►
Description
Sept 7, 2020
Status Updates
carlisia
RC this week
On community support
Nolan is on PTO
Working on the download request migration
ashish
1.5 release testing
restore hooks PR:
testing
PR 2804 reviews
getting that merged hopefully today.
bridget
Reviewing PRs for the 1.5 release
Discussion Topics
Phuong: Timeout for plugins. For app consistent backup, the pod will be quiesced while snapshot being taken. If the PVC plugin taking the snapshot and that plugin take a long time (say because it fails and retries), the application will be blocked for a long time which is unwanted behavior.
Contributor Shoutouts
@jenting: fix EnableAPIGroupersions output log format
@pawanpraka1: fix: rename the PV if VolumeSnapshotter has modified the PV name
@arminc: [velero] Adding imagePullSecrets to the chart
A
Welcome
everyone
to
the
valero
community
meeting,
slash
open
discussion
today
is
tuesday
september
8th.
It
is
1202
pm
here
on
the
east
coast
good
morning
to
all
your
all.
You
west
coasters
hope
you
all
had
a
lovely
weekend,
some
of
us
here
in
the
us.
Of
course,
we
we
had
a
long
weekend
with
labor
day
yesterday
and
with
that
I'm
gonna,
I'm
gonna
kick
off
with
some
status
updates,
kalisia.
B
B
A
C
B
B
B
C
B
D
C
You
so
we
we
can
talk
about
it
now
or
later
it
doesn't
really
matter,
but
now,
for
basically
there
are.
We
don't
have
like
a
e2e
test
that
we
run
it's
basically
coming
up
with
scenarios
and
making
sure
the
changes,
the
new
new
features
are
tested
properly
and
also
there
is
no
regression,
so
that's
pretty
much
what
we
do,
but
I
I
can
share
more
afterwards
or
offline.
A
A
E
E
E
It
is
not
working
without
a
hitch.
If
we
we
have
some
problem
that
we
go
through,
but
most
of
it
related
to
the
fact
that
we
have
to
update
the
kubernetes,
try
and
go
to
100.18.4,
and
I
think
I
remember
nolan
and
ashish
have
helped
me
get
over
that
him.
That
content
is
it
work
now.
So
so
that's
number
one
number
two
last
week
I
have
an
action
item
to
create
the
task
or
bug
for
the
backing
up
thing
in
parallel.
I
have
done
that.
E
I
and
I
post
my
update
on
the
on
the
action
item
as
well,
and
so
this
week
I
come
with
another
issue
that
I
bump
into.
So,
basically,
when
I
do
backup
app,
consistent,
backup,
right
and
then
every
time
I
want
to
when
I
consistent
backup
for
my
application-
let's
just
say
my
sql
right,
then
I
have
to
create
that
my
mysql
application.
E
Then
I
run
then
I
I
would
back
up
the
pvc
by
creating
snapshot
of
that
pvc.
It
happened
that
we
used
valero
to
do
all
of
this,
and
the
way
we
do
is
we
specify
the
little
plugin
to
create
the
vmware
snapshot
and
the
vmware
snapshot.
If
it
fail,
it
will
be
dry,
it
will
be
fine
and
during
that
time
the
whole
application
being
quit
is
locked.
E
E
You
know
consequences,
for
example,
so
we
want
to
have
ability
to
specify
a
timeout
for
each
of
the
plug-in
that
we
want,
for
example,
if
you
want
to,
if
you
have
logged
in
a
timeout
for
backing
up
pvc
or
tv,
I
just
want
to
say
just
have
to
get
it
done
within
10
seconds
or
5
seconds
or
whatever
right,
if
it
doesn't
execute
it
during
that
time
we
have
to
avoid
it.
We
have
to
release
the
customer
from
continue,
you
know
with
their
operation
and
we
will
try
again
later
time.
E
F
Yeah
long
term
is
probably
a
good
idea,
and
we
also
have
to
think
about
that
in
terms
of
running
parallel
snapshots
like
how
long
do
we
wait
for
everybody
to
to
actually
snapshot?
Why
so,
the
the
snapshots,
the
v-series
snapshots,
are
actually
crash
consistent
and
so,
unless
so,
you're
actually
like
flushing
all
the
buffers
and
everything
as
part
of
your
app
quiesce,
you
know.
E
F
C
C
Like
trying
to
pause
the
database
or
whatever
application,
is
writing
to
the
database
or
performing
create
operations,
they
want
that
application
to
stop.
While
the
snapshot
is
happening.
If
the
snapshot
takes
forever
for
a
long
runs
for
a
long
time,
then
that
means
their
application
is
blocked
or
stopped
for
a
long
time
and
which
is
not
desirable,
and
what
phong
is
asking
is
having
a
timeout
for
this
lock
operation
and
if
the
operation
does
time
out,
and
then
we
just
call
the
unquest
I'll
have
to
look
at
the
code
to
see.
E
E
F
C
E
Probably
probably
it
more
than
it's
just
more
than
just
the
snapshot
fail.
It's
probably
you
know,
because
the
snapshot
be
actually
being
taken
on
the
v
sphere.
No,
not
where
we
execute
the
the
the
pvc
plugin.
So
it
might
be
some
network
communication
issue
there
and
so
on
and
so
forth.
So
I
don't
know
for
sure
what's
going
on,
but
I
know
for
sure
that
we
experienced
some
problems
with
that
that
we've
been
holding
for
a
long
time,
and
we
don't
want
that
behavior
for
our
customer.
F
E
A
G
Today,
I
don't
know
if
I
really
have
a
topic.
I
just
want
to
say:
hi
my
name's
chris,
just
joining
in
first
time,
seeing
one
of
these
meetups
just
started
using
this
product
and
kind
of
wanted
to
see
how
this
all
went,
and
I'm
almost
done
actually
implementing
this
for
a
very
large
cluster
for
a
client
just
running
the
pot
affinity
issues,
if
you
guys
have
seen
those
things
where
maybe
I'm
just
not
using
kubernetes
in
the
right
way,
where
I'm
having
dynamically
created
volumes
that
are
the
snapshots
and
everything's
great.
G
But
then
what
I'm
doing
restores
the
pot
affinity
we're
using
the
label.
The
kubernetes
label
gets
applied
to
it
where
it's
saying,
hey
you're
in
this
region,
so
the
pod
affinity
for
that
volume
is
saying,
let's
say,
north
central
u.s
and
then,
when
I
try
to
restore
it
in
south
central
us,
it's
saying:
hey
we're
having
volume,
pod,
ident
affinity
issues,
because
you
are
saying
during
the
restore
is
grabbing
the
the
persistent
volume
and
saying
hey.
You
have
to
be
within
north
central
u.s.
G
You
cannot
be
in
a
south
central
u.s
region
and
then,
when
it
tries
to
restore
that
volume
that
disc
to
that
specific
volume,
it's
it's
failing.
I
can't
get
the
thing
to
restore
ever
in
a
different
region,
so
I
don't
know
if
that's
a
specific
thing
specified
within
bolero
that
I'm
missing
or
if
it's
something
I've
missed
within
my
cluster
setup,
because
I
feel
like
this
should
be
supported.
Where
you
know
I'm
doing
disaster
recovery,
I'm
in
any
different
region,
then
I'm
essentially
failing
over
from.
G
C
G
So
it's
region,
it's
not
specifically
a
zone,
it's
not
really
a
snapshot,
I'm
able
to
bring
the
snapshot
over.
So
I
have
a.
I
have
two
clusters,
a
cluster
in
south
central
and
one
in
north
central,
the
the
north
central
is
the
main
cluster
and
that's
where
all
the
volumes
get
generated,
I'm
using
helm
to
generate
and
elastic
and
using
my
own
storage
class.
That's
a
premium.
G
You
know,
storage
class
within
azure
and
there
are
limitations
within
azure
and
aws.
You
can't
create
a
snapshot
outside
of
your
region,
but
the
problem
is
it's
restoring
the
snapshot
in
the
correct
region,
but
then
it's
restoring
from
the
con,
the
essentially
the
backup
of
the
cluster
and
within
that
the
volume
within
the
specifications
within
the
volume
that's
being
restored.
G
There's
a
pod
affinity,
section
that
says:
okay,
it
has
to
have
this
specific
selector
within
it
and
that's
the
north
central
region
and
that's
coming
from
the
label.
The
the
label
within
the
the
pod
and
the
pot,
the
node
I
mean,
and
the
node
will
have
a
label
of
south
central
us
in
it.
So
when
it
tries
to
associate
the
volume
back
to
all
my
my
data
nodes
within
manga,
it
says
I
can't
associate
this
this
volume
that
I've,
the
backup
volume
that's
been
generated
from
the
restore.
E
G
I
threw
an
issue
out
there
on
on
git
and
I
hope
that's
not
a
bad
spot.
To
put
that,
I
don't
know
if
I've
set
something
up
incorrectly,
but
it's
the
most
recent
one
that's
out
there
and
I
don't
know
if
you
guys
have
seen
that
or
if
there's
something
built
in
for
valero
to
be
like.
Okay,
I'm
moving
regions.
I
have
a
pod
affinity
set
in
place
within
this
volume
claim.
G
G
G
Okay,
that'd
be
great
and
yeah,
I'm
I'm
definitely
monitoring
it
and
I
can
take
time
out
of
my
day
and
sit
down
and
go
through
it.
I'm
about
to
the
point
where
I'm
just
gonna
start
going
through
the
code
if
it's
within
bolero
or
if
it's
within
my
system,
because
I've
been
trying
to
jam
this
thing
through
for
a
while.