►
From YouTube: 2020 08 11 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
All
right
welcome
to
this
edition
of
database
team
meeting
jump
right
into
it
see
the
container
registry
partitioning
saw
your
notes,
giannis
thanks
for
the
update
there.
It's
a
high
priority
item
for
next
week,
so
I
don't
know.
Were
there
any
hot
topics
we
should
cover
with
container
registry.
B
No,
it
was
a
preliminary
discussion
on
so
I
try
to
understand
that
write
some
notes
so
that
there
are
there
for
you,
patrick
and
andreas.
When
it
comes
back,
you
can
start
drilling
down
on
this
issue
and.
C
B
The
summary
is
more
or
less
what
we
discussed
last
week.
Also
that
there
is,
there
is
no
natural
partitioning
key
from
what
I
can
see
in
my
understanding
is
that
there
are
like
three
chunks
of
different
things
inside
the
registry,
so
you
have
the
repositories
and
the
stuff
around
the
repositories
you
have
manifests,
and
then
you
have
the
blogs,
and
unfortunately
there
seems
that
there
is
no
natural
way
to
combine
everything,
and
the
second
part
was
the
problem
of
the
access
patterns
is
that
there
is
no
not
a
single
type
of
access
pattern.
B
They
there
are
all
those
read
the
most
varied
queries
go
with
using
the
house
either
a
manifest
cast
or
a
configuration
has
blocked
us
digest
and
has-
or
there
are
some
queries,
especially
for
governor's
collection
that
go
usually
like
these,
so
we
have
to
think
about
it
and
check.
So
I
wrote
some
ideas
like
making
the
digest
the
primary
key.
This
is
very
expensive
or
doing
some
denormalization.
B
A
A
All
right,
but.
B
One
important
part
that
we
discussed
is
that
figure
that
we
should
figure
out
the
axis
pattern
and
that
if
we
don't
have,
even
if
we
have
a
partition
key,
if
the
the
partition
key
is
not
used
in
all
the
queries
or
at
least
the
major
queries,
this
is
a
problem.
A
Yeah
for
sure,
okay
yeah,
thanks
for
having
those
conversations
with
john
all
right
yeah.
We
have
the
performance
indicators
meeting
later
on
today
and
we're
not
done,
but
thanks
for
the
update
on
this
issue
be
honest.
B
Yeah,
I'm
sorry,
but
most
probably
we
won't
make
it
so
I
will
try
to
have
it
have
as
many
things
as
possible
ready,
but
my
prediction
is
that
we
won't
make
it
because.
A
Yeah
and
there's
quite
a
few
dependencies
there
right.
B
Yeah
the
the
bad
luck
there
was
that
after
we
initially
discussed
it-
and
I
I
left
it
alone
and
did
everything
else
until
I
came
back
when
I
came
back-
there
was
an
update
that
removed
the
one
of
the
packets
from
the
histogram
that
we
did
so
in
order
to
run
the
his
the
the
metric.
We
need
the
the
packet
for
250
milliseconds,
so
the
buckets
for
100
milliseconds.
B
B
It
is
bring
back
the
the
bucket
and
then
we're
going
to
run
and
now
our
full
hour,
the
next,
with
the
memory
team
that
we're
going
to
discuss
this
issue.
So
there
will
be
a
recording
where
we
will
discuss
this
issue
and
all
the
issues
whether
we
can
release
everything
together,
how
we're
going
to
do
things.
So
probably
there
will
be
a
nice
recording.
A
A
Yeah
memory
team
just
did
similar
work
over
the
last
probably
milestone
and
a
half.
So
there's
a
lot
of
context
there
that
they'll
be
able
to
answer
those
questions.
So
thank
you
so
the
next
one
up
is
the
verify
audits,
partitioning
migration.
It's
not
assigned.
I
know
that
we
validated
that
triggers
worked.
Everything
was
moved
and
I
believe
the
backfill
worked.
Is
there
anything
else
that
needs
to
be
done
on
this
issue?.
A
D
Yeah
yeah,
we
were
just
gonna,
that's
kind
of
what
I
had
been
looking
at.
I
can
assign
it
to
myself.
I
guess
we
have
both
done
some
work
on
it.
So
but
yeah
we
were
gonna.
Try
to
do.
We
had
talked
about
doing
like
a
batch
by
batch
kind
of
comparison
of
the
two
tables
and
go
through
and
make
sure
everything
matches
the
record
counts
and
everything
side
by
side.
B
And
we
are
going
to
use
similar
code
in
our
cleanup
migration
on
the
right
patrick.
So
we
are
going
to.
D
Well,
it'll
be
a
little
different
there
because
we're
just
going
to
use
what
the
the
tracking
table
now.
So
we
don't
need
to
do
the
but
it'll
be
a
similar
concept.
A
D
Yeah
there
was
a
follow-up,
that's
in
the
last
on
the
last
review
right
now,
so
once
that's
merged.
I
think
that
the
comments
there
have
pretty
much
died
down,
so
we
can
close
it
once
the
last
mmr
gets
merged.
Okay,.
D
A
A
Index
helpers
don't
respect
the
where
clause
it
doesn't
look
like
any
work's
been
done
on
this
one
and
considering
monday's
the
end
of
milestone.
I
was
suggesting
we
move
it
to
thirteen
four.
Unless
pat,
are
you
honest?
You
think
you
can
get
this
in
before
monday.
A
Keyset
pagination,
I
think
I
saw
one
of
them.
Thank
you.
A
Close
thank
you
rig
task.
D
A
Okay
yeah
and
this
one
there's
still
some
discussions
going
on
about
pg
repack
and
I
think
once
andreas
gets
back
pat
you
and
he
can
talk
about
it
with
jose
to
figure
out
what's
the
best
path
forward
to
make
sure
we
do
the
automatic
re-indexing
the
right
way
in
the
most
manageable
ways.
I
know
there's
some
concerns
about
monitoring,
either
solution
that
we've
talked
about,
but
it'd
be
nice
to
get
some
kind
of
solution
going
forward.
A
So
we
can
reclaim
that
space
so
yeah
once
we
wrap
up
this
rake
task
and
close
out
that
issue,
even
if
we
take
a
different
route
right
so
and
then
database
guides
under
the
database
directory.
B
Talkingly,
I'm
in
panic
mode
with
the
metrics,
so
everything
I
have
I
will.
I
will
go
to
reviews
and
the.
A
B
B
A
So
a
couple
things
came
up
yesterday
from
the
postgres
checkup
report.
There's
still
some
concerns
about
the
growth
of
the
ci
builds,
and
I
know
we
have
some
related
issues
that
aren't
our
direct
responsibility.
There's
the
hard
delete
issue
and
I
believe
that's
on
the
continuous
integration
team.
I
commented
on
that
one.
The
events
table
primary
keys
at
40
for
the
m4
which
isn't
pressing,
but
it's
an
ongoing
concern.
B
A
B
Maybe
triple
the
number
of
rows,
I
don't
remember,
but
maybe
we
should
start
with
the
access
patterns
and
make
sure
at
least
some
of
the
work
is
done
so
that
we
are
ready
personally,
so
I
think
that
this
will
go
above
the
name
spaces,
for
example,
issue.
So,
even
though
we
have
this
namespace
or
front
namespaces
without
the
parent,
this
was
a
and
all
the
issues.
B
We
moved
it
at
some
point,
but
maybe
it
will
go
if,
if
you
have
me
doing
something-
maybe
it's
better
for
me
to
identify
as
patterns
for
events,
table
etc.
D
A
C
Yes,
great,
I
would
like
just
to
add
that,
for
my
analysis,
I
created
a
new
issue
based
on
some
queries.
I
did
and
the
report
from
nikolai
that
I
think
we
will
run
again
the
pg
repack.
I
believe
we
need
to
discuss
with
andreas
in
the
future.
What
happened
there?
Because
I
still
see
a
lot
of
bloat
and
at
the
moment
the
extension
under
estimate
is
not
is
telling
us
that
there
is
nothing
to
be
done.
A
All
right
and
you're
gonna
catch
up
with
andreas
next
week,
when
he's
back
on
that.
A
One
okay:
we
kind
of
covered
some
of
the
goals
for
this
week,
so
the
the
work
that
giannis
is.
A
A
What
else,
what's
the?
What
are
the
plans
for
this
week?
What
are
we
trying
to
accomplish
other
than
wrapping
up
13-3.
B
One
note
on
the
database
instrumentation
whatever
mrsa
will
have
ready.
I
will
pass
them
to
you,
patrick,
so
I
will
try
to
have,
even
if
they
don't
make
it
for
this
release.
B
A
Okay,
there
any
other
goals
outside
of
what
we
discussed
and
trying
to
wrap
up
issues
for.
A
All
right,
I
will
look
at
the
events,
epic
and
mark
a
couple
issues
for
13
4,
so
that
we
have
that
lined
up
is
probably
just
the
there's
a
couple
of
analysis
issues,
or
at
least
one
in
there
to
identify.
What's
the
best
axis
pattern
and
partitioning
keys
for
events
table,
I
think
it's
pretty
straightforward,
but
also
make
sure
that
we
have
the
primary
key
issue
in
13
4
as
well,
so
that
we
upgrade
the
integer
on
that
one.
So
all
right.
B
B
There
were
no
updates
and
yeah
in
the
issue.