►
From YouTube: 2022 11 29 Jenkins Infra Meeting
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
Welcome
to
the
chenkins
project
infrastructure,
weekly
team
meeting:
we
are
the
29th
of
November
2022..
Today
around
the
table
we
have
diamond
reportel
and
villamer
Mark,
wait
who
seems
really
motivated
to
take
notes
today:
Stephen
Merrill
and
Bruno
verarten,
thanks
Mark
for
already
updating
the
notes.
A
Let's
get
started
with
announcement,
so
the
weekly
2.380
of
Jenkins
score
is
out,
or
almost
at
least
the
war
packages
and
Docker
images
have
been
built.
The
docker,
if
many
images
was
built
five
minutes
ago.
So
that's
really
fresh
news
as
written
by
Mark.
Thanks
for
that
change,
log
is
merge,
but
there
are
Corrections.
Kevin
is
out
today,
so
that
might
have
to
wait
until
tomorrow.
A
A
special
thanks
for
everybody
to
double
check
that
the
infamous
plugin
pipeline
utility
steps
was
installed
again
on
release
CI
before
the
weekly
release
started.
Otherwise
it
would
have
been
broken
because
we
we
missed
some
usages
of
the
keyword
provided
by
by
this
pipeline
when
we
try
to
remove
it
a
few
weeks
ago,
so
they
are
walk
under
some
people
should
try
to
remove
these
keywords
as
a
general
matter.
A
We
don't
want
pipeline
to
read
the
ml
properties
file
on
the
controller
for
a
lot
of
good
reasons,
but
still
we
don't
want
to
break
the
weekly
release.
So
until
these
changes
have
been
finished,
reviewed
merge
better
to
keep
the
plugin
than
release
CI.
The
risk
is
quite
low
because
it's
not
a
public
facing
instance
so
and
it
has
been
updated,
so
that
should
be
okay,
but
let's
keep
an
eye.
So
thanks
survey
for
that.
A
I,
don't
have
a
other
announcement.
Do
you
have
some
nope
a
word
on
the
upcoming
calendar?
So
next
weekly
will
be
the
6th
of
December
2022
next
week
as
usual
tomorrow
we
have
an
LTS,
so
please
don't
merge
or
try
anything
the
infrastructure
which
could
be
a
vehicle
impact,
the
release
process
or
CI
Jenkins
IO
itself.
A
A
Any
questions
so
far.
Okay,
let's
proceed
with
the
work
we
did
this
week.
That
was
fully
completed
a
lot
of
Jenkins
recover
account.
Username
team
and
Alex
have
fun
interesting
things
survey
as
well,
so
we'll
speak
about
that
a
bit
later
on
the
new
topics,
it's
about
the
Eco
of
aconjo
and
kinsayo,
which
is
a
good
Eco.
That's
the
core
problem,
because
each
time
someone
is
locked
out
of
their
own
private
Jenkins
instance,
and
search
on
Google
on
how
to
recover
Jenkins
password.
A
B
C
Nothing
but
I
think
canonical
page
on
thinking
stuck
okay,
nice.
D
But
even
if
it's
too
late
could
do
something
as
we
do
in
the
community
I
mean,
could
we
get
a
template
or
boilerplate
something?
You
know
we
can't
answer
that
we
can
copy
and
pay
the
answer
saying.
No,
it's
not
account
on
jenkins.io.
Do
something
on
your
own,
and
here
is
the
answer.
C
You
can
make
predefined
answer
in
your
own
account
and
foreign.
A
Yeah
yeah
here
it's
only
writing
the
same
message.
The
idea
will
be
to
have
the
same
message
for
everyone.
There
I'm
not
sure
how
we
could
raise
either
a
bot
command
or
something,
but
at
least
having
the
template
written.
So
we
can
copy
it.
Past
will
be
a
first
step
because
the
Bots
did
not
coat
a
lot
of
the
issues,
because
we
have
some
that
are
localhost,
but
some
aren't.
C
A
And
some
don't
even
answer
the
correctly
the
question,
because
all.
A
A
C
C
A
C
A
D
A
Really
need
I
really
need
help
if
you
think
that
the
GitHub
app
or
whatever
automation
system
could
help
please
Implement
one.
It's
just
I,
don't
know
to
do
that
and
my
past
experiences
with
that
kind
of
system
are
100
of
the
time
catastrophic
when
I
do
it.
So
that's
why
I
prefer
relaying
on
someone
else
that
will
do
it
better
than
me.
A
I
take
on
me
putting
a
standard
message
and
communicating
to
everyone
on
the
team
as
short
term
that
we
have
a
standard
message
if
they
want
and
then
in
parallel
we
can
automate.
The
answer
looks
good
to
you.
Yes,
so,
let's
add
both
of
us
here.
A
I
will
update
these
shoes
open
by
team
about
the
SEO
we'll
see
if
we
need
another
issues,
but
the
goal
is
to
have
a
centralized
place
about
all
these
problems,
so
we'll
list
the
tasks
I'm
taking
over
that
one
I
won't
do
it
now
here
it's
not
the
location,
so
a
lot
of
login
issue
for
some
we
still
for
one
yes,
but
we
have
we
have
one
or
two
per
week.
We
still
have
a
HTTP
50,
so
you
have
captured
the
private
GitHub
gist
of
the
rawlog.
It's
about
the
config
property
missing.
A
We
might
need
help
on
that
one.
The
issue
is
already
erased
on
the,
so
just
keep
an
an
eye
on
that.
Most
of
the
time.
If
you
have
a
user
that
added
this
issue,
they
just
have
to
recover
the
password.
So
most
of
the
time
I
go
as
admin
I
find
because
the
ACT,
their
account
is
created.
It's
only
the
answer
which
fails,
so
you
search
their
account
or
the
email,
and
you
click
reset
password.
They
should
receive.
A
The
email
thing
is
that
they
never
answer
so
I,
don't
know
we
could
check
the
connection,
but
yeah
we
had
an
issue
about
slow
download
of
Jenkins
MSI
so
that
the
second
issue
about
someone
located
in
India,
who
is
in
fact
using
redirected
to
the
China
mirror
and
the
network
is
really
slow
from
India
to
China.
A
Despite
the
proximity,
we
have
that
project,
since
we
have
some
digital
ocean
credits
to
create
a
virtual
machine
that
would
act
as
a
mirror
right
now,
since
we
are
operating
a
lot
on
the
Azure
Network
that
takes
some
bandwidth
for
Airway,
I'm,
not
sure
Stefan.
If
you
will
want
to
try
to
set
up
the
virtual
machine
with
terraform
and
set
it
up
as
a
mirror
that
should
be
Apache
or
engineering
server,
I
think
Chrome.
That
should
be
whole.
C
C
D
D
A
Yes,
absolutely
we
can,
but
that's
the
same.
The
MJ
was
created
by
Olivia
I
removed
it
because
it
wasn't
used
anywhere
because
you
need
a
kubernetes
crystal
on
the
location
where
you
want
it
and
most
of
the
time
our
kubernetes
cluster
are
not
able
to
under
the
output
bandwidth.
Well,
when
you
have
a
virtual
machine
you
can
put
on
the
region
where
you
want,
that's
that
that
was
the
initial
reason
and
the
rest
is
only
a
nursing
command.
So
yeah,
are
you
okay
to
try
to
work
on
that
topic?
Yes,.
A
I'm
not
sure,
if
let
me
add
it
as
this
to
the
new
topic,
I
have
to
raise
an
issue.
An
issue
for
India
mirror
hold
the.
B
C
C
We
will
be
able
to
know
what
this
kind
of
service
consumer.
A
Yes
and
just
a
note,
we
took
the
the
bond
with
from
archives
in
kinsayo,
which
will
be
the
fallback
mirror,
and
it
was
only
a
few
bucks
per
month,
like
80
bucks
per
month
at
digital
ocean.
The
bandwidth
is
really
cheap.
It's
it's
AWS
who
make
the
band
with
really
expensive.
A
Most
of
the
time.
The
model
is
the
third
wing
AWS
you
have
like
until
two
terabyte
for
free,
but
once
you
pass
the
threshold,
you
pay
really
really
expensive,
While
most
of
the
ovah,
the
digital
Lucian
or,
let's
say
smaller
cloud
provider.
The
threshold
is
clearly
lower,
but
the
price
when
you
go
over
the
fresh
release,
chip
overage,
is
close
to
nothing
or
record
as
well.
So
that
should
be
okay,
and
if
it's
too
much,
we
can
always
shut
down
the
mirror
or
ask
others.
A
But
yeah
every
correctly
negotiated
with
digital
Essence
I'm
sure
that
will
be
really
fine
thanks.
A
Work
about
separated,
terraform,
backends
and
repository
Azure
net
and
Azure.
So
now
we
have
two
terraform
repositories.
So
that's
a
low
level
safety
issue
in
the
past.
The
reason
why
terraform
for
Azure
was
not
kept
up
to
date
by
Olivier
and
Tyler.
It's
because
any
contributor
or
any
terraform
issues,
and
most
of
the
time
it
was
terraform
issues
with
the
Azure
provider
when
it
was
breaking
or
destroying
resources.
In
an
let's
say,
unplanned
way.
A
The
risk
was
to
also
destroy
the
world
Network
and
DNS
of
the
system,
which
is
not
something
you
want
to
have.
So
the
idea
of
of
Airway
that
date
from
a
few
months
ago
was
just
implemented.
The
idea
is
to
have
two
separated
terraform
projects,
so
it
has
been
implemented.
So
the
Azure
only
reference
to
the
network
as
data
sources.
So
it's
a
read-only
mode
and
we
are
going
to
improve
the
security.
A
So
now
we
continue
we
put
on
the
other
Nets
the
networks
and
the
resource
is
strictly
related
to
network
that
shouldn't
mean
managed
on
the
other
repository
such
as
in
the
upcoming
future.
The
DNS
zones,
which
mean
you
can
create
a
DNS
record
on
the
other
repository
for
an
application,
but
the
Zone
in
itself
is
should
be
managed
there,
like
networks
and
right
now,
arva
is
working
on
the
new
VPN
for
the
private
Network,
so
the
virtual
machine
and
its
related
resource
are
on
Azure
Nets,
because
it's
strictly
required
to
access
the
private
Network.
A
So
thanks
every
for
that.
For
that
work,
the
job
continue
with
other
issues
that
are
on
the
currently
open.
On
the
milestone,
GitHub
permission,
someone
didn't
read
the
doc,
but
Alex
helped
on
this
one.
Thanks,
edx
Mark
I've
confirmed
that
Windows
slave
plugin
is
uninstalled
without
any
issue
on
all
controllers
and
still
present
after
updating
upgrading
controllers
and
plugins.
A
B
A
We
finally
were
able
to
open
pull
requests
back
on
cigen
games
IO
for
the
website
stories.
Jenkins
Ayo,
please
note
that
the
job
has
been
moved
to
another
location.
It's
now
under
the
website
directory
on
CI
jenkinsayo,
it
used
to
be
under
the
infra
directory.
A
The
pipeline
is
the
same
for
both
sea
agent
kinsayo
and
in
Frasier,
and
it
deploys
on
infra
as
usual,
so
that
one
is
okay.
We
have
a
pull
requests
that
has
been
updated
and
is
now
deploying
and
everything
works
as
expected,
thanks
for
the
album
that
one
Stefan
and
also
giving
finally
last
big
one,
is
the
component
alkay.
So
thanks
survey
for
taking
care
of
that
with
giving.
A
A
But
as
a
Jenkins
infra
team,
we
were
able
you
you
did
all
the
heavy
work,
but
I
mentioned
the
world
team.
We
now
have
an
official
npm
namespace
for
the
Jenkins
project,
which
allow
developers
contributors
to
puts
and
JavaScript
or
any
node.js
library
inside
that
repository,
which
is
the
default
entry
points
like
dockerbies
for
Docker
images
right
so.
C
Just
for
now,
a
token
has
to
be
set
on
every
repository
where
we
need
to
where
we
are
building
the
app
with
nuclection
you
might
have.
If
there
are
more
important
JavaScript
project
we
might
have
to
put
in
place
a
billion
VM
pipeline
or
something
like
that,
I
think
well
at
least
some
credentials,
so
people
could
access
use
them
in
the
apartment.
Yes,.
A
So
same
thing
as
stories
or
other
pipelines,
the
deployment
and
with
the
credential
will
happen
on
infrasti
by
contribution
could
still
have
a
crimson.
Fine
may
I
ask
you
to
write
down
a
quick
run
book
about
that.
A
A
Okay
so
handbook
for
npm,
just
a
note,
so
we
will.
We
won't
forget
about
that.
But
thanks.
That's
a
cool
word.
I
hope
it
helps
a
lot.
B
C
C
B
C
It
should
he
could
be,
it
could
be
include
as
Vedic
security
issue
or
worrying
about
this
component,
when
we,
when
we,
when
given
included
in
or
Wiki.
A
Thanks
a
lot
so
now
work
in
progress,
then,
so
we
have
two
top
level
priorities.
The
first
one
is
realigned
repo
Jenkins
ciog
Mission,
so
Mark
and
high
worked
on
so
many
months.
I
did
some
more
advanced
elements,
I'm,
not
really
pleased
with
the
experience
with
Chief
Rogue.
So
it's
really
hard
to
to
have
something
working,
because
each
time
we
make
a
change
on
a
repository
tree.
It
takes
some
time
to
propagate
and
to
have
an
on
all
the
let's
say,
the
instances
of
the
service.
A
So
we
have
weird
behaviors
so
sometimes
so
we
have
to
apply
configuration
and
wait
at
least
24
hours
that
everything
is
walking
and
behave
as
expected.
So
sometimes
it's
hard
when
you
test
things
so
I'm,
it's
I'm
really
slow
on
that
topic.
Tomorrow
we
will
have
the
meeting
with
g-frog,
so
we
don't
have
any
other
progress
since
last
time,
but
we
will
have
to
see
with
them
to
tell
them
it's
really
hard.
A
Yeah
I
I
I
did
yeah
too
much
thing
to
reproduce
the
setup.
Nothing
is
as
code,
so
everything
is
manually,
managed
that's
nightmare.
A
Honestly
trust
me
on
that
one
please,
the
M
chart
they
provide
is
is
the
one
of
the
worst
was
to
of
one
of
the
worst
I've
seen
that's
terrible
and
you
don't
want
to
run
artifact
or
inside
kubernetes
at
all.
I
think
maybe
postgresql
or
MySQL
is
was
to
run
in
production
with
kubernetes,
but
that's
close
right
yeah,
but
the
the
service
works
very
well.
They
will
have
I
think
it's.
They
will
have
to
migrate
it
to
another
platform.
A
We
haven't
heard
from
them,
yet
so
I
assume
that
will
be
in
January
I'm,
not
sure
they
will
be
able
to
do
it
in
December.
That
could
be
a
great
moment
to
do
it,
though,
if
they
have
the
ability,
because
I
don't
think
the
team
in
charge
will
have
Christmas
on
their
usual
early
days.
A
So
that
could
be
really
interesting
if
they
are
available
during
Christmas
on
you,
you
have
because
not
a
lot
of
activities,
so
that
could
be
a
nice
topic,
a
nice
time
window
for
maintenance,
same
second
thing
and
I
think
we
will
have
to
rush
that
one.
So
it's
just
a
note.
No
action
required
I'm
taking
this
based
on
last
year.
The
certificate
on
the
G4
got
effectively.
Repository
is
going
to
expire.
We
already
talked
about
that
last
week.
A
That's
something
generated
manually,
I,
think
I
will
I
just
need
to
validate
everything
because
UK
did
last
year,
but
it
looks
like
that.
A
The
plan
will
be
the
following:
I
will
use
the
our
search
manager
on
on
the
Azure
cluster
to
generate
manually
certificate
for
repojenkins,
ci.org
I
think
I
will
add
jumia
Ingress
that
should
do
the
job
once
the
certificates,
the
new
certificate
is
generated
with,
let's
encrypt
as
a
provider,
I
will
have
to
send
encrypted
to
people
like
g-frog
the
certificate
and
the
key,
like
kosuke,
did
last
year,
I
encrypted
that
with
gpg
key.
A
So
that's
why
I
will
have
to
generate
it.
Why
using
certain
managers,
because
we
don't
own
the
web
server
behind
the
report-
Jenkins
here.org.
So
we
need
to
prove
to
to
the
certificate
author
and
we
have
access
to
the
Jenkins
CI
org
DNS.
Everything
is
set
on
search
manager,
so
we
can
generate
it
using
DNS
Challenge
on
Nets
encrypt.
That
should
be
quite
easy
and
we
won't
have
to
bother
or
rely
on
kosuke
for
this
one.
A
Next
topic:
that's
working
progress,
recreate
networks
on
Azure,
first
private
and
then
the
publics
to
fix
two
main
issues.
The
Quran
public
networks
doesn't
as
overlap
you
with
the
IP,
and
we
want
to
support
IPv6,
at
least
for
external
users.
So
there
is
a
work
in
progress
on
that
area.
As
we
said
now,
we
have
a
fully
code
managed
with
the
the
work
of
Harvey
Avi
was
able
to
recreate
from
scratch
a
new
sets
of
network
of
private
Network
that
could
be
turned
to
dual
stack,
ipv4
and
IPv6
or
not.
A
We
can
change
them
without
recreating
them
with
the
new
Networks,
a
new
set
of
ips,
so
it's
created
from
scratch,
so
we
will
be
able
to
do
a
b
migration
and
the
branch
will
be
the
same
for
the
public
networks
as
well.
Regarding
IPv6,
it's
still
complicated
to
find
the
find
solution,
because
as
soon
as
we
go
on
kubernetes
it's
we
we
are
going
to
test
that,
but
it's
hard
to
have
internally
only
ipv4
and
external
aipv6,
it's
hard
to
find
an
applicant
or
resource
that
should
be
able
to
do
the
Gateway.
A
The
Gateway
between
both
the
thing
is
that
if
we
say
we
need
to
have
a
full
IPv6
routing
kubernetes
supports
that
AKs
as
well.
But
in
that
case
we
need
to
change
the
network
mode
use
inside
kubernetes
clusters
and
that
new
network
mode
has
two
downside
for
us.
First
one
it
doesn't
support
Windows
Walker
Walker
nodes.
A
That
should
not
be
an
issue
because
we
don't
need
IPv6
for
the
private
clusters
where
we
have
Windows
not
pulled
in
the
future.
But
the
second
thing
is
that
it
expose
some
endpoint
of
the
kubernetes
agent
named
kiblet.
It
exposed
the
main
points
of
the
Walker
notes
and
we
don't.
We
don't
really
want
that.
So
right
now
we
haven't
pushed
further.
What
have
we
proposed
and
we
are
going
to
try
this?
A
A
A
So
work
in
progress
right
now.
The
status
working
on
the
VPN
virtual
machine.
A
Two
layers,
one
infrastructure
as
code
and
then
prepared
updates,
so
we
should
be
able,
in
the
future,
to
have
two
vpns
one
for
the
private
Network
for
CI
access
release,
CI
access
and
infrastructure,
tooling
access
and
the
VPN
for
public
will
allow
us
to
let
access
to
the
public
kubernetes
cluster,
for
people
like
Adrian,
for
instance.
So
they
could.
A
A
Next
Step
publish
pipeline
step,
dog
generator
and
back-end
extension,
so
last
last.
Step
of
this
issue
is
to
have
Sia
agents
IO
able
to
build
the
same
pipeline
as
in
Frasier
for
both
projects.
One
of
the
project
has
been
done
so
this
morning.
So
thanks
Stefan
for
that
walk
and
I
think
the
order
is
almost
done.
Yeah.
A
Yes,
so
that
one
is
considered
almost
closed
and
it
was
a
nice
exercise
for
Stefan
to
merge,
Pipeline
and
play
around
without
configuration.
One
note:
Stefan
was
able
to
start
improving
the
labels
that
we
provide
for
developer
on
cig
and
kinsayo
to
step
by
step
going
to
labels.
That
makes
more
sense,
such
as
Linux
imd64,
because
Linux
only
doesn't
make
sense.
Linux
is
a
platform
and
as
a
label
it's
complicated
given
we
provide
different
CPU
different
kind
of
Linux
instances,
containers
or
everything.
B
So
so,
if
I,
but
if
I
mean
that
Linux
could
be
any
Linux
that's
available,
how
do
would
I
continue
to
say
that
would
I
continue
to
use
that
label?
Why
would
you
want
to
have
such
labor
because
I
like
that,
there's
intentional
variation
running
the
get
client
tests
on
arm
or
on
AMD
64
or
system
390
wherever
I
like
that?
It
sometimes
finds
failures
for
me
so
but
I
I
also
understand.
B
B
Which,
which
is
exactly
what
I'm
getting
and
I
I,
actually
have
that
in
my
in
my
personal
setup,
where
I
intentionally
allow
Linux
means
either
Centos
or
centos7
or
Ubuntu,
22
or
Etc,
and
and
that
variation
actually
helps
me.
But
that
may
not
help
others
and
I
understand
if
it
doesn't
help.
Others
I
happen
to
have
a
platform
specific
plug-in
that
I
like
getting
variation
as
I
go.
B
But
it
sounds
like
the
the
the
plan
is
not
to
have
that
kind
of
variation
automatically
and
most
people
probably
don't
think
of
it
and
would
be
surprised
if
they
got
it.
So.
A
I
support
the
use
case
is
interesting
because,
as
I
always
said,
we
could
provide
a
new
attributes
to
the
build
plugins
or
equivalents
but
being
able
to
say
I
want
to
either
randomly
select
a
cprchitecture
or
a
Linux
flavor.
A
Linux
flavor
is
is
not
an
option
here,
because
we
don't
provide
some
figures
than
Ubuntu.
We
could
in
the
future,
though,
but
that's
clearly
not
right
now,
but
we
have
different
CPU,
so
that
could
be
a
feature
of
the
vibrant
libraries
functions
that
say:
Okay
explicitly
as
a
maintenance.
A
I
want
to
either
build
an
old
CPU
flavors
for
this
platform
or
select
randomly
and
change.
So
you
can
introduce
cows
on
your
builds,
but
that
thing
must
be
first
explicit
and
second
not
defaults,
because
that
will
bite
a
lot
of
users.
Remember.
The
RM
and
plan
builds
that
were
nice
for
the
cost
of
the
infrastructure,
but
that
weren't
nice
for
the
debug
surprise
for
the
plug-in
maintenance
and
that's
the
main
issue
is
that
we
don't
want
labels
that
are
not
deterministic
okay.
A
So
there
are
some
part
that
we
cannot
remove
like
when
we
say
your
Linux
same
d64
with
Docker.
Still,
we
can
schedule
that
on
a
pod
or
virtual
machine
on
ec2
or
Azure
VM,
that's
up
to
us.
That's
already
a
surprise,
surprising
change
that
can
happen.
So,
let's
try
to
not
add
more,
but
we
are
not
removing
Linux
label.
Yet
Stefan
did
the
work.
That's
the
other
issue,
a
Windows
agent
disconnect
prematurely
the
work
is
going
on
on
Windows
that
has
less
usages.
A
E
If,
if
we
look
that
from
the
big
picture,
I
I
really
think
that
to
to
provide
the
best
working
area
for
your
plugin,
you
should
build
on
any
on
all
of
them
and
have
a
failure
on
on
the
one
on
anyone
which
fell,
because
if
you
really
want
to
be
able
to
have
your
plugin
working
on
any
kind
of
Linux,
you
should
try
each
time
with
any
kind
of
Linux.
But
it
would.
C
We
would
need
to
build
foreign.
E
If
you
think
of
is
plug-in
and
and
with
the
random
solution,
I
think
that
means
I
think
one
every
eight
or
ten
times
it
will
find
a
bag
that
is
probably
coming
from
like
eight
eight
times
before
it
did
the
change.
So
it's
it's
very
hard
to
debug
everything.
I.
E
D
A
The
order
of
magnitude
will
be
when
we
will
cross
10
axes
for
the
Matrix.
We
can
think
about
cost.
That's
one
order
of
magnitudes
compared
to
the
time
Wasted
by
the
plug-in
maintainers,
trying
to
understand
why
the
plugin
is.
D
E
A
Eventually,
for
Windows
that
might
be
an
issue.
However,
here
first
we
have
the
cpu-z
machine,
which
is
one
machine,
so
we
don't
pay
more
for
doing
more
Builds
on
that
machine.
It's
a
permanent
agent.
So
that's
the
first
exotic
one.
The
irm
builds
are
20
to
30
percent
cheaper
than
the
AMD
builds,
and
that's
a
test
issue,
though
I'm
not
sure
about
how
it's
start
detected.
A
But
right
now,
if
you
build
once
the
Java,
the
Java
binary
you
archive
it,
and
then
you
have
a
set
of
smoke
tests
that
say:
I
I
have
smaller
and
shorter
tests
for
each
platform.
You
have
one
way
more
efficient
than
not
building
or
randomly
building
and
the
other,
the
other.
Let's
say
common
thing:
that's
all
of
the
major
open
source
project.
Does
we
say
we
support
this
platform?
We
need
to
smoke
test
on
this
platform.
Absolutely
otherwise.
It's
not
support
it's
random
guess
at
best.
A
Mean
wow:
we
are
missing
a
lot
of
tests
here
because
we
don't
want
to
run
a
full
test.
Suite
on
that
area.
You
want
to
run
a
smoke
test
and
eventually
specific
issues,
so
I
mean
for
us
as
infrastructure.
We
provide
the
platform,
so
we
don't
have
to
say,
build
more
or
build
less.
We
say:
okay,
we
are
right.
Now
we
have
a
still
submerging
so
better
to
provide
that
I.
Don't
say
it
it's
urgent,
but
that's
the
direction
right
now.
The
issues
I
mentioned
is
just
we
added
more
labels,
more
specific
labels.
A
B
Yet,
no
and
and
and
I
I
think
you're
you're
I
I
fully
support
the
determinism.
The
deterministic
is
much
better
for
the
consumers
right
this.
This
randomization
use
case
is
is
quite
exotic.
I,
even
given
Stefan's
comments,
it
is
quite
exotic
and
it
would
be
perfectly
fine
for
us
to
tell
people
like
Mark
Waite
who
have
platform
specific
things.
B
A
The
feature
of
a
random
selection,
as
always
said,
if
you
are,
if
you
provide
an
attribute,
let's
say:
random,
random
test
or
random
builds,
and
you
provide
yourself
a
list
of
labels.
If
we
provide
these
labels,
that
could
be
that
that
should
fulfill
the
same,
there
is
a
need
and
we
can
provide
that
needs.
That's
what
I
want
to
say.
B
A
A
Toward
a
cleanup
of
resources,
so
there
are
two
resource
groups
that
cost
us
money.
So
first
prod
Confluence.
So
are
they
your
content
that
you
removed?
One
of
the
two
databases
right.
C
C
C
A
C
I
I
have
to
put
this
somewhere.
Okay
when
compressed
the
weight
about
500
gigilates.
We
I
can
open
a
guitar
paper
and,
put
it
there,
a
private,
triple
or
I,
don't
know
as
we
want
yeah.
A
A
Looks
good
to
you.
Yes,
that
should
be
300
bucks
per
month.
Almost
freeze,
3060
bucks
per
year
saved
thanks
to
that
work.
So
thanks
survey
for
that
same
for
broad
Community
function,
I'm
late
on
that
topic,
I
have
to
ask
on
the
developer
mailing
lists.
If
anyone
is
aware
of
any
functions
as
your
functions
still
being
used,
some
are
evergreen
directly
related
and
can
be
removed
and
deleted
from
Azure,
but
some
other
are
like
infra
function
and
we
weren't
sure
if
they
were
used.
So
let's
send
an
email
before
deleting
the
cost.
A
So
continue
working
on
that,
but
these
two
should
help
just
a
quick
one
that
I've
opened
the
two
last
one:
that's
an
old
one
that
the
dog
was
spamming
on
search
side
that
was
cooked
by
Daniel
and
I.
Have
the
issue
yesterday
on
a
personal
machine
with
data
dog
agent.
It
was
sending
a
lot
of
unable
to
get
this
Matrix
logs
on
syslog
by
data
dog
agents.
A
More
specifically,
these
were
always
related
to
VAR,
lib
Docker
overlay,
so
one
of
the
layers
of
one
of
the
containers
or
multiple
and
some
on
the
docker
net
NS.
These
two
are
two
specific
file
system
that
can
be
blacklisted
from
what
the
data
dog
agent
change.
It
used
to
work
differently
with
the
version
6
of
the
data,
the
gauge
and
o5,
but
since
the
version
the
latest
that
we
have
since
a
few
version,
it
generates
a
lot
of
error.
Clearly.
A
So
since
at
least
yeah
yeah
I
think
it's
it's
clearly
older
than
that,
I'm,
not
sure
that
initial
message
was
related
to
Docker,
but
still
now,
no
more
message
with
the
fix
I
provided.
So
it's
currently
waiting
for
review
for
someone
and
for
deployment
I
took
the
opportunity
to
to
take
that
issue
down.
Anyone
interested
on
pairing
and
merging
in
that
pair,
so
we
can
show
knowledge
is
welcome
to
mention
themselves
and
the
last
one
Java
19
Stephen.
What's
up.
E
I
did
two
on
three:
it's
it's
looking
good
I'm
I
studied
the
third
one
this
morning
and
right
now,
I'm
stuck
because
the
inbound
agent
on
Jenkins
is
not
prepared,
as
with
GDK
19,
so
I
I
might
have
to
work
on
a
repository
somewhere
with
drinking
C.
I
know
no.
C
D
E
A
The
same
model
that
model
will
work
okay,
so
that
means
in
the
upcoming
Milestone.
Once
this,
let's
say
minor
and
ions
will
be
fixed
Stefan.
You
should
be
able
to
announce
on
the
developer
mailing
list
that
anyone
will
want
to
try
early
builds
of
their
Jenkins
project
with
gdka
19
will
be
able
to
do.
E
B
A
E
A
B
A
A
If
it's
okay,
if
but
if
Daniel,
can
confirm
what
just
the
Mark
said
and
if
we
don't
find
any
independency
with
the
group
ID
artifact
ID
on
Java
other
Java
projects,
then
I
propose
that
we
simply
archive
the
repository.
We
can
always
unarchive
and
make
it
change
somewhere,
but
yeah
that
will
at
least
silence
the
security
issue.
And
since
the
code
hasn't
been
changed
since
two,
almost
three
years,
I
will
say:
I'm,
not
sure
that
project
is
used
anymore.
A
B
C
C
A
A
B
So
you
had
mentioned
the
the
expiration
of
one
of
the
certificates.
There
are
two
other
certificates
that
are
pending
expiration,
jenkins.ci.org
expires
in
38
days
and
jenkins.io
expires
in
58
days,
but
I
believe
both
those
will
be
Auto
renewed
and
if,
if
we
get
under
28
days
and
they
haven't
been
renewed,
I'll
go
talking
to
Tyler
Croy
because
he's
he
he
has
the
one
he's
the
one
who
processes
the
automated
renewal.
B
Names,
oh
sorry,
is
there
a
difference?
You're
right
domain
names?
Yes,
sorry:
okay,
not
certificate
domain
names,
you're.