►
From YouTube: Velero Community Meeting - October 4, 2022
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
Okay,
okay,
we're
recording
hello.
Everyone
today
is
October
4th
and
that's
official
community
meeting
for
Valero.
My
name
is
Ruben
vasilif
and
I'm
the
community
manager
for
Valero.
A
So
please
follow
the
code
of
conduct
and
be
nice
to
each
other.
Very
much
with
that.
I'm
gonna
kick
off
by
saying
I've
pasted
in
the
chat.
The
link
to
the
community
meeting
notes.
I'm
gonna
send
it
again
for
the
rest
that
just
joined
I,
don't
know
if
you,
if
you're
receiving
it.
If
you
join
later
with
this
one,
I'm
gonna
share
my
screen
for
a
second
thank
you.
Everyone
for
signing
up
into
the
attendance
for
today.
B
B
It's
actually
needed
for
one
of
the
main
things
we're
focusing
on
for
the
next
release,
so
we
kind
of
need
to
get
this.
You
know
wrapped
up
and
improved
relatively
soon.
This
is
going
to
allow
us
to
go
forward
with
some
data
mover
work
as
well
as
other
things,
basically
any
plugin
that
will
wants
to
initiate
some
other
action
in
an
external
controller,
so
that
could
happen
without
blocking
bolero
and
then
Valero
comes
back
and
checks
status
later,
so
the
Design's
been
up
for
a
while.
B
We've
had
some
comments
and
updates
and
then
related
to
that.
There's
the
because
we
do
have
110
is
going
to
have
the
basic
functionality
around
allowing
for
plug
and
API
versioning
110
won't
have
any
new
plugin
apis
defined,
but
the
infrastructure
is
in
place
to
allow
additional
kind
of
changes
to
the
plug-in
design.
This
will
allow
us
to
modify
return
signatures,
add
methods
to
plug
in
apis
without
breaking
existing
plugins
and
that's
kind
of
the
goal
there,
and
we
have
a
couple
of
designs
up
for
the
actually
yeah.
A
B
B
B
We
need
to
modify
the
plug-in
apis
for
backup,
item
action,
restore
item
action
and
the
volume
snapshotter,
and
so
this
is
kind
of
the
big
overall
design,
and
these
are
and
I'll
check
after
the
meeting
to
double
check
a
visit
there
when
I
need
a
link
because
I
there's
the
back.
There's
the
backup
item,
action,
design.
B
Oh
actually,
I'm,
sorry
I'm,
sorry.
This
is
I
linked
the
wrong
thing,
a
pulse
mental
meeting,
so
this
was
just
a
refactoring
for
me.
One
that's
already
done.
I
will
update
the
meeting
notes.
But
if
you
look
at
open
PR's
by
me,
there
should
be
a
couple
of
plug-in
yeah
if
you
just
look
at
sorry,
yeah,
PR's
and
then
search
for
yeah
and
I'll
update
the
meeting
notes
to
link
to
those
properly
after
the
meeting.
But
I
have
two
designs
up
for
backup.
B
Item
action,
restore
item
action,
the
backup
item
action
is
basically
implementing
or
calling
for
the
implementation
of
the
changes
that
we
need
for
that
async
plug-in
design,
we're
adding
a
a
stat
I.
Think!
That's
what
we
call
it.
It's
oh
yeah,
progress
method,
so
so
we
can,
after
the
backup
is
done.
We
call
in
progress
on
the
for
the
plugin
to
see
if
the
external
action
is
completed,
there's
also
a
cancel
function
and.
B
Yeah
there's
some
comments
here
from
Lyndon,
because
there
are
some
additional
things
they
were
thinking
of
that
we
might
want
to
be
too
around
there's
some
concern
about
some
of
the
storage
related
plugins,
specifically
around
the
the
CSI
snapshots
kind
of
manually,
calling
into
Valero
and
looking
up
things
relating
to
Rustic
and
kind
of
the
thinking
was.
We
might
want
those
to
be
able
to
have
it
built
into
the
API
rather
than
just
kind
of
applying
it
manually,
but
there's
no
proposed
design.
Yet
for
that
so
I.
B
The
basic
approach
that
we've
been
taking
with
kind
of
how
we
wanted
to
to
handle
plug-in
API
designs
is
that
we
start
with
the
design
for
a
feature,
and
so,
for
example,
we
had
the
item
action.
Progress
feature
that
we
looked
at
there's
another
feature
that
lyndon's
referencing
here,
where
we
want
to
add
some
explicits
kind
of
changes
to
the
plugins,
so
that
plugins
can
know
whether
or
not
you
know
rustic
is
relevant
and
I.
B
The
concern
there
is
that
request
doesn't
have
a
design.
Yet
we
haven't
come
up
with
what
those
API
changes
would
be
so
I
guess.
The
point
here
is
that
there's
kind
of
a
two-step
process,
with
the
API
changes
for
plugins
one,
is
to
define
the
functionality
to
find
what
we
need
get
that
agreed
upon
An
approved
design,
and
the
next
step
is
to
propose
an
explicit
change
to
the
API
that
says:
hey.
We
need
this
new
function.
B
B
So
this
design
is
really
just
what
we
need
to
handle.
This
asynchronous
plug-in
work
and
I've
already
I
have
a
I
had
a
proof
of
concept
PR
that
I
put
together
not
to
merge
just
to
make
sure
that
I
could
Define
a
new
plugin
for
backup
item
action
that
added
a
new
dummy
function.
It
worked,
so
this
is
basically
implementing
the
same
idea,
but
with
real
and
actual
needs
for
Valero.
So
we
see
we're
adding
the
here.
B
B
Oh
this,
the
we
have
in
mind
immediately
in
fact
we're
kind
of
doing
something
on
the
ADP
side
of
red
hat.
Is
it
did
they
remember?
Well,
we
have
a
data
mover
for
the
current
oadp
release
that
uses
the
existing
backup
item
actually
startim
action.
B
But
you
know
we
have
some
performance
issues
because
it's
a
blocking
action,
so
whatever
the
plugin
does
you
know,
we've
got
to
wait
till
that
finishes,
and
so
this
is
a
design
that,
once
it's
in
place,
we're
going
to
be
able
to
basically
use
on
on
that
side
and
the
the
idea
also
is
eventually
to
have
a
fully
functional
data.
B
Mover
Upstream
in
Valero
would
make
use
of
this
as
well,
the
idea
being,
for
example,
you
have
a
plug-in
for
you
know,
for
for
a
volume
snapshot
and
that
might
create
some
additional
CR,
that
the
data
mover
controller
would
pick
up
and
and
go
and
actually
move
that
data.
So
the
Valero
plug-in
would
create
that
CCR
that
the
data
mover
plugin
needs.
Add
it
to
the
backup
and
exit
go
back
to.
B
Valero
control
goes
back
to
Valero,
so,
while
Valero
is
going
to
the
next
item
to
back
up
this
plug-in,
rather
this,
this
data
mover
controller
in
a
completely
different
thread,
is
acting
on
that
CR.
That
was
created,
then,
at
the
end
of
the
backup,
when
all
the
things
have
been
backed
up.
Valero
goes
back
to
all
those
plugins
that
had
registered
hey.
B
We
created
this
thing:
it's
an
external
operation
and
we're
checking
status,
and
so,
if
everything
is
completed
at
that
point,
so
if
if
this
data
mover
action
for
example,
starts
and
finishes
before
the
Valero
backup
is
done,
then
at
the
end
of
the
Valera
backup
everything's
complete
backup's,
complete
we
go
on,
but
the
main
point
of
the
async
plug-in
design
is
if
those
actions
aren't
complete.
Yet
you
know
you're
backing
up
this
huge
volume,
so
Valero
is
done
with
the
backup,
but
the
volume
data
mover
work
is
still
happening.
B
B
So
that's
that's
the
point
of
these
changes
here
and
so
there's
a
cancel
operation
here
as
well
and
the
re.
The
purpose
of
that
is.
If
Valero
hits
a
back,
it
hits
a
configurable
cancel
kind
of
a
timeout.
This
is
kind
of
analogous
to
The
Rustic
timeout.
We
have,
for
example,
so
that
if,
if
The
Rustic
backup
is
goes
longer
than
a
certain
period
of
time,
an
existing
Valero
releases,
then
we
time
it
out
and
we
feel
you
know
we
we
add
that
as
a
backup
or
restore
failure.
B
This
is
kind
of
analogous
to
that.
If,
if
the
timeout
has
hit
on
this
async
actions
and
they're
still
you
know
running,
then
Valero
will
declare
it
a
failure
and
the
full
partial
failure
in
this
case
and
then
call
the
cancel
API
call
the
method
which
will
then
allow
a
plug-in
to
you,
know:
optionally,
cancel
that
option
operation
if
you
can
and
plugins
may
not
be
able
to
cancel-
and
maybe
some
external
thing
has
no
control
over,
in
which
case
it
would
just
be
no
op.
C
Yeah,
no
so
this
was
like
one
of
the
original
requirements
was
like
for
to
be
able
to
like
say
if
you
are
spawning
a
process
to.
Why
is
a
database
or
something
that
is
stuck
to
be
able
for
the
plugin
to
be
able
to
cancel
those
kind
of
calls.
B
In
the
context
here,
when
the
when
you
plug
and
execute
action,
happens
in
the
first
place,
so
so
again,
if
we're
looking
at
a
volume
snapshot
creation,
for
example,
Valero
plug
and
execute,
creates
the
data
mover
CR,
which
then
at
that
point
the
data
mover
can
start
acting
on
this
volume
and
doing
what
it
needs
to
do
to
back
it
up
and
copy
it
into
whatever
you
know,
storage,
it's
copying
to
control
returns
to
Valero
to
go
on
to
the
next
item
at
the
end
of
the
backup
and
then
periodically
Valero
is
checking
hey.
B
Is
this
thing
done
so
it
calls
the
progress
method
passes
in
that
operation
ID
and
the
plug-in
returns
back?
Yes,
it's
done!
No,
it's
not
done.
Here's
what
you
know,
percentage
done
or
whatever
and
Valero
keeps
calling
that
and
kind
of
pulling
it.
There's
some
configurable
timeout.
Maybe
it's
an
hour!
Maybe
it's
three
hours
whatever
you
decide,
you
know
and
there's
going
to
be
some
default.
B
If
you
hit
that,
you
know
one
hour
three
hour
whatever
time
period
and
it's
still
not
done
at
this
point,
Valero
is
going
to
just
declare
the
backup
partially
failed
and
declare
this
part
not
successful,
and
it
will
call
that
cancel
operation
on
the
plugin,
which
will
give
the
plugin
the
opportunity
to
cancel
the
cancel
the
operation.
If
you
can't-
and
it
may
be
an
operation
that
Valero
can't
cancel
this
kind
of
the
plugin
is
at
this
point
at
the
mercy
of
whatever
external
process
it
kicked
off.
B
If
it's
a
process
that
can
be
canceled,
then
the
cancel
operation
and
that
Valero
plugin
would
then
go
update
that
CR,
you
know
set
cancel
as
true
or
whatever
the
API
calls
for
to
cancel
the
operation.
If
it's
an
operation
that
can't
be
canceled
once
it
started,
then
it
won't
do
anything.
It'll
just
return
foreign,
but
that's
that's
up
to
the
plug-in
to
implement
this
is
this
is
just
declaring
that
you
know
the
API
has
this
function
in
it
that
needs
to
be
implemented.
B
Yeah
and
then
so
and
restore
item
action
is
slightly
different.
It's
got
two
things
in
it.
Actually,
we
have
the
same
kind
of
corresponding
changes
relating
to
the
async
plugin
works,
so
we
have
that
progress
and
cancel
we're
modifying
the
execute
method,
return
to
be
able
to
return
that
optional
async
operation
ID.
B
All
of
that
is
pretty
much
one-to-one
correspondence
with
what
we're
doing
on
the
backup
item
action
again
initially
to
support
the
data
mover
use
case,
but
there
you
know
there
could
be
other
external
plug-in
use
cases
that
you
know
basically
any
plug-in
that
needs
to
run
some
operation
that
could
take
a
long
time.
It
doesn't
mean
to
block
Valero's,
backup
or
restore
process
on,
would
be
a
candidate
to
use
this
API
and
again.
The
restore
item
action
basically
includes
the
same
changes
as
the
backup
item
action
for
that
feature.
B
One
of
the
things
that
happens
on
with
restore
plugins
is
there's
a
there's,
an
additional
items,
slice
that
a
restore
plug
in
x.
You
can
return
and
the
meaning
of
that
for
Valero
is
if
I'm,
if
I'm,
restoring
a
particular
item
and
I
return,
a
list
of
additional
items,
I'm
telling
Valero
these
things
in
the
backup
need
to
be
backed
up
first
before
we
restore
or
sorry
it
needs
to
be
restored.
B
First,
before
restoring
the
current
item-
and
this
is
to
handle
kind
of
internal
dependencies
at
a
per
item
level,
this
was
generated
in
response
to
a
very
specific
bug
we
had
on
the
openshift
side,
but
it
could
be
applied
to
anything
where,
where
it
matters
most
plugins
won't
need
this,
because
the
way
Bolero
actually
the
way
kubernetes
works.
With
these
internal
references
from
one
to
another,
is
you
know
the
notion
of
kind
of
eventual
consistency.
B
B
So,
for
example,
I
can
create
a
deployment
that
references,
a
PVC
PVC
doesn't
exist
yet
I
create
the
BBC
later
once
everything
exists,
the
controllers
reconcile
everything's
fine
problem
is
there
are
certain
some
controllers
out
there
that
don't
follow
this
principle
very
well,
and
if
you
try
to
create
something
that
references
a
non-existent
item,
there
are
some
controllers
out
there
that
will
fail.
B
We
ran
into
this
with
the
openshift
image
stream
tags,
where,
if
I
create
an
image
stream
tag
that
references,
another
image
stream
tag
and
that
reference
tag
doesn't
exist
to
create
call
fails.
B
So
our
plugin
for
that,
when
there
is
a
reference
like
that
and
our
Valero
plugin
returns,
those
additional
items
saying
hey,
restore
these
things
first,
because
we
need
them
before
this
thing
is
created,
but
there's
still
a
race
condition
there,
because
you
know
you,
you
issue
the
create
call.
You
immediately
issue
the
next
create
call
that
depends
on
the
first
create
call,
but
sometimes
with
kubernetes.
If
you
do
a
get
immediately
after
you
create,
you,
don't
see
it
yet.
B
So
there
are
still
cases
where,
if
you
immediately,
you
know,
run
these
two
creates
one
after
the
other
things
still
fail.
B
So
we
have
the
notion
of
a
a
wait
for
additional
items
to
be
ready
flags
that
we
pass
here,
the
idea
being,
if
a
plug-in
chooses
to
return
that
flag,
then
after
it
goes
through
the
additional
items,
creation
it'll
call
back
into
the
plugin
saying:
hey
are
the
items
ready
and
the
plugin
will
check
to
see
if
everything's
ready
and
if
it's
ready
it
turns
true
and
then
Valero
creates
the
item,
and
then
it
goes
on.
Does
it
just
finishes
the
store?
If
it's
not
ready?
B
Yet
we
go
into
this
weight
Loop,
where
we're
going
to
wait
and
try
again.
So
that's
a
feature
that
was
a
design
that
was
already
approved
some
sometime
back.
We
originally
were
hoping
to
get
it
in
one
nine
and
maybe
even
one
eight
I
think,
but
because
we
didn't
have
plug-inversion
and
we
couldn't
implement
this
because
without
plug-in
versioning,
you
know
if
we
added
a
new
function
to
the
plugin
API,
it
breaks
all
previous
existing
plugins
and
that
wasn't
acceptable.
B
So
now
that
we
have
plugin
version-
and
we
can
add
these
new
fields
in
the
new
method
to
the
plugin
V2-
that
way
old
plugins
that
don't
implement
this
are
okay.
B
So
this
design
combines
two
plug-in
changes,
because
one
of
the
things
that
we
had
agreed
to
when
we
approved
the
plug-in
design
is
that
when
we
create
a
new
plugin
API
version,
you
know
we're
not
going
to
create
a
V2
V3
before
for
every
single
thing.
We
add,
because
if
I
add
two
or
three
things
within
one
Valero
release,
it
makes
no
sense
to
release
Unity
at
the
same
time.
So
we
collect
all
those
changes
within
a
Valera
really
so
Valero
110,
for
example,
will
not
have
any
of
this
it'll.
B
Just
have
you
one
for
everything,
but
Valero
111,
hopefully,
will
include
both
of
these
changes
to
the
restore
item
action.
You
know
the
one
change
to
backup
item
action
and
then
you
know
back
to
the
the
comments
on
the
previous
PR
we
looked
at
or
I
think
was
Lyndon.
That
was
suggesting
hey.
We
might
want
to
think
about
handling
this
issue
with
rustic.
B
B
So
so
that's
what
these
are
in
terms
of
and
kind
of
where
they
came
from
and
there's
one
more
design
that
needs
to
be
added
to
these
and
that's
the
volume
snapshotter,
because
that
also
needs
the
same
kind
of
changes
because
we're
supporting
the
asynchronous
actions
on
volume,
snapshot
uploads
as
well,
and
so
and
then
this
then
this
was
the
refactoring
for
the
V1,
it's
a
kind
of
preparation
for
it,
and
so
we'll
we'll
need
a
V2,
API
design
and
once
that's
approved,
then
we
can
actually
put
the
B2
in
place
and
I'll
update
the
notes
here,
because
I
I
was
looking
at
the
wrong
here.
B
B
So
right
now
actually
merging
these
is
I,
think
still
dependent
on
that
first
PR.
We
looked
at
the
the
overall
item
action
progress
because
this
this
is
a
specific
plug-in
API
to
design
change,
that's
dependent
on
the
kind
of
feature
API
in
a
fishery
feature
design
here,
so
this
one
will
need
to
be
approved
and
merged
before
the
others
are
merge
ready.
Just
because
you
know
if
this,
if
they're
requested
changes
here,
for
example,
that
and
end
up
with
me
updating
those
designs.
B
B
You
know
action,
I
I
know
when
we
were
kind
of
approaching
the
110
future
freeze,
the
other
maintainers
were
you
know
busy
with
other
things,
and
then
they
didn't
have
time
to
get
to
this,
and
so
we
kind
of
need
to
and
I
know,
I
guess,
I
think
most
of
them
are
on
holiday
this
week,
because
I
know
China
has
a
week-long
holiday,
so
hopefully,
basically
next
week
when
the
rest
of
the
maintainers
are
back.
B
These
are
things
that
really
could
use
some
additional
attention,
but
at
this
point
we're
kind
of
waiting
on
getting
the
requisite.
You
know
people
on
the
maintainer
side
to
kind
of
approve
the
design.
There's
been
some
comments,
but
and
also
obviously,
we
need
to
make
sure
that
you
know
previous
comments.
A
A
B
Know
I'd
like
to
see
this
merge
in
the
next
couple
of
weeks,
just
so
that
I
can
then
get
to
a
point
where
okay,
the
because
once
we
get
the
V2
plug-in
API
designs,
you
know
approved
at
that
point.
You
know
these
aren't
going
in
110
so,
but
we
can
start
implementing
those
basically
as
soon
as
the
110
release
branches
cut,
because
at
that
point
we
can
merge
things
to
main.
You
know
for
111..
B
This
is
something
we
want
to
do
early
in
the
release
cycle,
because
there's
features
that
are
then
dependent
on
these
API
changes
so
that
this
is
kind
of
you
know.
My
thinking
here
is
that
we
get
all
this
stuff
approved
by
the
time.
110
is
released
so
that
you
know
first
thing
in
the
111
release
cycle.
We
can
get
these
changes
kind
of
put
in
place.
A
I'll
I'll
get
to
note
on
this
one
and
upcoming
Monday,
when
our
folks
from
China
are
back
I'll,
try
to
address
that
to
to
everyone
into
the
maintenance
team
to
take
a
look
to
try
this
forward
great.
Thank
you.
A
Okay,
all
right
anyone
questions
on
these
two
actually
Four
PRS.
A
Oh
okay,
so
next
one
I've
added
a
topic
to
this
course.
I
know
I
spoke
with
jaolin
during
my
vacation.
We
have
discussed
that
and
he
summarizes
it
over
here.
Oh
actually,
there's
a
new
command
USD
about
the
time
change
for
the
community
meeting
to
change
instead
of
having
it
like.
It
is
right
now,
every
first,
every
second,
every
third
and
every
fourth
to
be
bi-weekly,
so
we
don't
have
the
situation
as
we
had.
A
We
have
two
weeks
in
a
row
without
a
specific
meeting
and
also
I
think
Scott
or
someone
from
overhead
proposed
to
move
that
meeting,
that
we
have
right
now,
two
hours
earlier,
so
that
will
allow
the
Beijing
team
also
to
be
able
to
join
that
that
meeting
every
now
and
then
not
always,
but
it
will
be
at
least
a
bit
more
comfortable
for
them
to
join.
A
So
can
you
everyone?
Can
you
take
a
look
at
this
proposal
from
Geraldine
and
maybe
share
your
thoughts
right
now?
It's
it's
great
for
me.
Personally!
It's
a
European
Time
Zone,
it's
great!
If
you,
if
we
move
that
earlier.
B
So
so
the
original
reason
for
just
having
it
at
noon
was
to
accommodate
West
Coast
time.
At
this
point,
I
don't
believe
we
have
any
regular
attendees
on
the
west
coast
I.
You
know
crack
if
I'm
wrong
here
but
I
don't
know.
C
B
Currently
maintainer
but
I,
don't
believe
he
has
attended
in
a
few
months.
I
don't
know
what
his
plans
are.
Dave
you
know,
Dave
is.
B
As
a
maintainer,
but
he's
has
not
participated
in
the
meetings
recently,
I
don't
know
that
he,
if
he
plans.
B
Okay,
so
so
that
that,
but
that
that
was
my
thinking
of
just
based
on
actual
participation
moving
it
two
hours
earlier,
you
know
it
would
make
it
harder
for
people
in
the
Pacific
coast
to
come,
although
they
could
still
come
to
the
the
other
meetings
you
know,
but
it
would
allow
the
majority
of
the
maintainer
team
more
flexibility
in
coming,
because
the
current
time
is
midnight
for
them
and
then
they're
not
going
to
ever
join
that
really
and
10
o'clock
is
still
late.
But
it's
you
know
it's
not
insane
late.
B
It's
you
know
it's
it's
one
of
these.
If
it's
an
important
topic
they
want
someone
wants
to
come,
they
probably
can
yeah
so
so
that
was
my
thinking
and
I
think
the
bi-weekly
versus
one
you
know
first
through
fourth
yeah,
that
that
makes
it
easier
to
handle
the
whole.
You
know
here
there
kind
of
meetings,
and
once
we
pick
a
start
time,
you
know
it.
It
makes
it
harder
to
guess
without
looking
at
the
calendar.
B
You
know
which
meeting
is
today,
but
as
long
as
those
everyone's
calendars
has
those
bi-weekly
meetings
on
there
I
mean
you
know,
we
all
have
other
meetings
that
are
bi-weekly.
You
know,
that's
something.
That's
relatively
you
know.
Common
I.
B
Yeah
yeah
yeah,
yeah
and
yeah,
it's
a
little
confusing
when
we
you
know
about
half
the
month,
have
a
fifth
you
know
Tuesday,
which
but
there's
no
meeting,
and
you
know
it's
like
okay
and
then
it's
you
know
so
I
I
I
think
that's
a
good
suggestion
too.
A
C
Specific
study,
so
what
does
the
time
look
like
so
7
AM
for
same
name
and
then
the
other
would
be
it
7
PM?
Is
it.
B
Well,
no
because
it's
well,
it's
9am
right
now,
but
if
we're
talking
about
moving
it
to
ours
earlier,
that
was
at
the
7am.
That
would
be
like
this
yeah.
A
B
B
Like
here,
yeah
yeah
the
Beijing
time,
so
it's
right,
so
it's
8
P.M.
So
so
that's
5
p.m!
For
okay,
so
West
Coast!
Yes!
So
so,
basically,
with
that
proposal
it
would
be
one
week
is
5
p.m.
West
Coast!
The
next
week
is
7
A.M.
C
C
B
My
thinking
with
the
change
is
that
you
know
7am
is
kind
of
in
a
similar
category,
I
think
to
the
to
the
for
the
Beijing
people.
This
puts
it
at
a
time-
that's
slightly
inconvenient,
but
it's
not
entirely
impossible,
whereas
at
the
current
time
it's
more
convenient
for
West
Coast
9
A.M
is
obviously
better
than
seven,
but
it's
midnight
for
China,
where
half
the
more
than
half
the
maintainer
team
is,
and
so
it's
basically
impossible
for
them
to
join
at
all.
So
that
was
my
thinking
of
moving.
B
It
is
to
make
you
know
it's
kind
of
a
trade-off.
It
is
slightly
more
inconvenient
for
some
people,
but
it's
you
know
it
gets
to
it
to
a
point
where
I
think
it
kind
of
maximizes
the
ability
of
maintainers
and
other
community
members
to
participate.
You
know
and
if
there's
other
suggestions,
that's
fine
too
I
just
that
was
just
my
thinking
was
because
I
mean
last
last
time.
For
example,
when
we
had
this
meeting
there
was
one
or
two.
B
A
I
I
find
your
proposal
super
reasonable
and
I
I.
Think
if
everyone
is
happy
with
this
one,
someone
said
that
funk
left
you
mean
left.
C
A
C
Yeah
he
left
down
so
yeah.
He
won't
be
participating.
Okay,
so
I
think
Scotty
have
taken
almost
of
the
plug-in.
B
B
B
To
him
over
the
summer,
because
I
know
he
he
had
some
other
things.
You
know
Dell
that
we're
taking
up
his
time.
He
was
otherwise
occupied
and
he
didn't
have
time
to
finish
the
plug-in
versioning.
So
I
took
over
that
for
him
because
of
that
I
guess
about
a
month
and
a
half
ago.
B
Oh
sure,
that'll
be
good,
but
yeah
yeah
I
talked
to
phone
quite
a
bit
in
the
in
the
handoff
to
make
sure
that
I,
you
know
I
kind
of
caught
up
to
where
he
was
in
the
vlogging
version
and
kind
of
got
everything
in
place
that
he
was
still
had
remaining,
so
the
pr
that
he
started
that
he's
been
working
on.
We
got
that
merged
a
few
weeks
ago
and
then
what
I've
done
Stinson
was
kind
of
following
on
to
that
work,
kind
of
completing
it.
C
A
It,
okay
great
so
you
say
some
folks
will
join
and
they'll
be
also
working
and
assignments.
C
Yeah
so
right
now
so
we'll
have
one
or
two
folks
joining
Daryl
again
our
team
and
then
yeah
we'll
have
more
active
participation
at
the
Valero
development.
Perfect.
A
Talking
about
new
folks
and
old
folks.
I
spoke
with
Dave
Wichita
about
the
the
architecture
role
and
his
involvement.
Obviously
he
won't
be
active.
So
we
have
to
figure
out
election
for
new
architect
and
to
take
someone
to
take
over
that
and
to
lead
that
effort
out.
I'll
speak
with
pradeep
the
the
PM.
How
we
can
drive
this
forward,
but
but
I
got
the
message
from
from
Dave
that
he
won't
be
able
to
take
the
architectural
role
I'm.
Just
waiting
confirmation
came
from
his
side.
A
If
he's
gonna
continue
to
be
part
of
the
maintenance
team,
because
I
think
he
has
a
structural
role
into
the
team,
but
he
cannot
play
the
architect
anymore.
So
we
need
a
new
phase
there
to
drive
this
effort.
A
I'll
keep
you
updated
on
this
one,
how
to
be
honest,
I
as
as
of
now
I'm,
not
sure
how
we
have
to
approach
that
thing,
because
it
was
always
a
VMware
folk
to
be
the
architect.
So
yeah
give
me
time
to
figure
this
out.
What's
the
proper
way
to
to
approach
this
one
and
we'll
do
that,
is
there.
A
That's
that's
the
thing
that
I
don't
know.
No,
because
I
got
the
message
from
Dave
like
five
minutes
ago,
so
I'm
I'm
unprepared
for
this
one
I
just
wanted
to
show
that
Dave
yeah
responded
to
to
my
question.
If
he's
going
to
be
active
or
not
and
but
I'm
unprepared
for
the
rest
of
the
question,
how
are
we
going
to
proceed
so.
B
Yeah
I
mean
we,
we
do
have
a
governor
stock,
so
I
don't
off
top
of
my
head
I,
don't
remember
where
that
doc
redresses
how
this
is
handled
it
needs
to
address
it.
I
think.
The
first
thing
we
need
to
do
is
figure
out.
You
know
what,
like
you
said,
you
need
to
figure
out
how
to
proceed,
whatever
your
going
to
do
with
that.
If
it's
not
in
the
governance
stock,
it
needs
to
be
in
the
governor's
dock,
yeah
and.
D
B
Like
you
said,
you
know
in
the
past
it
was
VMware,
and
maybe
it
still
is
maybe
that
changes.
Maybe
it's
been
flexible
whatever,
whatever
that
process
is
needs
to
be
in
the
governance
Dock
and
that
changed
the
government
stock
needs
to
follow
the
you
know,
the
procedures
that
we
have
to
follow
for
those
changes,
yep.
A
A
C
I,
don't
see
him
here
so
so
this
was
a
issue
we
saw
at
customer
site.
It
is
they're
using
openshift
and
Blair
perverty
uses
of
ADP
right.
So
I
was
hoping
to
get
some
input
from
the
community,
so
buggy
is
open
for
a
while,
so
I
know,
I
know
we
have
red
hat
folks
here,
so
they
are
using
this
Matrix
application
and
I
got
that
error
in
the
front.
I
don't
know
maybe
Scott-
or
you
have
comments
on
this.
C
If
you
scroll
up
right
earlier
in
the
beginning
of
the
bug
which
they
are
back,
we
are
Bank
of
crd
and
then
it
is
looking
for
a
V1
beta1,
but
I
didn't
see
any
API
Resources
with
even
better
ones.
I
was
kind
of
confused.
B
That
was
defined
so
so
well,
first
of
all,
are
we
so
it's
failing
in
the
backup
is
that
is
that
right,
yeah.
B
Okay,
so
so
yeah
yeah,
I
I
haven't
seen
this,
but
it
looks
like
it's
it's
trying
to
back
up
something
and
it
can't
find.
C
Yeah
he
he
DM
me
on
slack
and
yeah
from
the
last.
It
looks
like
the.
D
B
B
Valero
has
code
around.
You
know
what,
when
you're,
when
we're
backing
up
a
resource
it
tries
to
back
up.
You
know,
multiple
versions.
If
they're
available
I
don't
know
how
that
works
in
the
context,
whether
that's
an
issue
with
the
Cod
or
whether,
maybe
that's
it's.
This
may
actually
be
in
that
code,
where
it's
trying
to
back
up
all
the
versions
of
the
crd
that
are
defined
in
the
in
you
know
in
that
CR,
but
the
crd
itself
doesn't
Define.
B
C
Be
something
so
the
so
the
CR
is
basically
this
one
right,
MTX
engines,
matrix.com,
but
I.
We
didn't
see
any
reference
to
V1
beta1.
It
was
V1,
so
that's
very
confusing
where
it
was
getting
even
better
one
reference
from.
B
Yeah
yeah
I
I'm
thinking
this
I'm
just
kind
of
thinking
about
now.
This
makers,
because
one
of
the
things
that
Valera
tries
to
do
is
that
you
know,
because
we,
when
you
just
do
a
you,
know,
Cooper
to
get.
You
know
on
a
resource.
You
get
the
the
kind
of
the
default
version,
but
you,
if
multiple
versions
of
that
of
that
resource
type
exist,
then
you
can
retrieve.
You
know
the
V1
or
the
one
beta
one
and,
and
one
of
the
things
that
Valero
tries
to
do
to
handle
compatibility
and
restore.
B
Is
that
you
know?
For
example,
if
you
have
a
deployment,
that's
got
V1
beta,
1
and
V1
Define
in
the
cluster
you
can
blur
will
back
up
both
of
those
that
way
on
the
restore
side,
if,
if
you're
restoring
to
a
new
newer
kubernetes,
that
only
has
the
one
not
B1
beta,
one
it'll
still
work,
so
it
may
be
in
that
code,
where
Valero
is
recognizing
that
this
Sierra
was
created
with
you
know,
B1
beta1
is
a
valid
version,
so
it's,
but
then
that
version
was
removed
from
the
cluster.
B
The
CR
still
works
because
it
also
has
V1,
but
Valero
is
trying
to
back
up
all
the
versions
of
it
so
and
if
that's
correct,
maybe
what
Valero
needs
to
be
doing
here
is
instead
of
erroring
out.
B
You
know
if
you
have
a
resource
where
it's
it's
trying
to
back
up,
say
two
different
versions
of,
and
only
one
of
them
works
and
one
of
them
doesn't
exist
that
should
probably
be
logged
as
a
warning
and
Velar
move
on
again
I'm,
just
guessing
right
now,
I
haven't
I,
haven't
reproduced
this
myself,
but
it
looks
like
what's
happening,
is
Valero
is
trying
to
you
know
back
it
up
both
versions
of
it,
but
it
can't
because
one
of
those
versions
it
doesn't
have
access
to
the
definition.
B
Yeah
OC
normally
just
shows
you
one
I
mean
I,
think
I
think
you
can
specify
a
version.
I
figured
I
figuring
out
how
to
do
the
CLI,
but
because
I
know
with
and
again,
if
you
have
a
resource
that
has
multiple
versions
defined.
One
of
those
is
the
storage
version.
That's
what's
actually
stored
at
natively,
and
then
you
know
the
cluster
can
convert
that
to
different
versions.
B
If
there's
multiple
ones
defined
for
to
use
it,
you
know
in
different
ones,
so
Valera
tries
to
use
all
those
in
the
backup
that
way
it
gives
it
maximum
flexibility
for
restoring
in
different
versions
of
kubernetes.
So
but
I
think
the
issue
here
is
that
there's
some
mismatch.
You
know
it's
been
defined,
maybe
maybe
one
of
the
versions
that
the
cluster
thinks
this
object
has
is
valid.
C
I
think
so
the
other
option
would
be
to
just
delete
the
V1
beta1
instance.
If
we
have
the
V1
instance.
B
Well,
the
thing
is:
if
it's
is
it
I
think
what's
going
on
here
is
the
is
the
matching
between
you
know,
there's
the
crd,
which
is
defined.
It
sounds
like
your
CD.
Crd
only
has
one
only
has
to
be
one,
not
the
one
beta
one,
and
then
you
have
your
actual
custom
resources
and
there's
only
one
CR
for
that
name.
It's
just
that
it
can
be
retrieved
in
multiple
versions.
B
You
know
you,
you
can
say
hey
give
me
the
you
know
the
the
B1
beta
one
version
of
this
give
me
the
V1
version
of
this,
and
if
all
those
versions
are
defined
in
the
crd,
you
can
access
that
Resource
as
multiple
versions,
so
I'm
thinking.
What
might
have
happened
here
is
that
at
one
point
in
the
past
you
had
a
crd
that
defined
both
versions.
B
B
But
you
know
it's,
but
the
cluster
knows
that
it
was
at
one
point
available
as
V1
beta1
and
Valero
they're,
using
the
the
discovery
code
Valero
uses,
for
you
know,
finding
all
the
resources
of
resource
versions
for
a
given
resource
it.
It's
still
seeing
B1
beta1
is,
should
be
an
option
and
then
it's
failing
to
find
it.
So.
B
That
and
again
that's
just
a
guess:
I
I'd
have
to
see
it
in
the
cluster.
That's
in
reproducing
it
to
confirm
that,
if
that's
true
I
think
the
fix
probably
is
to
look
at
is
to
modify
that
Valero
code
so
that,
for
this
specific
error
of
hey
I
can't
back
it
up
as
D1
beta
one,
let's
just
log
out
as
a
warning
and
not
fail
the
backup,
because
we
still
have
the
V1
version
of
it.
C
Okay,
yes,
I
guess
first
thing:
so
you
mentioned
we,
we
have
an
option
in
OC
to
get
all
versions
I'll
check.
So
at
least
we'll
see
right.
There
is
another
resource
of
the
even
better
one
right.
B
And
again
you
get
all
versions,
I'm,
not
sure
whether
you'll
see
the
one
better
one
or
not,
since
it's
not
defined
I'm,
not
sure
how.
B
That's
that's
that
that
was
my
next
next
point.
Is
that
I
don't
know
whether
the
code
that
you
know
that
Osa
uses
for
retrieving
something
by
version
is
the
same.
It's
been
a
while,
since
I've
looked
at
the
Valero
multi-version
backup
code,
so
I
don't
remember
off
the
top
of
my
head.
What
Valera
uses
to
determine
the
list
of
group
versions
that
it
needs
to
back
something
up.
As
you
know,
once
we
can
confine
that
we
can
kind
of
figure
out.
B
Okay,
you
know,
let's
look
at
this
object
and
to
see
it
if,
if
the
cluster
thinks
that
this
is
both
V1
and
B1,
beta
1,
but
then
V1
beta
1
doesn't
exist
as
in
the
crd
there
there
may
be
some.
You
know
some
change.
We
need
to
make
to
the
Valero
code
where
it
determines
what
versions
you
know
if
things
are
valid.
C
A
Get
back
yeah
yeah.
Can
you
guys
structure
a
bit
this
outputs
here,
because
or
you
know,
if
someone
wants
to
join
the
discussion,
it
would
be
hard
to
understand
why
this
is
here
in
this
one
I
mean
more
meaningful
information.
These
locks
and
stuff
thanks.
D
Yeah,
this
is
an
issue
we
opened
last
week,
so
our
performance
and
scale
team
attacked
at
did
some
research
on
creating
backups
for
100
000
PVCs.
So
we
have
Emily
on
the
call,
hey
Emily.
Do
you
mind
giving
a
summary
of.
C
C
So
ashibam
was
mentioning
our
skill
and
performance
team
was
testing
with
backups
with
a
large
number
of
PVCs,
and
we
were
hitting
that
error
log.
You
can
see
that
I
pasted
at
the
top
of
the
issue
and
after
I
looked
into
it.
It
looks
like
it
was
basically
just
due
to
the
fact
that,
on
the
recreate
volume
snapshot
content
function,
there
is
a
default
timeout
for
that
is
currently
set
to
one
minute.
So
we
tested
this
by
increasing
the
time
to
five
minutes.
C
It
got
further
along
in
the
backup
process
before
failing
and
then
with
a
10
minute
timeout.
It
eventually
worked-
and
this
was
tested
several
times
just
to
excuse
me-
make
sure
so
I
think
what
we
concluded
for.
What
needs
to
happen
with
this
issue
is
to
make
this
timeout
configurable
create
a
new
value
for
it.
C
D
Yeah,
so
basically,
we
just
wanted
an
opinion
whether
we
should
go
ahead
with
making
this
configurable
via
some
API
value.
Or
do
we
want
to
reuse
any
spec
that
we
already
already
have
for
CSI
snapshot,
timeout.
C
Yeah
I
can
I
say
that
or
I
understand
that
so
I
think
that
may
be
the
best
path
to
take.
B
This
is
this
is
at
least
the
second
case
of
a
hard
credit
timeout
that
we've
seen
you
know
causing
bugs
for
us
actually
do
we
ever
get
the
other
one
resolved
I
think
we
I
think
we
heard
I
think
we'll
let
the
other
one
by
the
timeout.
We
just
moved
to
a
five
minute:
hard
go
to
timeout
yeah.
D
B
I
I
mean
I
I.
Think
the
the
general
Point
here
is
that
so
we
we
have
identified
at
least
two
places
where
these
hard-coded
timeouts
you
know,
worked
great
for
tests
in
low
volume
environments,
but
with
large
backups
and
busy
clusters
some
suddenly
stopped
working
on
us.
It
might
be
worth
spending
some
time
trying
to
identify
all
of
the
places
where
we
have
timeouts
to
find.
You
know
which
ones
are
configurable,
which
ones
are
not
configurable
and
for
those
that
are
not
configurable.
B
A
We
can
create
a
book
bash
for
this
one
if
you
want,
if
you
want
to
play
a
game
on
this
one
and
some
rewards
on
this
one
I'm
kidding,
of
course,
but
that
that's
a
volume
point
to
make
it
more
structured
like
a
general
issue
and
then
link
tissues
to
this
one
or
right.
B
I
mean
I
mean
we
had
these
two
specific
issues
where
we
already
identified
hey.
This
is
a
bug
for
us.
We
need
to
make
this
specific
one
configurable,
but
right
the
general
issue
is
hey.
We
should
really
at
some
point,
spend
some
time
you
know
going
over
the
code
base
and
identifying.
Let's
look
at
all
the
places
where
we're
doing
you
know,
Polo,
mediate
or
whatever
check.
What
that,
what
value
we're
using
for
the
timeout
and
figuring
out
does
that
value
come
from
something
hard-coded
like
it
is
here
or
does
that
value
come
from?
B
You
know
some
configuration
file,
you
know
and
kind
of
go
from
there,
because
you
know
if
we
can
identify
something.
You
know
something
else
where
we're
waiting.
One
minute,
you
know,
maybe
we
say
hey,
let's
fix
this
before
you
know
we
have
a
customer
who's
failing
their
backups
schedules.
You
know
over
it.
C
C
B
Know
so
far
the
cases
we've
had
are,
you
know
relatively
unrelated
to
each
other
in
the
conditions
that
might
cause
you
to
you
know
there
might
be
some
cases
where
hey
we
need
to
increase
the
timeout,
because
we
have
a
huge
number
of
volumes.
There
may
be
other
cases
where
it's
some
other
load
related
thing,
or
maybe
the
scale
is
different.
B
You
know
in
one
case
the
timeout
needs
to
go
from
one
minute
to
five,
but
there
may
be
another
one
that
needs
to
go
from
10
to
30.,
so
those
are
cases
where
the
same
timeout
would
not
make
sense,
but
I
think
what
I
think
the
first
step
is
to
identify
the
cases
and
then
we
can
decide
whether
grouping
some
of
them
to
make
a
common
configuration
value,
makes
sense.
B
You
know
if
we,
if
we
can
add
three
new
configuration
values
instead
of
you,
know,
10.
That
would
probably
be
better
assuming
that
yeah
and
then
there
may
only
be
three
more
cases.
So
you
know
we
don't
right
now,
I,
don't
think,
there's
a
huge
number
of
these
in
the
code
base,
but
there's
certainly
probably
more
than
one
of
them
that
we
haven't
seen
yet
you
know,
but
you
know
how
we
handle,
that
is
going
to
depend
on
whether
we're
talking
about
you
know,
2
or
20
or
or
what.
A
B
That
I
mean
you're
right.
We
need,
we
certainly
need
to
false
values,
because
we
don't
expect,
you
know
users
to
suddenly
have
to
start
defining
them,
I
think.
In
most
cases
the
report
should
probably
be
what
it
currently
is,
but
in
some
cases
like
the
one
that
we
identified,
that
that
we
just
changed
in
192.
B
At
that
point
we
made
a
quick
change.
I,
don't
remember.
D
B
So
so
192,
for
example,
the
first
thing
we
did
was:
we
realized
that
the
the
default
of
one
minute
probably
was
too
small
to
be
the
default,
and
so
we
made
it
we
made
a
you
know
one
line,
change
to
to
update
that
default
to
35
minutes,
which
happened
to
solve
all
of
our
use
cases
for
it.
The
next
step
is
to
make
that
configurable,
so
that
if
someone
doesn't
want
to
do
one
five,
they
can
go
back
to
one,
but
but
the
point
is
I.
B
Think
in
all
these
cases,
there's
two
parts:
one
is
to
identify
the
hard-coded
timeouts
that
we
that
we
really
want
to
be
configurable
and
kind
of
I.
Think
a
secondary
issue
is,
you
know,
are
the
the
hard-coded
numbers
we're
using
now?
Are
those
same
values
to
to
be
the
default?
I
think
you
know
initially,
that's
that's
what
you
do.
You
know
you'd
make
the
default
whatever
it
is
the
hard
code
value,
but
then
we
can
also
have
the
opportunity
when
we're
making
the
change
to
say
this
particular
one
doesn't
make
sense
as
a
default.
B
It's
too
small,
it's
too
big
whatever,
and
we
can
change
it,
but
at
a
minimum
we
would.
We
would
want
to
make
these
things
configurable
with
a
default
value.
The
same
as
the
current
behavior
that
way,
a
user
who's
not
making
any
changes
in
the
functionality
has
not
changed.
They
don't
have
to
specify
any
of
these
things.
B
You
know
every
time
out
here.
It
needs
to
default,
obviously,
because,
especially
if
you're
just
starting
out
or
not
going
to
know
what
they
need,
they're
just
going
to
assume
that
the
defaults
are.
You
know.
C
B
That
are
reasonable
for,
for
an
average
you
know
customer
average
user,
and
you
know
you
know
that
if
I'm,
if
you're
backing
up
thousands
of
volumes
or
thousands
of
something
else,
that's
when
you
know
going
into
this
you're,
probably
going
to
want
to
update
some
of
the
faults.
A
Nope
I
had
a
message
from
excuse
me
if
I'm
not
pronouncing
the
name
nikhil
to
discuss
this
one.
D
Yeah,
so
just
I
wanted
to
check
regarding
so
previously
we
were
having
version
1.6,
which
was
fine.
So
recently
we
use
the
latest
versions
like
Upstream
of
1.9.2
as
well.
So
so
we
are
not
facing
any
such
errors
in
the
logs,
but
our
pod
volume
backups
are
getting
completed
with
the
error,
as
so
which
I
have
specified
here,
wrong,
password
and
and
our
backups
are
getting
partially
failed.
D
So
so
I'm
just
taking
help
on
this,
but
yeah
I
don't
see
much
any
updates
on
this,
so
just
wanted
to
check.
So
if
someone
has
any
idea
on
this.
A
D
D
So
are
there
no
errors,
investor
clocks,
no
I,
don't
see
any
I'm.
B
Just
looking
at
the
issue,
it
looks
like
I
I
see
the
The
Rustic
error
is
unable
to
read:
root
certificates,
the
search
value
directory.
D
B
So
I
mean
it
looks
like
this
is
having
a
problem
either
accessing
the
volume
or
something
I'm,
not
sure,
because
it's
just
saying
it's
unable
to
read
the
certificates,
the
Tim
CA
cert,
whatever
no
search
value
directory,
so
I,
don't
know
if
it's
not
there
or
if
there's
file
access
permission,
problems.
D
B
Because
I
see
they
were.
A
B
I'm,
just
like
yeah
I
was
just
sure.
I
was
just
looking
at
the
the
issue
that
was
linked,
that
you
linked.
D
So
if
you
mind
I
can
just
share
my
screen
and
show
you
so
I
have
just
entire
thing,
which
showed
the
logs
so
recently
I
just.
A
D
So
so
I've
just
did
this
backup
I
just
done
and
it's
the
status
has
been
like
partially
failed,
and
here
we
can
see
like
everything
the
restic
backups
has
been
completed.
So
yeah.
B
D
B
B
B
If
you
just
search
for
level
equals
error
that
should
because
otherwise
you're
going
to
get
debug
logs,
because
you
can
see
very
line
here-
you
see
level
equals
debug
at
the
beginning
and
we're
looking
for
level
equals
errors
yeah.
So.
B
B
Yeah
yeah
yeah
yeah
yeah.
We
need
to
look
at
the
backup
logs,
otherwise
you're
gonna
see
other
errors
that
they're
gonna
get
in
the
way,
because
we
we
want
to
look
at
just
there,
the
the
logs
for
this
backup.
So
you
can
do
Valera
backup
logs
in
the
backup
name.
B
So
that's
so
that's
so
so
this
is
probably
the
air
again
because
yeah,
it's
saying,
okay,
so
wrong,
password
or
no
key
found.
So.
D
Body
Yeah,
so
basically,
this
is
the
normal,
my
developer
mode,
which
I
have
only
one
rested
board,
and
now,
okay.
B
D
Because
if
we
check
this
pod
volume
back
up
for
status,
so
here
it
is
showing
the
faith
has
been
completed
and
progress
is
also
completed
like
by
its
turn
total
by
itself.
But
this
is
the
message
it
is
shown
right.
Yeah.
B
Yeah
that
that's
what
I
mean
I
mean
because
when
you're
looking
at
rustic
logs
that
succeed,
I
think
The
Rustic
command
line,
the
rest
of
CLI
succeeded
and
then
I
suspect
that
error.
You,
you
that
we
we
saw
was
happened
when
trying
to
upload
that
rustic
backup
to
the
backup,
storage
location,
though
I'm
not
100.
Sure,
because.
D
B
D
Versions,
yeah
1.6
we
have
currently,
we
have.
D
D
Maybe
like
just
as
an
approach
which
can
follow
more
too
like
troubleshoot.
This
issue
was
also
like
another
thing
and
just
suggest
me.
B
Yeah,
do
you
know
if
any,
are
there
any
any
other
changes
around
I'm?
Just
thinking?
If
this
is
a
problem
and
uploading
the
test
for
free,
because
I
mean
the
backup
itself
is
succeeding?
Otherwise,
so
that's
you
know
if,
if
the
backup
search
location
was
not
working,
you'd
think
it
would
fail
at
that
level.
B
B
Code
to
see
what
conditions
cause
you
know
lead
to
that.
B
Maybe
an
issue
with
the
rest
of
the
repository
itself
looks
like
it's
in
the
it's
in
that.
D
So
it's
no
I
guess
I'm
using
version
1.3
of
SD
since
yeah.
D
And
not
really
I
am
I'm
I'm
checking
form
from
1.9
itself,
so
I
thought
of
discussing
this.
Then
I
was
thinking
of
like
with
1.7
or
1.8
Maybe.
B
Yeah
and
the
other
thing
I
know,
do
you
remember
when
we
were
having
some
rustic
repository
issues?
What
what
those
look
like
I
mean
I,
don't
know
I'm
wondering
if
deleting
the
master
repository
and
trying
again
would
might
resolve
this.
If
this
is
the
problem
with
the
rest
of
the
repository
because
I
know,
Valero
creates
that.
D
Yeah,
like
I,
think
this
is
maybe
during
the
resting
init
process.
It's
failing.
B
D
Restrict
data
in
your
bucket
yeah,
it's
yeah,
so
basically
I
have
tested
that
so
yeah,
it's
creating
the
restic
folder
in
Pocket
itself:
okay,
okay,
yeah,
some
fault
as
a
config
and
case,
but.
B
The
other,
this
error
message
is
from
rustic
when
it's
trying
to
decrypt
the
repository
Keys
it
looks
like
but
I,
don't
know
what
could
lead
to
that.
I've
never
seen
this
particular
error
myself.
D
D
B
B
Do
you
get
the
same
error
message
if
you
try
a
second
backup
did
this?
Did
this
backup
failure
happen
right
like
when
you
first
created
the
I
was
looking
on
kind
of
searching
for
any
other
Valero
issues
that
had
this
mirror
message
and
I
found
one
from
several
years
ago,
where,
like
the
first
backup,
would
fail
with
that
message
and
then
after
that
would
succeed,
there
was
some
concern
that
the
rest
of
repository
was
taking
longer
to
be
initialized,
and
maybe
the
weight
code
was
not.
You
know
waiting
properly
for
it.
D
Yes,
not
basically
so
I
tried
testing
that
as
well,
because
I
was
thinking
like
maybe
Valero
is
taking
before
initializing
the
repositories
trying
to
take
the
snapshot
so,
but
the
community
guys
say
he
replied
that
it's
the
correct
like
way
how
it
works.
Basically,.
B
D
B
Just
I
just
wonder
if
you
tried
to
take
a
second
back
up
after
the
first,
the
you
know
for
the
the
same
names
or
whatever
did.
A
B
B
Maybe
that's
guess
what
it's
still
not
clear
is
what
might
be
different
about
your
environment.
So
I
was
just
trying
to
look
at
the
rest
of
Upstream
issues
to
see.
C
B
B
It
might
help
to
completely
empty.
You
know,
you
know
basically
delete
everything
in
the
bucket
and
start
over.
It
may
be
that
there's
something
about
the
way
Valero
Works
in
an
older
version
that
was
different
in
terms
of
what
still
got
stored,
and
so
maybe
there's
some
mismatch
there
between
some
old
obsolete
files
or
directories,
and
if
you
started
with
a
completely
fresh
bucket,
maybe
the
problem
goes
away.
Maybe
it
doesn't,
but
at
least
that.
D
Eliminates
the
possibility
yeah
so
so
this
is
like
our
developer
node.
So
we
are
installing
everything
from
fresh
air
like
so
we
are
also
testing
on
different
clusters,
but
this
is
our
fresh
cluster,
where
you
know
I'm
just
doing.
B
It
right,
but
it's
the
bucket
for
your
backup
storage
location,
have
old
stuff
in
it,
or
was
that
also
completely
new
when
you,
when
you
tested
this.
D
B
D
So
we
are
expecting
all
the
data
it's
on
inside
this
backup
way
to
folder.
So
here
we
are,
we
are
having
backup
folder
resting,
folder
and
yeah
restore.
So
that's.
C
B
Although
I
guess,
although
actually
I
guess
if
you're
using
ODP,
then
probably
not,
we
didn't
have
version
one
eight.
So,
but
if
you
go
to
like
the
ODB
one
zero
which
we'll
use
one,
you
know
Blair
one
seven,
but
then
again
that
might
be
another
point
for
comparison.
B
D
B
Sure,
okay,
so
I
guess
that's
everything
we'll
go
ahead
and
in
the
meeting
now
and
you
just
let
us
know
if
anything,
changes
with
a
different
version.