►
From YouTube: 2022-05-03: Object Storage WG
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
I
would
like
to
check
what's
the
status
of
the
object
surge
working
group,
comparing
to
our
exit
criteria
to
figure
out
if
we
are
in
a
situation
where
we
can
start
wrapping
up
and
and
declare
this
effort
done,
and
so
I
was
looking
so
the
extra
criteria,
the
first
one,
was
about
documenting
status,
quo
of
object,
storage
and
classify
its
usage
by
feature
vertical
integration
patterns.
Since
we
know,
there's
there'll
be
drift
between
features.
I
think
this
is
done.
B
I
would
maybe
so
one
thing
that
might
be
worth
doing
and
I'm
not
sure
how
to
best
go
about
this,
but
I
I
agree
like
the
basic.
I
think
we
all
understand
how
it
all
kind
of
ties
together.
Well
enough
now
and
I
think
the
main
points
we
have
written
down
and
jacob
confirmed
with
a
team
member
as
well
that
it
seemed
to
have
helped
so
that
that's
that's
good
feedback.
B
So
it's
good
to
know
it's
kind
of
the
litmus
test
right,
writing
documentation
is
like
putting
it
in
front
of
someone
who
has
never
worked
with
it
and
then
getting
feedback,
but,
like
one
thing
I
did
notice
is
that
when
we
first
started
to
document
the
individuals
like
on
a
per
feature
level,
how
do
they
perform
upgrades?
B
I
I
because
I
I
drove
that
issue.
I
didn't
understand
well
enough,
and
I
wasn't
really
in
a
position
to
review
what
users
were
adding
to
this
table.
That
is
no
part
of
documentation
and
I'm
not
sure
it's
correct,
I
think,
maybe
maybe
to
wrap
this
up
completely.
We
could
go
over
this
table
together.
B
Maybe
even
now
there's
enough
time
in
this
meeting
today
and
do
a
quick
check
if
that
makes
sense
what,
because
we
had
just
like
distributed
this
kind
of
fan
this
out
throughout
the
org
and
then
kind
of
relied
on
what
everyone
contributed
was
correct,
and
I
I
think
I
mentioned
before.
I
think
there
are
some
combinations
in
there.
That
can't
really
be
true
like
direct
upload,
but
sidekick
upload
is
the
uploader
in
the
uploader
column
and
I
don't
think
that's
a
combination
that
can
ever
happen.
A
Enough
yeah,
probably
all
of
us,
now
know
more
about
this
since
we
started,
so
we
can
just
look
through
this.
So,
okay,
let's
go
to
the
to
the
list
of
items
and
then,
during
this
meeting
we
try
to
go
through
the
list
and
see
to
the
table
and
see
if
there's
something
that
doesn't
make
sense
good
cloud.
Thank
you.
Matthias
point
b
outline
a
path
forward
by
designing
a
new
simplified
architecture
for
object.
Storage.
A
Now
here
we
have
a
couple
of
suggestions
as
well
as
what
the
scalability
framework
team
is
doing,
which
is
taking
from
what
we
started,
describing
and
trying
to
build
a
plan
so
yeah.
A
Maybe
here,
is
just
a
matter
of
wrapping
up
things
linking
all
the
proposals
and
things
like
that,
and
I
will
consider
this
done
as
well
as
identify
high
level
step.
We
need
to
take
for
the
architecture
to
be
realized.
That's
kind
of
in
line
with
that
thing,
because
here
I
here
just
put
the
exit
criteria,
but
then
there
was
another
section
in
the
page
which
said
what
we
are
not
going
to
do,
and
there
are
things
like
we
are
not
going
to
do.
A
full
implementation.
A
We
are
not
going
to
micromanage
the
steps
needed
to
give
us
a
full
solution,
so
I
think
that
b
and
c
are
in
line
with
what
we
have
done
as
well
as
d,
because
the
prototype
individual
aspect
of
it
by
exploring
both
new
technology,
such
as
active
storage
or
reworking
existing
code.
I
think
this
is
where
david
did
a
lot
of
work.
We
did
a
couple
of
tests
on
active
storage
and
other
single
authorization
endpoints.
We
learned
a
lot
from
this,
so
I
would
say.
B
This
is,
and
also
there
was
quite
a
bit
of
work
done,
like
I
think,
five,
mrs
or
so
to
go
through
workhorse
as
well
and
yeah
kind
of
go.
It
didn't
add
new
features,
but
like
it
kind
of
cleaned
up
the
oh
cool,
so
that
was
actually
quite
a
bit
of
work
done.
So,
okay,.
B
A
So
maybe
if,
let's
let's
do
something
like
this,
which
is
it's
easier,
everyone
who
did
some
type
of
merch
request
related
to
this,
please
send
it
to
me
so
try
to
just
write,
build
the
collection
of
what
we
have
done
and
try
to
categorize
so
that
we
can
just
write
out
what
what
we
did.
A
Okay
and
then
identify
features
and
configuration
that
should
be
duplicated
to
reduce
the
maintenance
complexity.
So
we
removed
the
pseudonymizer
bucket
and
feature
we
removed
background
upload,
and
there
is
a
point
from
last
week
that
we
didn't
discuss
about
the
direct
upload
toggle,
which
right
now
really
makes
little
sense
to
still
have.
So.
A
That's,
maybe
something
we
can
think
it's
already
on
jakob
rather,
but
let's
try
to
figure
out
if
this
is
another
thing
that
it's
part
of
what
we
can
remove
so
to
me
here
we
are
in
a
kind
of
a
wrapping
up
situation,
so
we
want
to
clean
up
to
just
and
so
yeah.
I'm
really
happy
with
the
status.
B
I
I
I
would
agree,
I
mean
like
I
think
we
got
you
know
like.
Is
there
more?
We
could
have
done.
Maybe
right
I
mean,
I
guess
that's
always
true,
and
I
think
we
got.
I
think
we
I'm
really
glad
that
we
have
something
to
show
for
each
of
these
points.
It's
we
have
like
no
area
really
where
we
like
left
all
stones
unturned.
Basically,
so
we
really
looked
at
everything,
so
that's
good
and
yeah.
B
I
also
agree
we
should
start
to
wrap
up
because
with
15
0
coming
up,
I
think
there's
also
less
availability
yeah
in
working
on
these
orthogonal
kind
of
kind
of
aspects,
yeah.
B
A
B
Yeah
we
had
briefly
touched
on
this
like
a
couple
weeks
ago.
I
think,
okay,
I
don't
know
if
at
this
point
it
even
like
makes
sense
anymore,
but
I
was
thinking
my
experience
was
that
the
americas,
I
think
it
is
the
the
one
that,
for
me,
at
least
in
the
evening,
used
to
have
more
attending
the
attendance,
the
slot.
So
at
some
point
the
idea
was
floating.
We
just
always
use
that
slot.
Actually
we
have
four
people
today.
So
that's
that's.
That's
actually
pretty
good.
A
Yeah,
but
all
of
us
could
attend
the
more
or
less
the
the
american
meeting,
because
we
we
did
the
emea
one,
especially
for
people
which
are
which
lives
easier
than
us.
We
in
the
original
memberships,
we
had
a
lot
of
people
from
either
australia
or
or
so
in
that
area
or
in
russia.
So
it
was
kind
of
shifted.
A
B
C
B
Now
so
we
had
added
this
to
so.
This
is
the
general
guideline
for
if
you
are
a
developer-
and
you
just
want
to
add
a
new
thing-
that
you
can
upload
kind
of
what
are
the
steps
you
need
to
go
through
last
week,
we
also
said:
maybe
we
also
want
to
review.
B
You
know
this
for
completeness
and
kind
of
up-to-dateness,
because
I
think
jacob
also
said
it's
actually
not
super
clear
like
how
do
we
actually
implement
an
uploader?
We
don't
really
have
a
good
guide
for
this,
but
that's
maybe
separate,
so
that's
actually
something.
Maybe
we
can
also
consider
documenting
it.
It's
just
like
it
seems
like
we
haven't,
found
anyone
so
far.
Who
would
actually
write
this
documentation
because,
apparently
no
one
on
this
working
group
has
ever
written
an
uploader.
But
okay,
that's
maybe
a
separate
point.
So
so
we.
B
Tables
there's
two
one
is
about
this
mapping
from
what
is
kind
of
the
thing
that
you
upload
into
which
bucket
does
it
go
and
how
does
it
get
there
and
then
we
have
the
the
carrier
wave
integration,
which
is
like
yeah,
is
this
like
like?
Where
is
this
hook
defined?
Maybe
that's
just
more
for
yeah
developers
to
look
at
rather
than
guiding
like
future
development
but
yeah.
So
let's
go
through
this
so
with
job
artifacts
upload
technology
yeah.
Maybe
we
should
also
maybe
another
thing.
B
I
can
look
at
this
as
well.
If,
if
we
need
to
align
this
a
bit
with,
I
think
we
use
slightly
different
terminology
now
in
the
uploads
yeah.
A
B
Think
we
just
called
it.
I
forgot,
I
think
we
just
called
upload
strategy
or
something
that
wasn't
necessarily
intentional,
but
I
don't
know
I'll
find
out,
but
we
have
some
kind
of
definition
for
that.
So
this
looks
fine
though
I
would
expect
I
mean
this
is
really
the
only
combination
that
can
be.
B
B
Yeah
then
next
one
so
so
this
is
already
where
I
struggle
a
bit
because
upload
technology
carrier
wave,
I
think
what
we're
more
interested
in
is
like.
Does
this
go
through
a
rails?
So
if
it's
not
direct
upload,
then
we
need
to
think
about
what
are
the
other
alternatives
right.
Otherwise
we
have
this.
A
Yeah
here
counterweight
means
that
we
are
relying
on
cardio
wave
callbacks
to
upload
this
from
sidekick,
because
this
is
something
that
is
generated
from
an
asynchronous
job
on
sidekick,
okay,.
B
The
background
upload,
no
okay,
so
we
actually
have
a
manually
defined
sidekick
worker
specifically
for
pipeline
artifacts
yep
that
performs
okay,
then
this
is
correct.
All
right,
yeah,
that's
also
an
interesting
one:
life
job
traces.
This
is
a
unique
thing:
okay,
because
so
this
completely
has
its
own
separate.
B
A
Implementation
and
it's
really
complex
so
because
you.
A
Of
data
and
they
are
stored
in
reds-
if
I
remember
correctly,
and
then
when
we
have
enough,
they
are
uploaded
or
something
like
this.
This
changed
over
time,
but
this
is
completely
implemented
from
scratch,
is
using
none
of
the
other
technologies.
Okay
got
it.
B
C
B
Workers
actually
live.
That
could
be
something
yeah.
A
A
Yeah,
but
I
mean
no,
that
one
is
really
hard
to
find,
especially
then,
if
you
want
to
compare
with
the
next
one,
which
is
the
job
trace,
archives
right,
because
they
are
basically
doing
the
same
thing.
So
the
first
one
live
job.
Traces
is
handling
chunks
of
log
and
the
next
one.
If
I'm,
if
I
remember
correctly,
is
the
one
that
collects
those
chunks
and
and
create
a
single,
stable
final
log
file
for
us
makes
sense.
B
A
But
yeah
this
is
this
is
nothing
to
do
with
what
we're
doing
it's
it's
part
of
the
product,
but
it's
handled
by
the
runner
itself.
To
my
understanding,
that's
why
I
was
here
to
say:
okay,
it's
not
something!
I've
seen
uploading
technology
not
available,
so
it's
not
something
that
was
in
line
with
this
working
group
effort.
Yeah.
B
I
guess
same
for
the
backups,
then
yeah
exactly
then
we
have
yeah
get
lfs
files,
so
they're
coming
through
workhorse,
yet
download
yeah.
I
remember
actually,
okay,
that's
that's!
That's
fine!
Okay!
So
then
design
management
file.
So
this
means
okay.
So
it
goes
to
the
rails
controller.
But
it's
workhorse
assisted
right,
yep,
yep.
Okay,
all
right
actually
like
yeah
sure
is
that
something
we
should
do.
Should
we
rename
this
node
to
workhorse,
assisted
or
I
think,
that's
what
we
call.
A
A
A
Important
to
there's
an
important
distinction
here
that
maybe
was
not
important
at
the
higher
level
when
we're
just
explaining
how
things
work,
but
here
there
is
an
important
things,
which
is:
I
mean
you're
right.
It's
not
really
that
important,
yeah
you're
just
I
mean
as
long
as
you
know,
that
rails
controller
is
uploading.
It
means
that
it
is
not
directly
uploaded
by
workers.
Yeah,
that's
the
point,
so
you
can
be
workers
assisted
yeah.
It
makes
it's
the
same
problem
when
we
do.
B
C
No,
I
would
assume
that
both
the
files
and
the
thumbnails
goes
to
upload
design
management
and
whatever
structure
they
have
below
yeah.
The
design
management
files
into
the
ls
lsf
lfs
object
bucket
sounds
not
correct.
Yeah.
B
A
B
And
like
whoever
has
any
insight,
just
start
an
mri
and
we
can
go
from
there
all
right,
but
I
assume
this
is
correct,
then
so
the
thumbnails
they
go
through
sidekick.
I
guess
because
there's
some
post
processing
yeah
with.
B
All
right
cool!
No,
then
that's
correct
as
well.
This
looks
right,
yeah,
that's
the
general
uploads
personal
snippets
right.
B
Because
it
has
actually
a
different
bucket
structure:
yeah
that
looks
fine
appearance,
settings
right.
Jesus.
A
B
C
C
Yeah,
well,
I
don't
know
the
the
actual
code
that
uploads
this
or
process
this,
but
I
recall
seeing
some
spec
examples
for
the
multi-part
middleware
that
processed
the
direct
upload
object
from
workhorse,
and
there
were
some
examples
for
the
avatars
because
they.
B
B
C
A
B
All
right,
migration-
I
don't
know
what
this
is
there's
also
import
really
is
that
maybe
that
is.
A
I
think
it's
when
you
import
from
live
from
github
or
other
data
sources.
Okay,
but
that's
not
the
split.
A
B
Yeah
because
I
was
just
not
able
to
find
all
of
these
custom
hand
rolled
sidekick
workers
that
do
all
of
these
uploads.
I
don't
know
why.
Maybe
I
looked
in
the
wrong
place,
but
so
I'm
really
surprised.
That's
why
I
thought
it's
wrong
in
the
table.
I
thought
like
all
of
these
would
rely
on
backgrounds
upload
but
yeah.
Maybe
I'm
wrong.
B
C
So
on
packages
it's
direct
upload
and
workhorse,
except
for
one
format,
but
I
guess
we
can
just
ignore
it:
yeah,
which
is
npm
and.
B
C
It's
a
standard
upload
by
rails
because
it's
well
it's
an
upload
in
a
json
body
and
it's
not
currently
supported
by
workhorse.
So
that's.
C
But
for
all
the
other
formats
is
direct
upload
workhorse.
A
A
C
A
B
Just
break
out
a
new
row
which
somehow
we
can
just
say
pack
like
npm
package,
manage
manager,
archives
or
something.
C
We
could,
although
it's
like
out
of
the
nine
formats,
it's
only
npm
that
is
not
supported,
so
it's
more
of
an
exception
than
anything
else.
If
you
want.
B
A
C
A
More
knowledge
and
think
about:
how
can
we
rewrite
those
three
lines
in
something
that
makes
sense
yep.
Okay,
thank
you.
C
Yeah,
it's
the
dependency
proxy
cache
which
caches
blobs
and
manifests
from
container
images
so
yeah.
I
will
fix
that
role
too.
I
guess
okay,
the
other
ones.
I.
B
Mean
how
how
about
like?
Maybe
you
can
like
open
an
mr
with
these
changes
and
if
we
have
other
things
to
contribute,
you
can
just
all
assign
us
as
reviewers
as
well.
We
can
all
contribute
fixes
where
we
think
something
can
change,
and
we
can
have
a
single,
mr,
that.
B
All
right,
terraform
state
fights
that
I
think
was
the
most
recent
one.
I
think
we
had
already.
A
B
A
Another
one
I
have
no
idea,
so
this
is
oh
no.
This
is
easy.
I
can
tell
you
what
is
happening
here,
so
this
is
for
when
we
migrated
pages
from
nfs
to
object
storage.
A
So,
basically,
before
we
had
a
sidekick
job,
there
was
unpacking
the
artifacts
supposed
to
be
deployed
by
gitlab
pages
into
an
nfs
share
that
was
shared
with
the
gitlab
pages
fleet.
Now
those
things
are
kept
as
a
zip
archive
into
object.
Storage
and
pages
is
reading
the
content
of
the
website
directly
from
object
storage
inside
the
zip
file,
because
we
have
a
way
of
getting
things
inside.
A
B
Okay
and
his.
B
Looks
like
all
right
so
to
recap
and
yeah,
so
we
can
just
collaborate
on
an
mr,
so
that
was
mostly
so
fixing
these
three
breaking
out
an
export
row
yeah.
Maybe
we
can
rename
this
to.
I
think
upload
strategy.
I
think
strategy
is
what
we
use
just
to
be
consistent
with
the
terminology,
and
I
think
the
rest
was
actually
correct
and
what
I
would
maybe
propose
is
whenever
it
says
sidekick
to
the
extent
we
know
where
it
is,
and
I
can
find
it.
B
C
A
B
C
Have
to
do
it,
can
I
ask
a
question
for
since
we
have
a
workhorse
maintainer,
so
it
will
be
more
efficient
on
the
sections
about
implementing
a
new
route?
That's
a
section.
B
C
C
A
C
D
Right
question:
if
I
mean
if
we
are,
if
we
are
done
with
that
part
of
the
of
the
document,
if
not,
I
can
help,
I
can
wait
a
little
bit.
A
So
I
I
think
we
have
the
next
list
on
the
document
and
then
we
are
good
to
move
on
the
next
table,
which
is
the
which
is
the
category
of
integration.
A
It
kind
of
give
us
where
we
don't
know
something,
because
there
are
some
of
those
carrier
wave
mounted
stuff
that
have
no
yes
mark
in
the
categorized
table.
Let
me
just
share
it.
D
Well,
I
think
it
should
definitely
be
owned.
It's
the
secure
team,
actually,
they,
I
think
they
own
those
kind
of
features.
D
Go
ahead,
yeah
like
everything
like
like
dashboards,
security,
dashboards
and
so
on,
like
when
you
go
on
security
and
compliance
menu
on
the
left.
You
actually
like
a
bunch
of
security
related
stuff
and
all
those
features
there.
There
are,
if
I'm
not
mistaken,
normally
owned
by
the
security.
D
Checking
but
normally
a
secure
team
actually.
D
Ya
yeah,
so
the
secure
the
secure
the
team
actually
owns
like
static
analysis,
dynamic
vulnerability
so
so
yeah.
It
should
normally
be
owned
by
them.
A
Yeah
I
mean
right
now
we
did.
We
already
did
all
the
effort
in
trying
to
figure
out
categories,
so
it
should
be
quite
easy
to
get
those
two
values.
So
how
is
it
uploaded
that
that's
basically
the
question?
So
maybe
we
can
try
to
to
squeeze
this
in
in
this,
mr
that
we
that
david
is
going
to
open
and
try
to
also
tick
those
new
elements.
So
we
say:
okay,
we
nothing
was
left
uncategorized,
but
yeah
I
mean
we
we
can.
We
can
handle
this
when
we
do
the
dmr.
A
I
think
it's
just
a
few
of
them
and
just
a
matter
of
pinging
some
em
on
those
on
those
feature
category
and
getting
the
information
out.