So
sorry
that
was
probably
confusing
yeah.
I
don't
know
that,
there's
anything
that
we
can
do
other
than
just
try
to
get
these
merged,
at
least
for
these
next
couple.
F
C
Yeah
and
we
don't
have
to
do
things
in
parallel-
I
was
just
wondering
like
what
is:
what
is
the
best
approach
to
help
with
that?
We
can
also
do
it
in
a
way
that
you're
keeping
up
those,
mrs
and
I'll,
help
getting
them
merged.
With
the
background
containers
like
we
did
yesterday,
I
don't
know
if
that
was
too
helpful,
but
it
did
get
merged
today.
So.
C
F
C
Go
ahead,
maybe
we
can
try
getting
them
in
one
by
one
really
focusing
on
on
getting
getting
short
review
cycles,
because
that
that
really
can't
take
a
long
time
to
wait
for
the
responses.
F
F
Makes
sense
and
yeah
plus
you
know
the
time
zone
factor
too,
if
you
have
an
opportunity
to
like
respond
in
the
morning
and
then
monday's
already
over
whatever,
and
you
can
get
it
so
we
cannot
miss
a
day
on
the
recycle
or
whatever.
That
would
be
helpful
too
just
so
we
can
try
to
get
them
like.
If
we
have
to,
you
know,
send
them
back
for
re-review.
We
can
get
it
done
the
next
day
or
whatever
as
soon
as
possible.
C
F
Yeah,
I
kind
of
thought
we
could
send
at
least
through
the
first
round,
the
back
end
review
at
the
same
time,
kind
of
and
because
probably
there
won't
be
too
many
large
changes
that
would
impact
it.
F
No,
I'm
going
to
add
that
from
the
oh
and
then
the
second
one
that
adds
the
migration
where
that
kicks
off
the
migration
I'll
add
the
crunch
up.
There
probably.
C
To
the
to
the
same,
mr,
that's
the
migration
right
that
makes
sense.
Okay,.
F
I
did
do
the
mr
for
that,
but
I
guess,
assuming
that
we're
just
going
to
run
this
and
then
maybe
like
delete
the
records
from
the
database
before
we
roll
it
out
to
the
wider
for
a
wider
release,
then
I
don't
think
we
need
to
have
that
merged
right
now,
if
we're
just
kind
of
going
to
try
to
do
this
under
a
feature
flag
on
gitlab.com
and
then
make
sure
it
works
and
we're
planning
on
removing
that
or
fixing
that
somehow
manually,
we
can
always
just
push
that
one
down
the
road
until
after
we've
merged
these
others.
F
C
No,
that
makes
sense.
Oh,
that
was
that
was
the
one
I
wasn't
sure
about
it,
that
if
it
did
make
sense
to
put
that
into
a
follow-up,
be
sure
not
it's
a
small
change,
but
yesterday
I
was
a
bit
radical,
like
trying
to
delay
as
much
as
possible
or
to
to
you
know,
create
follow-ups
as
much
as
possible.
D
It
was
just
for
for.
D
In
in
the
agenda
so
yeah
the
web
clocks
partitioning
has
been
merged,
so
there
is
an
epic
that
would
allow
us
to
follow
all
the
steps
and
the
first,
mr
for
this
margin
has
been
merged.
It
will
be
most
probably
deployed
tomorrow,
so
we
will
be
able
to
follow,
and
I
think
that
everything
will
go
according
to
plan.
D
We
will
be
done
at
14.01
and
then
the
only
thing
remaining
will
be
to
replace
the
existing
prone
worker
with
a
worker
that
will
prune
partitions,
but
we
will
work
with
the
future
team
on
that.
E
Yeah
super.
I
have
two
comments.
First,
I
think
I've
said
it
to
janice
already
and
pointed
it
out.
I
think
this
is
actually
a
really
great
example
of
an
epic
and
how
it
is
structured.
So
I
really
like
that.
I,
like
the
description,
it's
very
easy
to
understand,
so
I'm
I'm
just
pointing
that
out.
I
think
it's
really
awesome.
I
wish
everything
gitlab
had
the
same
level
of
of
detail
when
you
look
at
it.
So
thanks
for
doing
that,
janice
and
the
other
thing
is.
E
D
Okay,
yeah,
the
only
problem
is
that
we
run
the
the
testing
pipeline
yesterday
and
due
to
the
lock
today,
we
cannot
deploy,
so
it
will
be
off
by
a
couple
of
jobs,
whatever
the
new
work
hooks
will
be
for
two
days,
but
I'm
willing
to
bet
that
in
two
fridays
from
now
march
26th
in
the
morning,
everything
will
be
done.
D
I
have
also
wrote
it
in
the
issue,
the
related
issue,
so
so
there
is
record
there.
D
B
A
How
much
influence
can
giannis
actually
exert
on
the
completion
date?
Because
I
think,
there's
a
fix
in
here
somewhere
yeah.
D
I
think
that
I
have
the
best
relationships
with
the
distribution
team,
so.
B
A
All
right,
yeah,
thanks
for
creating
that
epic,
that
is
very
clean
and
clear.
I
appreciate.
A
C
It's
less
than
it
looks
like
it's
mostly
about
the
testing
pipeline
and
a
couple
of
fixes
and
smaller
changes.
So
one
problem
that
we
had-
I
don't
know
if
you
if
you
thought
I
was
running
or
basically
have
those
two
runners
in
their
instances
and
they're,
basically
building
a
lot
of
poker
images
and
they're
constantly
running
out
of
space.
C
So
I
had
to
go
in
manually
and
free
that
space
up
again,
and
I
hope
I
have
a
method
now
to
do
it
for
the
registry
of
the
local
register
that
we
run
on
that
instance.
So
that
should
be
should
be
fixed
hopefully,
and
the
rest
is
really
small
updating
feedback.
C
C
Basically,
then
audit
events-
partitioning
that
was
mostly
summary
for
what
pat
already
had
and
we're
now
dropping
the
table,
or
there
is
a
there-
is
an
invite
demar
for
dropping
the
table.
E
C
This
is
waiting
for
feedback.
Basically,
so
I
think
people
are
using
it
or
picking
it
up,
but
haven't
given
feedback.
Yet
I
would
like
to
get
a
bit
more
feedback
before
we.
You.
B
A
All
right,
pat
your
up
migrations
for
primary
key
conversions.
F
C
Yeah,
that
is
probably
once
we
have
those
mrs
lined
up
and
working,
then
we
should
be
able
to
run
them
on
the
testing
pattern.
If
that
still
makes
sense,
it
should
be
relatively
easy
to
do
that.
We
already
have
it
sort
of
halfway.
C
So
maybe
once
we
have
those
mrs
working
now
we
can,
we
can
update
the
testing
pop-up
once
again
to
reflect
that
and
you
can
simulate
a
couple
of
runs
if
we
like
to,
but
I
would
wait
for
for
those
mrs
to
be
become
ready,
basically,
so
that
we
can
we
can
test
them.
Does
that
make
sense.
F
B
E
B
F
Right
this
is
the
one
we
also
just
discussed.
That
was
a
follow-up
actually
thinking
about
it,
a
little
bit
more.
I
think.
Maybe
we
probably
do
need
to
merge
this
now,
because
otherwise,
we'll
have
a.
We
might
have
an
issue
merging
it
because
the
data
in
the
database
and
then
what
if
we
were
trying
to
deploy
while
the
migration
is
running,
then
things
won't
we'll
have
basically
like
a
deployment
issue,
because
if
we
update
the
data
in
the
database,
then
some
service
will
be
running
some
code.
F
Some
will
be
running
the
other,
so
I
think
we
probably
should
just
try
to
do
it
now.
It'll
be
easier,
I
think,
than
trying
to
fix
it
later
and
it's
a
small,
mr.
Hopefully
it
won't
take
very
long.
F
C
F
D
This
affects
people
that
go
on
those
in
between
minor
versions
where
we
added
new
schemas
and
some
new
types
of
objects
like
views
and
functions,
and
there
are
some
edge
cases
where
people
try
to
downgrade
and
upgrade
again
and
rerun
migrations,
and
some
things
are
duplicate
or
missing
so
because
we
decided
not
to
to
go
back
back
and
pass
10
different,
10
minor
versions,
we're
going
to
add
a
troubleshooting
section
for
everyone.
D
A
A
C
A
That's
a
good
question:
we
are
going
to
have
the
pg-12
meeting
this
thursday.
It
was
supposed
to
be
last
thursday,
but
I
would
just
maybe
throw
it
on
the
agenda
for
next
time.
D
We
will
want
this
there
as
a
bonus
it's
not
like,
but
it
would
be
nice
to
have
it
deployed
by
the
time
they
run
the
the
test
so
that
they
can
test.
With.
With
that
update,
included.
A
D
Just
one
young
everything
else
is
actually
true
housed
and
we
are
dressed
everything
else.
So
yeah,
okay,
are.
B
Good,
I
will
add
an
agenda
item
for
testing
pg-12
for
thursday.