►
From YouTube: 2020 11 24 Jenkins Infra Meeting
Description
Jenkins infrastructure meeting November 24, 2020
B
Awesome
so
welcome
everybody,
so
the
first
topic
to
the
agenda
is
regarding
the
azure
storage,
the
azure
file
storage
outage
we
had
last
week.
So
basically,
what
happened?
Is
the
service
automatically
recovered
by
itself?
Last
thursday,
I'm
still
waiting
for
information
from
microsoft
to
know
what
happened
so
basically,
what's
happening
now
on
our
side.
Is
the
service
was
relying
on
azure
file
storage
to
distribute
to
files
to
generate
checksum
for
the
files,
and
now
the
azure
file
storage
is
usable.
We
can
mount
it.
B
We
can
list
the
file.
We
can
do
all
the
things
that
we
need
to
do
so.
Typically,
the
official
location
is
now
running.
We
are
still
running
on
default
bike
right
now,
because
I'm
waiting
to
know
if
the
official
location
is
reliable
enough.
So
at
the
moment
we
are
waiting
for
microsoft
to
confirm-
or
at
least
to
tell
us
what
was
the
issue.
They
asked
us
seven
to
ten
days
to
investigate
in
order
to
understand
what
was
wrong.
So
that's!
That's
the
current
state,
nothing
moved
there
since
it's
working
correctly.
B
I
would
be
interested
to
move
to
redirect
the
traffic
from
the
fallback
situation
to
the
official
location
I'll
keep
the
fallback
situation
in
case.
Something
goes
wrong
again.
C
B
The
problem
is,
I
would
prefer
to
to
redirect
the
traffic
to
arc
in
this
cluster,
which
is
more
appropriate
to
handle
the
loads
than
the
seagan
machine
that
we
are
using
as
a
fallback.
So
there
is
in
the
notes,
a
dashboard
where
you
can
see
that
by
using
the
foldback
service,
it
increased
the
loads
on
updated
jenkins
that
I
own.
So
that's
why
I
would
be
interested
to
to
read
words
that
change
any
question
mark
on
this.
A
So
the
the
you
have
a
timeline.
I
would
love
for
us
to
get
back
to
using
the
official
the
official
production
scale
rather
than
a
fallback.
Are
you
envisioning
doing
that
over
the
next
few
days?
We've
got
a
good
time
right
now.
This
is
thursday,
and
friday
are
u.s
holidays,
so
the
load
will
probably
be
lighter
over
those
two
two
national
holidays
here
in
the
us
at
least
or
or
is
it
you're
just
trying
to
find
the
time
now?
B
I
think
it
would
be
nice
to
do
it
on
thursday.
I
don't
want
to
to
do
to
work
on
that,
so
we
just
we
just
released
a
new
weekly
release,
so
I
would
not
change
anything
neither
today
or
tomorrow,
but
thursday
sounds
like
a
really
good
time.
There
is
no
risk
technically
because
it's
just
a
dns
record
that
we
need
to
update
and
that's
it
so
I
can.
I
can
definitely
switch
the
dns
on
thursday.
A
B
A
And
then
we
get
the
benefit
that
next
week
we
will
do
a
weekly
release
on
tuesday
and
the
lts
on
wednesday.
So
by
the
next
weekly
release,
and
if
there's
a
problem,
we
can
flag
it
and
say
we
have
to
do
this
or
that
in
order
to
assure
that
the
lps
arrives.
Okay,
yes,
exactly!
Okay,
then
wednesday!
Next
week,.
A
Oh
speaking,
of
which
I
I
just
need
to
remind
myself,
I'm
going
to
play
this
action
separate
topic
here,
lps
respect.
B
B
The
idea
was
to
monitor
the
maven
repository
reporter
jenkinsci.org
to
know
what
is
the
latest
changes
version
that
you
can
download,
and
so
from
that
maven
repository
we
know
we
have
a
way
to
know
what
the
latest
lts
release
and
what's
the
latest
weekly,
and
so
the
idea
was
to
create
a
data
dog
custom
check
that
fetched
that
information
from
the
maven
repository
and
then
monitor
get
the
jenkins
that
I
got
to
know
if
we
published
if
we
publish
the
appropriate
just
about
the
distribution
packages,
I'm
happy
to
say
that
it's
not
working.
B
So
there
is,
I
also
put
a
dashboard
here,
so
the
idea
is
right
now
because
of
the
way
we
release
weekly
and
lts
and
sound.
We
first
do
a
maven
release.
B
B
The
idea
would
be
will
be
to
to
reduce
that
gap,
to
be
sure
that
when
we
release
everything
is
ready,
and
so
we
don't
have
to
wait.
20
minutes
to
have
everything.
A
So
your
checks
will
not
be
harmed
by
the
acceleration
that
you'll
do,
of
bringing
those
things
closer
together.
They'll
they'll
still
be
okay.
What
do
you
mean
by
my
check?
Will
not
be
well
so
so
today,
latest
package
available
here
shows
war,
for
instance,
so
when
war
package
I
think
there's
a
way
I
can
change
the
this
view.
Isn't
there
because
we
we
should,
we
can
actually
see
this
thing
in
its
time
scale.
A
A
B
To
put
that
to
a
different
time
frame,
I
mean
feel
free
to
ask.
This
is
just
the
url
that
we
that
we
let
other
people
to
use.
Otherwise,
personally,
I'm
directly
looking
at
the
dashboard
inside.
B
Yeah
may
I
will
increase
the
size,
but
in
this
case
we
we
just
saw.
We
just
saw
that
today,
because
today
we
released
the
weekly
we
saw
a
few
hours
ago
that
there
was
a
gap
for
the
weekly
release
in
that
dashboard.
B
I
just
want
to
clarify
here
that
the
wire
that
we
are
monitoring
is
not
the
one
from
the
maven
repository
is
the
one
from
get
the
jenkins
on
your.
So
I'm.
So
there
is
a
pythons.
I
wrote
a
python
script
that
read
the
maven
repository
metadata
file
to
know
what's
the
latest
version,
and
then
I
do
some
tests.
Some
queries
on
on
get
dutch
into
the
logo,
based
on
suggestion
from
garrett.
I'm
going
to
switch
from
get
the
jenkins
that
I
o
to
archive
the
jenkins
that
I
o.
B
The
reason
to
that
is
because
get
the
jenkins
that
I
o
has
that's
about
how
many
times
we
download
the
package
and
I'm
not
sure
how
that
query
is
done
on
the
service.
So
I
what
I
fear
is.
I
fear
that
we
increase
the
number
of
the
number
of
time
we
download
a
specific
version.
So
that's
why
I'm
going
to
query
directly
amir
that
we
know
that
we
automatically
update
each
time.
We
do
release
great
excellent.
B
So
the
next
topic
is
regarding
the
windows.
Docker
image.
If
I
understand
correctly
on
the
current
situation,
is
the
docker
image
is
ready
to
be
built
and
published?
The
problem
is
it's
taking
a
significant
time
to
build
and
publish
the
images
which
affects
security
releases,
and
so
now
garrett
is
working
to
identify
how
to
reduce
the
time
of
that
process.
D
Garrett
yeah,
so
that's
a
pr
went
up.
E
To
remove
what
we've,
what
was
felt
to
be
like
the
unused
windows,
docker
images
that
were
being
created
and
tested,
so
it's
only
currently
doing
java,
11
and
hotspot.
So
it's
it's
very!
It's
very
quick
now
and
then
that
docker,
the
pull
request
there
just
adds
an
extra
one.
So
even
so,
it's
still
it's
down
from
four
images
to
two
images.
So
it's
significantly
faster
than
it
was,
but.
E
Longer
term,
I
think
that
pr
is
good
to
go
longer
term.
I
think
we
want
to.
We
want
to
work
out
how
to
better
paralyze
those
windows
builds.
We
kind
of
we
can
improve
those
definitely.
A
And
I
think
this
is
this
is
exactly
trying
to
address
that
concern
that
daniel's
expressed
right
that
hey
a
one-hour
build
for
the
total
set
of
docker
images
is
too
long
and
so
gareth
you've
your
reduction
from
four
to
two,
because
these
were
built
built
in
series
right.
So
it
was
building
each
image
one
after
the
other,
not
not
even
running
them.
In
parallel
on
the
same
machine.
E
It's
actually
the
testing
that
takes
the
time
or
it
seems
to
be
the
build.
The
building
is,
I
think,
that's
relatively
quick,
but
because
it
it
builds
them
all
and
then
tests
them
all
one
after
the
other.
I
think
that's
the
bit.
That's
taking
the
time
yeah.
A
C
A
Yeah,
I
thought
today
that
the
docker
image
generation
was
distinct
and
separate
from
the
release,
build
process
that
it
just
consumes
a
release
and
detects
it
on
a
clock.
So
I
assume
daniel
was
launching
those
docker
image
builds
when
he
had
the
security
war
built,
then
he
has
to
launch
that
interactively
to
reduce
the
time
as
much
as
he
can.
A
B
So
that
I
never
work
on
another
current
staging
environment,
so
we
already
published
a
private
warfight,
so
the
private
one
is
already
available.
The
thing
is
right:
now
we
have
different
marvin
repositories
for
each
security
releases,
but
the
pro
I
mean
what
we
would
have
to
do
here.
We
would
have
to
create
private
stack,
average
decker
repository
yeah,
and
probably
we
would
have
to
work
on
the
script
for
doing
that
as
well.
So
yeah,
I
think
we
are
quite
far
from
staging
darker.
A
B
B
A
A
B
C
On
docker
there's
a
plugin
installation
manager,
cli
pull
request,
that's.
A
C
A
That
yeah,
I
think
it's
that
one
great
so
there
was
we
had
a
conversation
yesterday
in
the
doc
sig
tim.
There
was
some
some
issue
or
concern
raised
by
vlad
silverman,
asking
if
it's
okay,
that
we're
using
plug-in
installation
manager,
cli
everywhere
in
the
documentation
and
continuing
to
use
it
even
more,
I'm
assuming
it
is
okay
and
that
that's
our
strong
preference
does
that
make
sense
to
you
or
are
there
places
where
we
should
not
use
it?
I'd
definitely.
C
Use
it,
the
the
install
plug-in
script
is
gonna,
get
located,
I'm
using
it
constantly
for
reproducing
all
of
the
tables
to
dev's
issues.
A
C
A
B
2.263,
so
so
we
have
the
audi.
Do
you
hear
me?
We
do
yes,
you're,
very
audible,
okay,
perfect
thanks.
Everything
is
explained
in
the
jenkins
enforcer
release
kit
repository,
so
there
is
a
bunch
of
instructions
with
a
list
of
things
that
need
to
be
done.
So
I
think
we
have
to
create
a
branch
on
the
jenkinson
first
release
inside
that
specific
branch
we
have
to
create.
B
We
have
to
update
a
bunch
of
environment
variables
and
then
we
are
ready
to
to
trigger
the
stable
release,
but
it's
better
to
the
bench.
We
start
with
me.
A
B
A
C
And
open
id
connect
now
that's
been
moved
over
and
using
key
cloak
more
widely
or
so.
What's
the
what's
the
plan,
apparently
we're
still
in
the
same
situation,
where
we're
reliant
on
elder
right
and
we
lost,
we
lost.
B
Okay,
so
basically,
what
now
that
jira
is
on
running
on
the
linux
foundation?
We
have
to
to
change
the
the
user
directory.
So
right
now
we
are
doing
a
generic
user
directory
and
we
have
to
configure
the
open
edap
one.
So
we
could
get
a
in
that
case
we
would
still
be
using
open,
ended
up,
but
we
could
directly
use
key
clock
as
well.
We
just
have
to
configure
jira
yeah.
This
is
something
that
we
should
do
in
the
coming
weeks.
B
I
was
I
was
playing
with
that
yesterday,
and
this
is
something
that
we
should
go
to.
C
B
Yeah,
I
totally
agree
with
you.
I
totally
agree
with
you
and
we
just
have
to
add
one
new
yeah.
I
I
agree.
B
So
this
is
something
that
we
can
change,
so
we
just.
We
still
have
access
to
the
jira
and
illustration
to
the
admin
accounts,
and
so
we
just
have
to
add
one
additional
configuration
so
the
first,
the
first
iteration
we
will
be.
We
will
have
two
connectors
one
using
the
correct
the
current
one
that
we
have
at
the
moment
and
the
second
one
that
would
be
using
the
our
hoods
on
our
toggle.
B
So
we
would
in
the
back
end
it
would
be
exactly
the
same
user,
because
in
both
cases
our
users
are
coming
from
the
same
ldap.
But
the
thing
is
once
we
once
we
are
able
to
connect
using
kiklow.
We
can
remove
the
olds
on
the
old
connector.
B
We
just
have
to
confirm,
with
the
linux
nyx
foundation
that
we
won't
overwrite
any
configuration
on
their
side,
because
I
noticed
that
some
there
were
some
configurations
that
were
done
directly
in
the
java
properties,
and
so
we
may
yeah.
We
just
have
to
double
check
with
linux
transportation,
folks
that
we
can
safely
do
implement
this
change,
but
it
should
be
okay,
oh
yeah
that'll
be
nice
to
do
it.
B
I
don't
know
if
you
already
did
the
the
experiment
tim
with
configuring
jira
and
an
old
token.
C
I
think
I
managed
to
get
it
working,
but
it
was
a
bit
difficult
because
I
was
running
it
in
docker
and
just
getting
the
redirect
urls
mapping
properly
was
a
bit
of
a
pain.
I
think
I
managed
to
get
it.
Working
was
just.
C
C
C
B
Yeah,
I
didn't
want
to
spend
too
much
time
on
it
before
we
migrate
the
instance
to
the
linux
foundation,
but
I
had
a
look
yesterday
because
I
was
investigating
some
user
issues,
people
who
had
issues
to
connect
by
the
way
what
we
are
talking
about
jiren
did
you
figure
out
the
email
he
made
issues
nope,
not
nearly
not
yet.
Okay,.
B
Because
you
I
had
a
lot
to
send
grids
and
same
grid,
says
that
you
that
you
received
emailed.
But
you
were
not
the
only
person
complaining
about
that.
C
C
B
B
I'm
not
I'm
not,
then
the
problem
is
maybe
elsewhere.
So
I'm
getting
notification,
I'm
not
I'm
not
receiving
notification
from
the
jenkins
project
because
I'm
not
actively
following
that
part.
So
maybe
that's
why
so
maybe
it's
yeah!
Maybe
there
is
configuration.
A
C
Sure
yeah,
so
it's
infra
project
and
I
get
notification
subscription
emails
as
well.
So
if
I
configure
the
subscription
that's
that
works,
but
I
don't
get
anyone
pinging
me
or
creating
issues
and
assigning
them
to
me
or
anything.
A
B
Okay,
okay,
so
let's
let's,
let's
finish
this
meeting
and
then
let's
investigate
for
this
issue
later
on
thanks
everybody
have
a
good
day
and
see
you
in
rc,
bye.