►
From YouTube: Velero Community Meeting - Sept 3, 2019
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
A
All
right
so
before
we
get
started
just
a
reminder
that
this
meeting
is
being
recorded,
we'll
post
the
recording
up
on
to
our
YouTube
channel
later
on
today
or
later
this
week.
So
just
please
keep
that
in
mind
as
you're
commenting
so
today
not
much
in
the
agenda.
We
would
like
to
just
do
a
quick
review
of
kind
of
what
we've
been
working
on
and
and
in
particular
reviewing.
A
What's
in
the
1.2
release,
scope
and
talked
a
little
bit
about
what
we've
been
doing
there
and
what's
coming
up
there,
and
then
we
do
have
several
issues
and
PRS
that
we
wanted
to
open
up
an
opportunity
for
some
discussion
on.
There
are
some
different
feature:
requests
and
ideas
that
we
need
to
collect
more
input
on
so
we'll
pull
those
up
and
see
if
anyone
has
any
input
or
any
opinions
on
those.
A
So
I
guess
I'll
start
just
by
pulling
up
the
Zen
hub
board,
which
we
used
for
planning
our
milestones
and
I
actually
have
it
up
in
this
tab
here.
So
right
now,
I
have
it
filtered
to
the
1.2
milestone,
and
so,
if
you're
interested
to
see
what
we're
currently
working
on
and
kind
of,
what's
in
scope
for
this
release,
you
can
take
a
look
at
this
view.
A
A
Yet
whether
I
will
be
able
to
declare
it
GA
at
the
end
of
1.2
or
whether
that
will
push
to
1.3
but
I
think
we're
definitely
continuing
to
make
some
significant
improvements
to
the
stability
and
usability,
and
you
can
dive
into
some
of
these
issues
that
are
labeled,
rustic
or
rustic
GA
for
more
details
on
those
we're
also
continuing
to
work
on
our
CSI
integration.
Maybe
Nolan
can
give
a
little
bit
of
an
update
on
that
in
a
minute,
but
we've
been
working
on
kind
of
iterating
on
our
design
for
how
we
want
to
integrate.
A
So
we
may
be
in
a
little
bit
of
a
holding
pattern
here,
while
that
sort
of
settles
but
definitely
still
open
to
input
here
on
the
design
proposal
that
we
have
up
beyond
that,
we
have
a
number
of
sort
of
housekeeping
items,
so
we
will
be
moving
the
velaro
github
repository
out
of
the
heavy
OU
organization
and
into
a
VMware
organization.
As
part
of
this
release
cycle.
It's
not.
A
We
want
to
do
this
so
that
they're
sort
of
on
equal
footing
with
any
external
plugins
that
are
contributed.
It
also
helps
us
make
sure
that
the
kind
of
the
development
experience
for
those
buggins
is
good
and
that
you
know
if
there
are
any
improvements
that
need
to
be
made,
that
we
can.
You
can
see
it
kind
of
in
the
same
way
as
any
external
contributors,
and
it
also
allows
those
things
to
kind
of
evolve
and
be
released
potentially
on
their
own
cycles.
A
A
C
Make
sure
we
have
to
make
sure
that
we're
not
doing
extra
work
that
doesn't
need
to
be
done
also
on
the
CSI
snapshot
ap?
Is
they
got
delayed
due
to
so
late
discussions
last
week
on
modifying
some
of
the
the
api's
that
are
involved
with
taking
snapshots
and
specifically
the
data
source
for
PVC,
and
those
discussions
ended
up
making
it
so
the
team
missed
the
deadline
for
moving
things.
A
Well
thanks!
Yes,
so
will
you
know
we'll
be
continuing
to
kind
of
keep
informed
of?
What's
going
on
upstream
see
if
there's
you
know
any
input
that
we
can
provide
that'll
be
helpful
there
and
hope
to
be
able
to
move
this
forward
over
the
the
coming
months.
So
I'll
switch
back
to
the
agenda.
The
did
anyone
else
have
any
other
specific
items.
I
wanted
to
cover
about
the
1.2,
your
release.
C
A
A
A
A
You
know,
for
example,
if
you're
running
on
AWS,
this
is
kind
of
a
sufficient
workflow,
because
when
you
create
a
snapshot
using
the
EBS
API,
you
get
a
point
in
time,
snapshot
that's
created
and
then
behind
the
scenes,
AWS
kind
of
replicates
the
data
that
existed
at
that
point
in
time
off
to
a
durable
back
end.
So
they
replicate
the
data
off
to
s3,
but
that's
all
kind
of
transparent
to
the
user
or
the
API
caller.
A
D
A
That
could
be
in
the
cloud
or
some
other
location
within
your
data
center,
and
so
this
this
design
proposal
basically
walks
through
enhancing
the
Valero
plug-in
API
to
add
a
separate
API
called
for
uploading
snapshot
data
into
that
that
durable
back-end
and
we've
we've
started
to
have
some
discussion
on
this
PR.
So
some
of
the
folks
on
the
team
have
started
to
add
some
comments
to
it,
but
we're
we're
definitely
interested
in.
You
know
making
sure
that
any
enhancements
that
we
put
in
here
make
sense
for
all
of
the
storage
providers.
A
E
I've
got
a
comment
based
on
my
experience
of
backup,
which
goes
back
a
long
time.
Let's
say
this
need
for
a
two-stage:
backup
has
been
perceived
as
essential
going
all
the
way
back
to
Microsoft's
volume,
shadow
service,
which
is
over
ten
years
ago,
I
actually
questioned
whether
using
the
ABS
API
is
adequate.
I
mean
it
gives
you
what
some
might
call
a
crash,
consistent
backup
equivalent
to
pulling
the
power
cord
out
of
a
wall
on
a
server.
E
But
there
are
many
stateful
apps
that
cache
they
in
potentially
memory
and
things
like
that,
and
they
don't
get
an
opportunity
to
flush
before
you
make
the
backup.
Also
any
stateful
lap
that
is
clustered
meaning
multiple
nodes
are
involved.
Typically
is
could
have
them.
You
could
have
a
real
mess
on
your
hands
with
something
like
this.
E
What
you
want
to
do
with
that
two
stage
is
inform
the
app
that
hey,
you
should
flush
all
your
state
down
to
the
persistent
disk,
ASAP
and
then
brownout
service
on
any
new
transactions,
because
you
don't
want
transactions
that
are
half-finished
you
know,
update
an
index
table
and
not
what
it
points
to
or
something.
So
what
they'll
do
is
they'll
have
inbound
API
calls
and
they'll
sit
there.
E
E
The
other
thing
that
this
two-step
call
is
useful
for
is
something
called
log
truncation,
where
a
lot
of
these
things
that
tolerate
yank,
the
power
cord
out,
build
up
a
transaction
log,
but
it
grows
to
infinity
and
you,
in
other
words,
there's
nothing
that
ever
garbage
collects
it,
and
it's
typical
for
backup
solutions
that
have
this
two-stage
to
use
that
you
sometimes
also
want
a
signal
back
to
the
app
that.
Yes,
we
made
the
snapshot.
A
A
F
A
C
Yeah
and
I
I
think
long
term.
We
need
to
look
at
that,
but
I
haven't
taken
the
time
to
look
into
it
because
I
don't
think
that
caps
accepted.
Yet
everything
might
be
closed.
Maybe
it
is,
but
I
think
it's
still
saying
that
it's
still
sitting
okay
but
yeah
you're
you're
right
that
there's
a
more
generic
way
out
there
that
we
should
you.
E
A
Yeah
that
makes
sense
and
I
think
we
we
you
know
long
term.
We
we
would
like
to
integrate
with
that.
If
there
is
a
standard
way
of
doing
it
upstream
or
you
know,
potentially
help
help
make
that
happen
and
I
don't
see
any
reason
why
we
couldn't,
you
know
swap
out
our
existing
exec
book
code
for
kind
of
the
equivalent
upstream
approach
to
it.
We
do
want
to
make
sure
that
you
know
where
we
don't
want
to
only
support
CSI,
at
least
in
the
in
the
near-term.
Until
that's
you.
B
And
so
I
know
that
the
open
abs
guys
have
been
really
interested
in
this
for
a
while.
Now
I
have
any
of
the
other
storage
providers
taking
a
look
at
this
I'm
thinking,
specifically
about
our
own
beefs
and
folks,
as
well
as
as
well
as
like
the
folks
in
Deltek
and
the
power
protect,
because
I
think
they
both
have
voiced
desire
for
this.
This
upload
signal.
A
B
I
think
I'll
ping
a
couple
of
folks
offline
on
this
as
well
to
draw
their
attention
to
it,
because
this
is
work
that
they're,
also
interested
in
and
I
think
that
if
we
can
get
a
good
deal
of
feedback
from
not
only
the
open
source
community
but
also
some
of
the
some
of
the
proprietary
storage
vendors
I
think
that
that
would
be.
That
would
be
helpful
to
make
sure
we
have
a
good,
well-rounded
solution,
not.
E
Just
I
think
there
may
be
potential,
not
just
for
them,
contributing
to
shall
we
say
the
spec,
but
also
given
that
both
of
those
people
have
working
backup
solutions
coming
out
of
the
VM
world.
They
have
developers
who
are
experienced
with
this
I
know.
The
Dell
EMC
has
an
inventory
of
plug-ins
for
a
lot
of
popular
stateful
apps,
and
they
might
be.
You
know.
The
more
interesting
thing
would
be
the
to
get
them
to
contribute
developer
hours
to
build
this
out.
E
I
did
talk
to
both
Dave
and
a
bunch
of
the
Dell
EMC
people
last
week
at
VMworld,
and
they
are
aware
of
this
and
interested
in
the
subject.
It
struck
me
that
the
VMware
storage
might
be
more
aligned
with
something
matching
up
to
CSI,
because
that's
clearly
the
way
full
going
forward
with
the
storage
support
and
it.
E
E
I
approached
the
possibility
that
perhaps
they
could
even
take
some
of
their
santa
clara
resources
and
host
a
face
to
face
meaning
to
get
people
to
physically
meet
and
interact
with
each
other,
because
we
used
to
do
that
over
in
the
scoober,
nettie
storage,
saying
and
I
think
it
was
a
good
way
to
break
the
ice
and
get
people
involved
on
a
project
anyway.
I
brought
that
up
and
they
seemed
to
you
know
they
didn't
reject
the
possibility,
they
didn't
say
sure,
we'll
provide
a
room
and
host
it
next
month.
E
A
Thanks
Anthony,
that
sounds
great.
You
know,
I
I,
don't
know
that
this
is
necessarily
something
that
we'll
be
able
to
get
consensus
on.
You
know
really
quickly,
but
it
may
make
sense,
depending
on
kind
of
how
many
folks
we
can
get
interested
in
it
and
set
up
a
little
bit
of
a
recurring
meeting
to
just
discuss
the
the
technical
design
and
what
the
workflows
look
like
for
some
of
the
different
providers
and
make
sure
that
we're
getting
to
a
design
that
makes
sense
overall.
A
A
A
A
So
the
kind
of
the
quick
background
on
this
is
that
when
velaro
restores
kubernetes
resources
prior
to
restoring
them,
it
actually
deletes
the
entire
status
field
from
those
resources.
So
any
of
the
the
subfields
that
exist
within
that
the
status
are
cleared
out,
and
so
essentially,
when
Valero
is
restoring
those
resources,
it's
just
taking
the
speck
of
the
object
and
applying
it
to
the
API,
and
there
are
kind
of
a
number
of
different
reasons
that
we
did
this,
but
at
least
for
in
turn,
for
for
the
upstream
api's
for
the
core,
kubernetes
ap
is.
A
H
Just
had
one
I
guess
love
a
question
instead
I'm
if
we
do
go
ahead
and
make
this
change,
is
this
something
that
would
be
an
optional
Cham,
just
wondering
like
if
we
have
plugins
already
that,
for
example,
were
written
under
the
assumption
that
status
was
not
restored
if
we're
suddenly
restoring
status.
That
could
change
behavior
for
existing
plugins.
A
B
Yeah
I
agree,
I
mean
it
well.
Of
course,
I
need
to
read
through
this
to
know
more,
but
I
always
think
of
the
status
fields
as
something
that's
a
runtime
attribute
and
not
necessarily
something
that's
supposed
to
be
restored,
but
rather
reconstituted
by
the
system.
So
I've
checked
to
look
at
like
what
specifically
is
being
proposed
here.
C
Yeah
and
I
agreed
like
it
seems
like
wanting
to
restore
the
status,
isn't
like
it
to
me.
It
seems
like
that's
an
anti-pattern
in
your
in
your
app,
but
it's
also
we've
got
a
couple:
people
that
it
that
don't
want
to
reorder
tech,
their
applications
just
to
get
it
backed
up
by
have
it
Valero,
which
I
think
is
fair
and
I.
Also
Scott
brought
up
a
good
point
that
like
if
we
just
switch
I,
think
that's
like
a
major
like
we
made
everything
globally,
restore
it.
D
A
H
H
I
learned
was
trying
to
write
a
plugin
was
you
know,
you
lose
status
and
you
know
we
have
some
of
for
openshift
our
image
stream
stores
a
lot
in
the
status
field,
and
that
was
actually
one
of
the
reasons
why
we
put
a
request
and
to
have
available
that
original
resource
with
status
field
not
to
back
up,
but
so
flooding
could
use
data
in
it
and
but
because
we
weren't
restoring
the
status
field.
You
know
our
plugin
uses
that
information
in
the
restore,
but
it
turns
out.
We
don't
really
need
that
in
the
restore.
H
So
it's
ok,
because
what
those
images
that
we're
restoring
again
this
the
cluster
is
gonna,
recreate
that
so
it's
kind
of
an
unknown
right
now,
if
we
did
restore
that
what
would
happen
because
one
of
these
things
were,
if
we
were
gonna,
add
support
for
this.
You
know
the
first
two
boys
should
be
just
leave
it
alone
for
our
plugins.
We
don't
want
it
because
we're
not
using
it
and
it
works,
but
it's
something
we
probably
look
into
once
it's
available,
you
know.
Ok,
does
this
help
us
out?
H
C
C
That's
like
a
boolean
that
controls
the
behavior
of
all
resources
covered
by
that
backup,
or
by
that
restore
it
may
be
a
actually
now
I'm
thinking
about
it
would
be
a
probably
be
a
restore
time
flag,
not
a
back
in
flag,
but
it
seems
to
me
like
it's
useful
for
people,
but
also
I,
think
we
need
to
talk
through
what
the
implementation
looks
like
to
get
a
read
on
like.
If
we
do
annotations,
what
do
we
annotate?
H
Step
in
my
mind,
hey
I
think
he
wanted
to
be
granular,
at
least
the
level
of
the
resource
or
the
CRT,
because
again
I'm,
just
taking
in
terms
of
you,
have
custom
plugins.
You
might
have
some
plugins
that
want
to
make
use
of
this
and
others
that
you
know
you
really
don't
want
to
mess
with
this.
You
want
to
leave
satis
empty
and
not
you
know
the
whole
restore
as
a
whole.
Everything
or
nothing
yeah.
C
E
Yeah
there's
two
use
cases
of
restoring
from
not
on
top
of
nothing
and
then
restoring
an
existing.
So
if
it
was
existing,
I
think
putting
the
annotation
in
the
existing
so
that
it
can
control
the
extent
of
potential
overwrite
would
make
sense.
But
if
it
was
empty,
then
a
less
granular
approach
makes
sense.
A
Yeah
I'm
definitely
interested.
You
know
and
I'll
comment
on
this
issue,
but
just
to
hear
from
the
couple
of
folks
who
have
requested
this
I'd
be
interested
just
to
see
some
of
the
specific
sort
of
fields
or
information
that's
stored
in
status
that
needs
to
be
restored,
I
just
I'd
like
to
and
those
details
before,
pursuing
it
because
I,
don't
I,
don't
know
how
you
know
widespread
this
need
is
the
original
issue.
Creator
talked
about
rancher
and
some
I
think
some
rancher
CR
DS,
and
then
we
had
another
comment
from
I.
A
A
So
this
is
an
issue
that
we've
had
in
the
backlog
for
a
while,
basically
around
being
able
to
use
bolero
with
self-signed
cert
and
potentially
provide
you
know
needing
to
provide
custom
CA
bundles,
as
as
connections
are
being
set
up.
More
recently,
we
had
a
PR
that
was
submitted
to
basically
add
an
insecure
skip,
verify
flag
that
would
allow
kind
of
TLS
verification
to
to
be
bypassed.
We
chose
not
to
the
the
PR
came
in
at
the
very
end
of
the
1.1
cycle.
A
A
Think
you
know
we.
We
definitely
want
to
have
a
solution
for
folks
who
are
in
this
situation.
It
seems
to
come
up
a
fair
amount
with
with
Mineo
endpoints,
and
so
you
know,
we'd
like
to
have
a
way
for
folks
to
be
able
to
to
have
a
good
workflow
here,
but
I
think
you
know,
the
concern
around
adding
the
insecure
Skip
verify
flag
is
is
that
it
will
end
up
being
sort
of
abused
or
used.
A
But
that's
you
know
that
places
additional
demands
on
the
user
as
far
as
configuration
and
potentially
has
input
implications
not
just
on
the
velaro
server,
but
also
on
the
clients.
So
if
we
do
go
down
that
path,
we
definitely
need
to
think
about
kind
of
what
the
UX
is
there
and
and
how
to
make
it
kind
of
a
seamless
process
for
users.
F
Yeah
to
me,
it
seems
like
we
might
just
want
to
do
both
and
just
recommend
that
people
use
the
custom
CA
option
over
the
insecure
scape
verify.
Maybe
just
have
like
a
warning
whenever
you
use
that
one,
but
I
wouldn't
necessarily
not
do
it
just
because
we're
scared
of
people
abusing
it
also
tools
like
give
CTO
and
and
I
was
also
support.
This
option
so
I
think
I.
Don't
think
it's
that
bad
to
necessarily
have
it.
A
Yeah,
it
would
definitely
yeah,
like
you
said,
like
cube
CTL
offers
it
as
an
option,
so
it
would
be
kind
of
consistent,
at
least
with
some
of
the
main
ecosystem
tools.
In
that
regard,
you
know,
I
do
think
we'd,
you
know
obvious.
We
can.
We
can
put
big
scary
warnings
in
the
in
the
help
text
for
the
flag.
I,
don't
know
if
they're
you
know,
if
we,
if
we
choose
to
include
that,
if
there
are
other
ways
that
we
can
make
sure
that
folks
are
aware
that
it's
an
insecure
option.
F
F
C
D
You
can
make
backups.
Whoever
has
the
permissions
make.
Backups
has
permissions
to
make
to
restore
things
and
I
don't
just
grow,
based
actions
that
the
little
is
allowing
and
I'm
just
trying
to
think
of
a
malicious
case.
Where
somebody
would,
you
know,
use
an
unsigned
cert
or
something
like
this,
to
backup
things
to
a
storage
and
be
able
to
just
restore
I,
don't
know
what
the
issues
would
be,
but
that's
the
sort
of
direction
that
I'm.
How
big
thinking
about.
A
Okay,
well,
it
sounds
like
yeah
we're
I'm
in
we're
good.
Obviously,
gonna
have
to
come
to
a
decision
here
on
on
which
approaches
we're
gonna
pursue.
Certainly
the
you
know
the
code
for
the
insecure
skip,
verify
flag
has
already
been
written
and
submitted
in
this
PR.
So
if
we
do
decide
to
accept
that,
you
know
we'll
just
need
to
make
sure
that
that's
that
code
looks
reasonable.
A
If
we
decide
not
to
accept
it,
then
I
think
we,
you
know
we'll
probably
want
to
put
forth
a
design
proposal
for
the
alternate
approach.
Make
sure
that
that
works
for
for
folks,
who
are
in
this
scenario,
so
yeah
again,
I
would
I
would
say
for
anyone
who
has
any
other
inputs
or
is
in
this
situation.
Please
please
add
comments
or
questions
on
this
issue
issue
1027
here
and
we'll
we'll
try
and
get
to
a
resolution
here
in
a
in
a
timely
manner.
A
A
We
just
closed
that
at
the
end
of
last
week,
and
so
we
will
be
working
on
kind
of
collating
the
responses
and
getting
those
in
a
format
where
we
can
publish
them
out
so
that
everyone
can
kind
of
see
the
results
of
those
we're
hoping
that
we
can
use
those
responses
to
help
them
form
our
roadmap
going
forward.
So
we're
we're,
definitely
excited
about
kind
of
analyzing
that
data
and
seeing
what
it
tells
us
so
stay
tuned
for
that
and
the
coming
weeks
and
months,
and
then
beyond
that
yeah
we're
we're.
A
Definitely
you
know,
excited
to
have
folks
continue
to
join
us
in
the
the
slack
channel
or
in
any
of
the
other
kind
of
community
channels.
So
if
you're,
if
you're
using
velaro
or
if
you're
new
to
Valero,
please
come
join
us,
you
know
ask
questions,
answer
questions
just
just
be
a
part
of
the
community.
They
are
know
and
that'll
help
us
continue
to
evolve.
H
Friday.
The
issue
was
when
you're
doing
a
restore
with
include
question
resources
set
to
auto
Justin
L.
My
expectation
was
that
you
know
he
had
a
CR
that
you
were
restoring
or
action
to
back
up.
If
you're
backing
up
a
CR,
the
appropriate
CRD
would
be
pulled
in
as
well.
I
know
if
you
include
close
to
resources
to
false
you,
don't
get
anything
and
I
can
set
it
to
true
you
get
all
of
them.
We've
going
into
the
code.
We
realized
that
there's
no
code
actually
written
to
do
that.
I
opened
an
issue.
H
A
Cool
yeah
I
was
a
PTO
on
Friday.
I
was
just
kind
of
catching
up
on
a
guided.
No
one
mentioned
it
to
me,
but
I
would
you
know
it
sounds
like
having
that
behavior
is
definitely
consistent
with
that
the
include
cluster
resources
flag.
Unfortunately
we
don't
have.
We
don't
really
have
a
generic
mechanism
for
identifying.
You
know
when
there's
an
Associated
cluster
scoped
resource
that
should
be
included.
Yeah.
H
So
v's
is
a
special
case,
and
so
my
PR
is
another
special
case
for
CR
DS
and
that,
like
you,
said,
I,
don't
know
of
any
way
to
doing
it
other
than
a
special
case.
So
you
know
have
a
look
like
you
know.
When
you
have
a
chance
some,
you
know
we'll
figure
it
out
whether
that
makes
sense
or
a
different
approach
is
better.
Okay,.
A
H
Slightly
related
to
this,
there
was
an
previously
existing
issue
where
and
of
CRD
is
stored
and
then
immediately
we
restore
CR,
but
the
co2
is
not
ready.
Yet,
as
soon
as
it
fails,
I
know
there
was
some
discussion.
I
think
this
is
supposed
to
go.
There
hasn't
been
any
movement
on
that
issue
in
awhile.
As
to
you
know,
longer-term
solutions
around
resource
I'm,
either
retrying
or
we
you
know.
I-
was
also
wondering,
if
would
seem
like
a
shorter
term,
and
a
more
targeted
solution
might
be
possible
if
we
had
a
notion
of
a
you
know.
H
After
restore
back
of
action,
there's
right
right
now,
the
restore
actions
ago
before
backup
I
know,
I
saw
an
issue
again
the
year
or
so
ago
of
you
know
the
possibility
of
having
a
plug-in
that
would
run
after
the
backup
all
right
after
after
the
restore
that
might
allow
this
to
work
for
making
sure
see.
Ids
were
ready,
yeah.
A
H
So
the
only
place
you'd
be
waiting
would
be
on
c
rd
restores
rather
than
you
know,
on
everything,
because
without
doing
that,
the
only
other
way
that
I
can
do
it
and
a
plug-in
would
be
to
have
a
plugin
that
that
you
know
runs
on
every
single
resource
restore
and
if
it's
a
custom
resource
wait
for
the
cid
and
that's
obviously
a
lot
more
waiting
and
places
that
we're
checking
and
the
other
thing
you
could
do
would
be
again.
A
treat
c
rd
is
a
special
case.
H
It's
handled
directly
in
the
code
and
at
some
point
we're
we're
restoring
those
cds.
You
know
actually
have
code
in
velaro
itself.
That's
waiting
for
that.
But
that's
that's
where
I
thought
some
of
that
longstanding
github
issue
was
talking
about
and
the
discussions
were
moving
more
in
the
direction
of
a
more
general.
You
know
linking
of
related
resources
and
all
that,
rather
than
something
more
targeted
to
just
see,
ids
yeah.
A
H
There
is
an
issue
I
saw
that
was
that
was
there
was
some
discussion
on
as
well.
It
looked
like
for
that
use
case
having
to
do
with
pod,
restore
and
databases.
I
got
the
impression
from
the
discussion
that
this
wasn't
really
the
right
solution
for
that,
but
in
a
more
general
case
of
something
like
the
restore
plugin
action
but
runs
after
restore
that
you
know,
went
because,
then
you
could
just
ride
up.
H
A
Okay,
yeah-
and
you
know
we
could
also
think
about
you-
know
having
kind
of
a
multi-phased
approach
here,
where
maybe
it's
it's
just
kind
of
expedient
to
implement
something
like
this
within
the
core
velaro
code.
For
now,
we
can
kind
of
make
sure
it
works
and
then
think
about
the
adding
post,
restore
plug-in
support
as
a
separate
thing
and
kind
of
have
a
plan
to
migrate
that
that
special
case
logic
in
there
you
know
with
adding
new
plugins.
It's
certainly
more
overhead
than
just
adding.
H
A
C
We
already
have
other
special
case
logic
around
PBS
and
PVCs
like
if
they're
terminating
and
you
try
to
make
it
back
up
or
I
forget
what
it
is.
But
there's
we've
got
like
a
check
for
if
they're
terminating
I
think
it's
on
restore.
So
if
the
terminating
wait
to
restore
them
and
it
like
walks
from
a
PvE
to
a
BBC
to
a
name
space
if
any
of
those
are
in
a
terminating
state,
so
yeah
we've
got
precedent
for
that
kind
of
logic.
It's
just
where
to
put
it
is.
H
Well,
I'm,
it
was
one
of
these
things
they
just
came
through
with
one
of
our
testers.
We
was
realizing
that
you
know.
If
you
had
me,
they
were
testing
a
case
with
an
operator
which
is
you
know
it
was
not
something
we're
actually
supporting
yet,
but
I
was
we
were
realizing
that
just
more
generally,
if
you
have
an
application
with
with
custom
resources,
we
realized
that
the
CRTs
weren't,
even
in
the
back
up
at
all,
and
so
that
was
what
precipitated
this
the
issue
that
I
put
in
and
then
as
I
was
testing.
H
My
attempted
solution.
I
realized
that
while
it
works
on
the
backup,
the
backup
included
it
when
it
needed
to
and
all
that
I
was
getting.
You
know
sporadic
restore
failures,
but
this
is
a
specialized
case
with
very
little
in
it.
And
it's
just
you
know:
here's
here's
a
resource,
here's
a
CR
D.
So
that's
whether
you
hit
that
race
condition
pretty
often
because
it's
not
else,
there's
not
a
whole
lot.
That's
doing
so.
It's
pretty
common
to
get
to
that
point
where
it's
not
ready.
H
So
you
know
I
hadn't,
gotten
any
further
than
that.
I
figured
I'd,
you
know,
get
the
CRT
in
the
backup
first
and
then
I'd
worry
about
the
next
step,
which
is
and
then
the
answer
may
be
something
similar.
You
know,
but
instead
of
because
the
the
PR
that
I
put
out
there
over
the
weekend
is
you
know,
on
the
back
on
the
back
up
side,
to
sort
of
pull
things
in
and
and
the
backup
resource
method,
and
it
would
probably
be
an
equivalent
spot.
A
Okay,
yeah,
that
makes
sense,
I
I,
think
we
at
you
know
we
have
at
a
couple
of
other
reports
of
this
issue
before
and
I
think
we've,
you
know
generally
recommended
at
least
as
a
workaround.
Just
you
know
kind
of
do
to
backup
or
to
restores
in
a
row
where
your
first
restore
is.
You
know
specifically
just
restoring
the
customer
resource
definitions
and
then
essentially
manually
wait
for
those
things
to
be
ready
and
then
trigger
another.
H
H
Gotten
to
the
point,
I
figure
out
whether
you
know
how
much
work
that
way
to
work
with
them,
with
our
controller,
to
do
that
versus
work,
to
get
something
in
velaro
directly,
that's
worth
because
and
then
tailed
it
until
they're
included
in
the
backup.
You
know
none
of
this
relevant
anyway,
so
that
was
the
first
step
is
to
get
those
Co
days
into
the
backups.
H
But
you
know
I'd
rather
avoid
the
multiple
passes
in
the
back,
because
we're
already
doing
two
to
stage
backup
and
restore
with
handling
PBS
versus
everything
else,
and
so
that
would
just
add
more
complexity.
On
our
end.
To
have
another
pass
just
to
handle
the
CR
d,
some
I
mean
it's
there's
nothing
that
would
prevent
us
from
doing
that.
Other
than
wanting
to
avoid
that
complexity.
If
we
could
yeah.
A
That
makes
total
sense.
We
can
yeah.
We
can
definitely
kind
of
pick
up
discussion
on
these
and
try
and
get
to
a
an
approach
and
then
figure
out
where
we
can
fit
it
in
or
see.
If
see,
if
you
know,
if
you
all
have
time
to
contribute
or
if
anyone
else
is
interested
in
contributing
here
sure,
okay.
Well,
thanks
for
bringing
that
up,
Bob.