►
From YouTube: Geo Scheduling Call 2021-08-10
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,
first,
one
again.
B
Got
this
now
to
the
point
where
it's
running
qa
against
instance,
that
it
spins
up
so
it's
coming
along
pretty
good.
It
should
hopefully
have
this
pipeline
cleared
up
today
and
then
I
will
just
need
to
add
the
finish
up.
The
omnibus
mr.
A
A
A
I'm
not
sure
where
exactly
that
is,
but
since
it's
a
doc
thing
and
cat
is
doing
it,
I'm
not
gonna,
maybe
bother
him
too
too
soon.
Unless
anybody
else
knows.
A
Being
done
by
support
here.
C
Yeah,
the
start
is
pretty
much
the
same
as
last
week
is
not
working
on
this
right
now.
There's
some
work
left
over,
but
he's
working
on
an
another
issue
which
is
somehow
not
on
this
board,
probably
because
we're
not
our
active
I'd,
take
this
moment
to
give
you
a
small
update.
C
These
are
now
the
last
three
weeks
of
gsoc
and
he
is
working
on
the
second
issue
here,
so
the
temp
directory
was
not
getting
cleared
up
in
the
restore
task,
which
is
what
he's
working
on
right
now,
and
I
think
this
might
be
the
last
issue.
Maybe
this
and
one
more
issue
he
can
get
through,
so
that
that'll
complete
his
project.
There's
going
to
be
a
demo
call,
and
I
will
thank
you
all
about
it.
A
D
I
don't
remember
any
details.
Sorry.
A
A
Okay,
well,
it's
been
five
days
since
the
last
discussion
so
I'll
leave
that
to
get
to
a
week.
E
Yep
this
is
me
this
is
just
so.
This
is
now
adding
the
hybrid.
So
this
is
all
waiting
on
review
now,
so
all
the
work's
done
and
everything
looks
to
be
working.
It
was
okay
for
the
most
part.
Most
of
our
stuff
for
charts
is
quite
simple.
I've
spoken
about
it.
I
think
one
of
the
the
only
really
annoying
part
was
when
you
change
a
secret
within
the
chart.
E
It
doesn't
redeploy
the
the
nodes,
so
you
have
to
go
and
destroy
them
all,
to
get
them
to
pick
up
the
new
secret
value
and
because
we
don't
automatically
pull
the
secrets
from
the
primary
there's
some
kind
of
awkward
redeploying.
We
had
to
do
to
get
the
correct
secrets,
but
other
than
that
it
was
all
quite
smooth.
So
it's
one
of
the
things
we
might
raise
an
issue
for
just
as
a
potential
thing
we
might
be
able
to
improve
later.
A
Nice:
okay,
thanks
nick.
A
This
was,
hopefully
it
was.
We
were
hope
I
was
hoping
it
was
going
to
be
the
last
part
of
the
implementation,
but
it's
not
quite,
but
it
is
closed
and
with
this
merge
request,
where
it
makes
it,
so
we
don't
make
too
many
api
calls
from
workhorse.
A
And
what
happened
was
when
I
went
to
test
in
staging
there.
There
was
an
old
bit
of
code
in
workhorse
that
just
said:
if
we
set
the
back
end
to
an
https
url,
then
just
die
panic
and
it's
basically
we
don't
know.
We
don't
know
why
exactly
that
was
added,
so
we
removed
it
and
that's
in
review
and
it
seems
to
be
working.
So
that's
that's
where
that
is.
A
And
in
dev
is
the
rollout
issue
for
that
feature,
and
so
we're
waiting
on
this
to
be
deployed.
Then
I
can
test
in
staging.
A
And
after
that,
fabian
suggested
that
we
get
some
customers
to
beta
test
it.
So
this
is
this
feature,
is
kind
of.
It
seems
like
it's
almost
done,
but
there's
a
lot
of
testing
yet
to
be
done,
and
probably
some
improvements
to
come
out
of
that.
A
C
Yeah,
I
can
talk
about
this
so
between
last
week
and
this
week
the
big
change
that's
happened
is
in
this
center
most
issue
resolving
association.
C
So
this
for
those
who
were
not
in
the
threads,
we
had
some
larger
discussions
with
the
database
team
amongst
us
as
well,
and
we
considered
some
options
and
now
we
have
a
solution
which
is
going
to
be
that
we
need
to
backfill
the
verification
data
for
all
the
rows
and
we're
probably
going
to
need
a
worker
for
that,
and
we
will
have
to
change
some
queries
and
eventually
also
optimize
some
of
the
verification
queries
which
were
which
are
a
real
problem
and
all
of
this,
along
with
thinking
about
what
is
going
to
happen
to
these
tables
when
sharing
happens,
because
even
though
it
might
not
happen
in
the
next
three
months,
it's
something
we
have
to
think
about
when
adding
a
new
huge
table.
C
So
now
we
have.
C
We
are
right
now
making
a
plan
in
the
issue
in
the
comments
later
on,
and
what
I'm
trying
to
do
is
start
with
something
smaller
like
lfs
optics,
which
is
an
mr,
which
is
stuck
that
ian
was
working
on
after
I
left
I'm
trying
that
we
start
pushing
out
one
thing
after
another
in
a
way
that
we
are
not
stopped
by
solving
the
most
complex
problem
first,
but
we
also
don't
release
any
of
this
work
until
we
are
very
sure
that
it's
going
to
work
and
we
are
not
introducing
a
regression,
so
I'm
most
likely
going
to
start
with
lfs
objects,
and
the
next
iteration
would
be
mrs
and
pages
deployment
that
nick
was
working
on.
C
The
overall
thing,
I'm
trying
to
say
is
that
now
this,
this
current
project
has
become
even
bigger,
and
it's
going
to
take
at
least
two
or
three
milestones
before
we
can
get
back
to
that
speed
that
we
had
before,
where
we
have
a
issue
template
and
we're
just
following
steps
and
adding
verification
quickly.
But
after
that
things
should
be
smooth
again.
So
this
is
where
we
are.
A
Yep,
thank
you
for
driving
that
yeah.
It's
really
too
bad
that
we
had
to
that.
We
have
to
use
the
summer
table
and
we
have
to
backfill
only
specific
rows,
not
all
of
them
so,
but
I
think
it
will
be
a
nice
solution
in
the
end.
C
Yeah,
but
it's
still
better
to
find
all
of
these
problems
in
one
go
now
than
having
not
so
well
thought
out,
database
solution
and
then
figure
out.
Oh
my
god,
our
queries
are
so
not
performant.
What
do
we
even
do?
Where
do
we
even
start
or
having
these
huge
tables
like
mrdisk,
is
one
of
the
biggest
tables
in
our
database,
and
this
new
verification
table
is
going
to
be
at
least
as
big,
so
we
are
literally
introducing
another
very
big
table.
A
A
Well,
this
is
closed.
Something
has
merged.
I
don't
know
where
that
is
exactly,
but
it
was
the
documentation
thing.
So
I'm
not
too
worried
thanks.
Nick.
A
Gabriel,
create
a
single
command
that
can
be
running.
You
know
to
promote
a
secondary
to
primary.
A
D
Yeah,
so
I
didn't
have
a
chance
to
spend
a
lot
of
time
on
this
issue
last
week,
so
I
am
I
I
am
close
to
finish
it.
I
can't
really
say
anything
more.
A
I
I
just
had
a
thought,
given
that
now
we
have
made
a
decision
on
the
verification
of
things
that
need
a
separate
table.
Well,
I
it's,
I
think,
everything's
going
to
be
using
the
separate
table,
probably
in
the
end
yeah,
that's
where
the
discussion
was
going.
Given
that
that's
happening,
maybe
it
is
worthwhile
to
release
the
migration
of
uploads
to
ssf
without
verification.
A
D
I
think
it's
it
like
it's
not
like
something
is
pulling
me
back
and
I
think
it
will
not
solve
any
problem
really
and
also,
moreover,
it
will
require
even
more
steps
to
do,
because
I
will
need
to
cut
lots
of
code.
I
I
already
added
so
I
think
significant
work
already
done
about
verification
been
added
to
this
particular
request.
So
also
I
have
some
other
troubles
with
name
collisions
and
lots
of
different
stuff.
D
So
maybe,
like
it's
not
already
finished
so
maybe
I
will
face
some
obstacle
that
will
be
hard
to
overcome.
Then
I
think
it
will
make
sense
to
consider
separation,
but
for
now
I
don't
think
so
so,
but
if
you
have
any
like
any
like,
if
you
think
it
will
solve
some
really
hard
problems,
then.
B
A
Yeah,
I
I
don't
yeah
if
I
think
I'm
happy
to
go
with
whatever
is
going
to
be
the
most
efficient
over
the
next
four
months.
That's
totally
fine!
So
if,
if
that
means,
you
know
not
not
necessarily
merging
it
until
that's
done,
because
we're
not
ready
to
merge
anyways,
that's
that's
fine
too.
D
Yeah
yeah,
it
actually
made
sense
when
I
started
because
one
benefit
I
see
is
it
would
be
a
lot
of
like
it
was.
It
would
be
easier
to
review
because
we
now
have
like
60
60
changes
and
it's
like
a
hell
to
review,
but
I
think
I
faced
some
actually
initially.
I
was
considered
to
do
that.
D
But
then
I
faced
something
I
don't
remember
so
in
theory,
it's
like
we
just
need
to
flip
feature
flag,
false
right
and
that's
it
right
and
we
don't
need
to
add
any
extra
code
for
verification
yeah.
I
think
one
problem
is
that.
Do
you
remember
if
you
know
our
guide,
we
have
a
separate
instructions
to
add
verification,
step
and
replication,
because
it
seems
to
me
like
we
have
a
common
instructions,
so
you
would
need
to
separately
check
every
line,
so
it's
like
would
be
harder.
D
A
A
A
But
for
the
most
part,
it's
like
overriding
the
verification
feature,
flag
thing
to
say:
false
hard-coded
and
not
including
a
couple
modules
and
not
including
some.
You
know
shared
examples
into
the
test.
A
I
think
it's
doable
but,
like
I
said
whatever
you
feel,
is
most
efficient
over
the
next
four
months.
So.
D
Yeah
so
in
general
I
think
that
made
sense
early,
but
when
there
is
lots
of
changes
now
and
yeah,
I
think,
unfortunately,
that
would
be
like
extra
extra
fee
or
to
go
back
on.
A
Yeah
yeah,
okay,
okay,
just
a
thought.
C
I
can
talk
about
this
because
I
believe
the
mr.
I
think
this,
mr
is
only
blocked
on
backfield
migration.
So
that's
why
this
is
in
my
plan
to
do
in
the
next
iteration.
Once
we
have
alphys
objects
through
all
right.
I
think
this
can
be
tagged
as
workflow
blocked.
Okay,
okay,.
A
A
I
think
so.
Yeah
so
stan
found
an
issue
with
a
relative
root.
Url
and
zach
zach
just
went
ahead
and
fixed
it
very
quickly.
So
thank
you
to
zach.
That's
closed
now,.
B
B
But
when
we're
looking
for
replication
we're,
not
we
weren't
telling
the
primary
that
we
were,
we
would
accept
a
manifest
list.
So
then,
the
doctor
industry
kind
of
assumes
that
you
just
want
a
regular
old
image
and
it
hands
you
an
image.
Rather
than
gives
you
an
error
that
you,
you
know
you're
asking
for
for
something
that
I
can't
keep.
So
that's
why
it's
always
only
replicating
the
one
container
right
now
so
once
this
is
fixed,
then
should
be
able
to
replicate
multi-arch
and
I'm
not
sure
how
much
we
want
to
think
about
that.