►
From YouTube: 2022 04 26 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
B
Sure
yeah,
so
the
tim
jacom
has
raised
the
question
in
the
jenkins
developer
list.
The
lts
baseline
has
not
yet
been
selected
for
the
what
would
usually
be
the
june
lts,
it's
being
discussed
on
the
developer
list.
I
had
proposed
about
a
week
ago
that
we'd
actually
intentionally
so
maybe
it
was
two
weeks
ago
now
that
we
intentionally
slip
the
selection
of
the
lts
baseline
two
weeks,
two
or
four
weeks
to
allow
us
more
time
to
resolve
regressions.
B
A
number
of
regressions
have
been
resolved.
A
very
nice
set
have
been
resolved,
but
there
are
still
a
few
that
I
feel
like
hey.
We
should
really
consider
fixing
these
before
we
choose
the
weekly
baseline
but
tim's
the
release
officer,
and
that
that
really
is
his
decision
to
make.
We
we
defer
to
him.
We
try
to
persuade
etc,
but
as
a
release
officer,
he
gets
to
make
that
decision.
B
A
That
was
all
I
had
damien
thanks.
That's
clear,
any
question,
one,
two,
three!
Okay!
I
don't
have
any
other
announcements.
Do
you
have
some
folks.
A
So
we
had
a
bunch
of
minor
tasks
that
we
won't
cover
in
details
that
are
usual
operationals
people
wanting
to
be
obtained
or
opt
out
of
plug-in
maintenance,
dns
updates
on
some
records
just
a
few
words,
so
you
have
the
direct
link
to
the
milestone
with
the
closest
shoe.
If
you
want
to
check
the
details,
I'm
gonna
just
cover
some
major
elements.
A
A
It's
a
centralized
repository
that
provides
for
the
ubuntu
and
windows
lines
that
try
to
provide
the
same
templates
either
you
are,
we
are
using
azure
or
aws
machine
that
might
also
build
docker
container
in
the
future.
To
be
sure,
we
have
the
same
set
of
toolings
whatever
kind
of
agent
we
have
right.
Now,
it's
only
virtual
machines.
A
Okay,
a
water
thanks,
a
lot
urban
stephan
for
contributing
on
the
vpn
tasks.
So,
as
usual,
every
six
months,
we
have
a
regular
task
to
update
the
crl
the
certificate
revocation
list
of
our
infrastructure
vpn.
That
was
a
good
opportunity
for
knowledge,
sharing
and
improvement.
So
the
goal
was
to
let
two
members
of
the
team
that
are
not
olivia
mark
or
I
who
usually
do
that
task,
so
they
did
that
in
complete
autonomy.
A
So
thanks
for
taking
care
of
that
for
the
improvement
on
the
documentation-
and
that
was
a
good
opportunity
for
taking
care
of
vadec
renewal-
I
mean
it's
not
as
if
vedic
had
the
need
to
access
the
vpn,
so
he
confirmed
that
he
was
able
to
go
forward.
So
thanks
folks,
a
word
on
issues
jenkins
io
that
has
been
upgraded
without
further
notice
to
the
latest.
A
Lts
sounds
like
we
had
a
miscomprehension
with
the
linux
foundation,
folks,
so
I've
written
that
on
the
issues,
but
I'm
saying
that
a
lot
for
everyone
there
when
they
propose
a
date
time-
it's
pst
west
coast,
good
to
know,
because
until
we
realize
that
it
was
that
time
zone,
the
team
already
proceeded
and
the
person
in
charge
of
communication
was
off
today.
We
should
have
delayed
so
better
to
know
next
time
for
once
it's
not
european
centered.
A
A
That's
something
that
happened
yearly,
that
lts
update
that
was
triggered
by
the
health
check
inside
the
instance,
so
everything
so
far
so
good
based
on
team
analysis.
He
was
able
to
finish
the
update
in
the
u
administration
ui,
I
haven't
seen
any
user
complain
so
either
no
one
can
pass
to
jira
or
it's
working
as
expected,
or
we
haven't
seen
any
problem.
A
A
A
Pipeline
library,
we
just
walked
around
some
improvements,
thanks
basil
for
starting
a
big
cleanup
of
that
groovy
code.
That's
a
good
learning
opportunity
also,
so
we
spent
some
times
with
stefan
last
week
and
today
around
the
groovy
string
in
depolation
fixes.
A
A
Finally,
infra
report
generation,
which
is
used
by
the
repository
permission,
updater,
that's
a
regular
job
that
takes
care
of
updating
the
authorization
and
account
and
permission
on
github
repository
for
plugin
developers
and
artifactory
has
been
fully
migrated
to
infra
ci.
It
was
running
on
trusted
ci
before
creating
a
lot
of
chaos
there
and
costing
us
a
lot
of
money.
So
now
it's
running
on
pods
on
one
cpu,
one
gigabyte
that
cost
almost
nothing
and
so
far
so
good,
took
some
time
to
compare
that
with
the
help
of
some
contributor.
A
A
A
That
should
be
publicly
hosted
to
focus
on
the
topic
of
the
postmortem,
because
we
are
not
mature
enough
as
a
team
on
the
post-mortem
exercise,
so
we
could
totally
improve
and
learn
from
bazel
and
there
is
a
technical,
a
set
of
technical
elements,
short
term
medium
term
and
long
term.
That
should
benefit
the
world
jenkins
core
or
what
is
some
plugins?
A
So
that's
interesting
to
discuss
that
to
share
that
knowledge
altogether.
That
will
be
the
goal
of
such
elements.
So
what
happened
briefly
is
that
we
had
a
wave
of
builds.
We
had
a
bunch
like
more
than
thousand
five
hundred
bills,
waiting
in
the
q
and
c
I
jumped
in
sayo
the
root
cause.
A
They
are
okay
to
let
all
of
our
organization
and
account
to
be
part
of
the
open
source
program
that
would
allow
us
to
have
more
than
5
thousand
pull
per
hour
instead
of
200
per
6
hours
today.
That
would
be
a
great
improvement
that
should
avoid
that
thing
to
happen
again.
There
are
other
solutions,
of
course,
using
local
proxy
using
other
registry,
but
right
now
that
was
the
root
cause.
A
A
That's
why
we
need,
let's
say
the
professional
experts
like
basil
to
step
into
there
to
help
us,
because
we
need
that
help
to
fully
understand
what
is
going
on
to
be
able
then
to
decide
what
to
do
in
short
term
if
it
happens
again
in
the
next
days
and
what
to
do
in
long
term
to
fix
that
behaviors,
because
if
we
have
that
behaviors
other
users
of
jenkins
for
sure
will
have
this,
we
are
not
exceptional.
We
are
not
specific.
A
B
So
damian
one
of
one
of
my
concerns
and
you
didn't
mention
it,
so
I'm
I'm
wondering
if
maybe
I'm
thinking
in
a
different
area.
We've
had
an
ongoing
issue
for
the
last
18
months
or
24
months
of
agents
that
disconnect
at
random
cloud
agents
did
that
have
any
impact
on
this
or
or
not.
As
far
as
you
could
tell,
the
the
random
disconnects
of
agents
at
unpredictable
times.
A
B
Okay,
all
right,
so
they
were
agent
allocation
failures,
not
jobs
due
to
disconnects
not
not
job
failures
due
to
the
disconnect
of
the
agent
okay,
but
to
be
more
precise,.
C
A
A
C
Your
mind
I
already
I
already
wrote
a
quick
document
and
I
think,
there's
five
different
issues
that
I
saw
during
that
incident.
So
I
have
a
draft
of
what
I
think
the
short,
medium
and
long
term
action
items
are
for
each
of
those
five
issues
and
I'd
be
happy
to
discuss
that
at
the
meeting
that
you're
closing,
I
could
share.
D
A
A
E
A
Where
should
I
post
that,
if
it,
if
you
want
it
to
be
publicly-
and
you
don't
mind,
they
reckon
the
associated
issue,
let
me
have
the
link
there.
That's
the
issue
you
opened
on
the
help
desk.
Sir
okay
sounds
good
to
you.
Yes,
thank
you.
Many
thanks.
A
There
is
an
upcoming
issue
with
the
docker
image
tags,
lts,
whatever
lps,
something
that
are
not
published
for
other
platform
than
intel.
It's
almost
fixed.
We
found
a
root
cause
that
was
a
chaotic
branch
being
built
in
production
on
trusted
ci
that
should
not
have
been
that
was
building
with
an
old
docker
version
overriding
the
manifests,
the
modern,
manifest
version
to
new
ones.
So
it
was
indirectly
making
unavailable
the
ram
and
cpu's.
A
A
It's
blocking
the
availability
of
today's
weekly
docker
image,
of
course,
because
that
script
of
publication
must
be
fixed
and
fully
working.
So
it's
continue
on
the
next
milestone.
If
anyone
is
interested
in
contributing
on
helping
me,
don't
hesitate
to
manifest
yourself
on
the
associated
issue
that
you
can
find
on
the
milestone.
B
A
We
have
the
kubernetes
upgrade
campaign
to
1.21,
so
it's
managed
by
jarvie
and
stefan
I
unders.
So
we
decided
two
weeks
ago
that
airway
is
leading
and
stefan
is
shadowing
rv
as
a
matter
of
knowledge
transfer.
A
Initially
it
would
have
been
all
the
cluster
without
me
due
to
last
week
cause
I
I'm
sorry,
I
I
stole
your
task
folks,
but
now
there
is
the
world
aks
that
should
happen
tomorrow
morning.
Is
that
correct?
Are
they
yes
cool?
So
I
let
you
fault
manage
that
part
thanks
a
lot,
so
we
keep
that
for
the
next
milestone.
Of
course,
stefan
and
I
are
working
on
the
doc
up
credential.
A
We
have
multiple
areas
there
documenting
the
different
accounts.
Securing
the
different
accounts,
ensuring
the
open
source
program
is
applied
for
the
rate
limiting
and
fixing
the
pipeline
library.
So
each
controller
of
the
infrastructure
use
one
credential
for
pool
and
another
credential
for
push
to
separate
concerns
and
writes.
So
that's
an
ongoing
task
that
should
continue
this
week.
A
Did
I
forget
anything
stefan
or
is
that
you
you
can
specify
that
we
we
may
use
credential
with
read-only
for
pull
to
avoid
any
yeah
stalling
credentials
and
pushing
wrong
image,
yeah
good
catch.
That's
the
reason.
Docker
up
feature
released
two
weeks
ago
now.
A
Accounts
are
able
to
have
a
read-only
token.
That
feature
was
only
for
enterprise
or
paid
account
last
month,
and
now
we
can
use
it
for
free,
even
without
the
open
source
program.
So
if
you
have
a
free,
docker
earn
up
account.
You
can
do
that
on
your
own,
so
that
the
good
news
there
is
read,
write
and
read,
write
delete.
There
are
four
or
five
sets
of
permission.
That's
really
useful
good
catch
steph
thanks,
so
we
continue
working
on
that.
Stefan
on
your
own,
you
have
to
continue
working
on
migrating
rating.
A
No
problem,
that's
a
non-priority
task,
but
a
good
knowledge
transfer.
So
we
delayed
to
that
milestone.
You
were
ill
and
off
most
of
the
past
two
weeks,
so
that
makes
sense.
Sorry,
the
gdk
17
campaign
that
has
been
updated
on
the
tools
for
ci,
junkins,
io
and
virtual
machines.
So
now
I
have
two
points.
First,
the
container
agents
that
are
still
partly
built
by
ourself
and
partly
built
from
the
community,
that's
related
to
another
task
survey
that
was
building
our
own
windows
container
inside
the
infra
ci.
A
A
A
So
the
ddk
17
campaign
is
parallel
to
that
task
that
the
same
area-
I
don't
know
if
someone
volunteers
to
track
all
the
docker
images
we
have
running
that
might
be
the
linux
are
easy
and
the
windows
might
be
a
more
a
bit
more
slow
because
related
to
external
requests
to
the
jenkins
ci,
but
we
have
to
track
them.
Is
there
anyone
volunteering
to
that
task,
or
should
I
take
it.
A
C
C
C
My
suspicion
is
that
we
may
be.
There
may
be
some
areas
that
we
can
optimize
our
memory
usage,
for
example,
depending
on
how
much
memory
the
agents
are
using,
we
might
be
able
to
turn
that
down
with
some
explicit
configuration
and
the
other
thing
that
I'm
trying
to
determine
is
once
we
understand
how
much
memory
we're
using
my
question
is:
are
we
giving
ourselves
containers
that
are
too
small
or
too
big
for
our
actual
workload?
So
I
don't
know
the
answer
to
that.
C
Yet,
because
I
haven't
quantified
every
single
jvm's
heap
size,
but
once
I
do
I
I
want
to
determine.
Is
this
eight
gigabyte
container
vm?
Is
it
too
large?
Is
it
too
small
if
it's
too
large,
can
we
save
money
by
shrinking
it
to
a
smaller
size?
If
it's
too
small,
is
that?
Why
could
that
be
one
of
the
reasons
why
these
builds
are
failing?
C
So
these
are
the
questions
that
I'm
seeking
to
answer
and
I'll
probably
have
more
of
an
update
on
that
in
the
next
week
or
so
once
I'm
able
to
do
some
more
analysis
there,
but
I
I
think
now
I
finally
have
all
the
information
that
I
need
to
start
digging
into
this.
So
that's
small
update
there
I'll
pause.
If
you
had
any
questions
for
me
about
that,
okay,.
E
Okay,
I
don't
think
there's
any
questions
so
the
other
thing
yep.
Just
so
one
question:
are
you
how.
F
Are
you
measuring
it
are?
You
is
using
open,
telemetry.
C
Much
more
primitive
than
that,
I'm
just
having
the
jvm
print,
it's
what
it's
decided
to
use
with
the
simple
command
line
flag,
and
then
I
have
a
piece
of
paper
and
I'm
adding
up
all
the
numbers.
So
it's
really
nothing
nothing
advanced
or
sophisticated.
C
All
I'm
all
I'm
doing
is
looking
at
the
the
heap
size
for
each
jvm
and
adding
them
all
up.
And
in
my
experience
that's
pretty
close
to
the
the
actual
usage
I
mean.
Obviously
it's
not
one-to-one
correspondence,
there's
other
other
factors
that
are
at
play,
but
I'm
not
gonna
dig
deeper
into
those
until
I
know
that
I
need
to
for
now,
just
a
bit
just
a
very
high
level
overview
is
sufficient
for
me.
A
C
A
To
the
ssh
machine,
so
you
so
I
will
open
an
l
desk
issue
and
mention
you,
because
there
might
be
some
tasks
on
your
own
to
be
done,
but
I
will
describe
them
that
should
allow
you
to
work
on
the
controller
itself
and
since
you
are
working
on
the
gvm
matrix,
I
propose
that
we
add
you
as
well
on
the
datadog
organization,
so
you
can
see
the
metrics
that
are
sent
to
datadog,
especially
the
memory
metrics.
That
should
help
you
a
lot.
I
guess.
Oh.
C
A
No,
no
problem,
just
one
thing:
if
you
need
anything
that
will
that
could
help
you
on
the
analysis,
please
shoot
and
ask:
we
can
give
a
lot
of
access
if
needed.
We
don't
have
an
exhaustive
list,
that's
an
improvement
for
the
documentation
team,
but
if
you
see
that
you
feel
completely
locked
out
of
the
system,
please
ask
us:
we
can
open
access
even
temporary
once,
but
if
it's
easier
for
you
that
will
help
everyone
inside.
B
C
Thanks
mark
great,
so
the
other.
The
other
thing
that
I've
been
looking
at
is
improvements
to
the
pipeline
shared
library.
So
you
might
have
seen
that
I've
been
refactoring
a
lot
of
things
and
I'm
not
just
doing
that
for
fun.
There's,
actually
a
change
that
I'm
working
on
making
that
has
required
some
refactoring
as
a
prerequisite,
and
that
is
enabling
the
use
of
the
maven
wrapper,
which
is
something
that
many
people
have
requested
in
jenkins
core
and
instead
of
instead
of
running
the
maven
that's
installed
in
the
virtual
machine
or
container.
C
You
know
we
had
to
roll
that
back
at
the
infrastructure
level.
But
you
know
if
we
were
using
the
maven
wrapper,
we
would
have
been
able
to
roll
that
back
at
the
application
level,
and
that
might
and
the
reason
that
that's
important
for
new
contributors
is
that
it
also
applies
to
local
development
as
well
as
ci
builds,
and
so
in
order
to
update
the
pipeline
shared
library
to
support
the
maven
wrapper.
C
C
C
Yeah
exactly
so
the
the
there's
two
consumers
of
this
red
maven
function.
Well,
not
just
two
but
two
primary
consumers.
There's
the
the
build
plugin
dot
groovy,
which
is
basically
the
the
jenkins
file
for
all
plugins
and
then
there's
the
the
core
jenkins
file.
I
think
those
are
the
two
primary
consumers.
There
are
many
other
smaller
consumers
as
well,
but
the
idea
behind
my
change
is
that
the
the
change
would
be
a
generic
change
to
this
red
maven
function
so
such
that
it
would
apply
to
all
consumers,
including
plug-ins.
A
The
reason
was
trying
to
avoid
downloading
most
of
the
time
maven
on
each
time,
but
that
will
go
against
the
case
of
is,
but
I
don't
have
that
exact
version.
So
having
that
feature
is
really
useful.
We
were
also
trying
to
work
on
the
ensuring
that
the
tooling
used
to
build
come
from
a
trusted
source.
So
I
guess
that's
completely
in
the
same
area,
because
pipelineable
is
something
that
is
trusted
the
people
committing
code.
There
need
review
approval
tests,
it's
not
perfect.
Of
course
there
are
always
flows,
but
that's
clearly
a
trusted
trusted
area.
A
C
C
A
Okay,
there
is
no
need
to
add
one
at
once,
because
we
don't
do
on
the
packer
images.
For
today.
That's
that's
a
medium
term
objective.
We
had
trying
to
generalize
so
as
soon
as
we
know
that
the
maven
version
can
be
has
to
be
changed.
If
there
is
a
cv
or
an
issue,
there
are
two
only
location,
the
template
for
the
agent
that
should
be
centralized
in
upcoming
weeks
and
the
pipeline
library.
That's
more
than
fine.
For
me,
that's
enough
because
it's
just
we
need
to
be
able
to
know.
C
Yeah
I
mean
there
could
be
a
concern
there
if
this
ends
up
being
adopted
broadly
by
plug-ins
by
plug-in
maintainers
who
decide
to
use
the
maven
wrapper,
but
then
declined
to
keep
it
up
to
date.
So
that
would
be
if,
if
that,
if
that
happens,
then
that
might
be
a
concern
for
us,
so
we
might
want
to
discourage
them
from
from
using
it.
If
we
don't
think
that
they're
responsible
about
updating
it,
but
that's
to
be
fair,
that
same
concern
applies
to
any
dependency,
that's
pulled
in
by
a
plug-in.
A
A
A
Yes,
that's
fine!
Many
thanks.
B
And
damian,
I
had
one
item
as
well
that
I
I
tucked
into
the
upper
upper
section
just
or
are
you
okay,
if
I
bring
a
topic
so
so
up
earlier
in
the
topic
in
the
completed
we've
now
added
a
a
c
name
record
to
the
dns
that
is
crowdin.jenkins.io.
B
To
help
us
do
a
more
effective
job,
translating
jenkins
components
into
various
local
languages,
so
french,
german,
italian,
japanese,
etc.
Alex
brandis
known
as
not
my
fault
on
github
is
helping
us
go
through
this
exercise.
We'll
do
a
jenkins
online
meetup
to
show
just
how
much
better
it
is
to
use
this
tool
to
translate.
B
B
And
yes,
no
no
need
to
show
anything.
I
just
wanted
to
be
sure
that
the
infra
team
is
aware
that
that
we're
doing
this
effort-
it's
not
done
yet.
It's
still
very
much
in
prototype
phase
right.
We've
had
a
few
plugins
that
we've
translated
it's
working
very
very
well,
but
we've
got
to
get
the
authentication
worked
out
correctly.
We've
got
more
changes
that
need
to
be
done.
It
just
looks
very
promising.
A
So
there
is
an
item
thanks.
That's
a
good
point,
because
I
was
alone
that
day
so
yeah
we
added
a
dns
record
to
make
it
better.
There
is
that
ongoing
work,
that's
really
important
to
to
share
it,
and
I
know
that
there
have
been
an
item
or
request
from
alex.
A
B
That
was
only
a
joke.
Yes,
no,
no
worry!
I
wasn't
serious
at
all,
so
so
to
advise
question
will
will
plug-ins?
Will
it
replace
the
translation
assistance
plug-in
easily
and
with
with
great
improvement
in
experience
yeah?
So
now
we
want.
We
want
testers
to
tell
us,
because
the
languages
I
speak
fluently
are
one
and
it's
already
the
native
language
of
jenkins.
Therefore
it
doesn't
help,
whereas
those
of
you
who
natively
speak
french
or
other
languages,
it's
it'll
be
a
great
help.
We'd
love
to
have
you
as
part
of
the
test.
F
Yeah,
I
noticed
it
on
the
dummy
plugin,
the
helpers
localize
this
page,
and
so
this
pop-in
this
pop-up
coming
in
and
yeah.
I
never
saw
it
before
and
I
wondered
about
the
crowding
integration.
F
B
Yeah
right
so
so
we
would
love.
There
will
certainly
be
plenty
of
places
to
make
that
more
visible,
but
given
the
success
of
the
hacktoberfest
2021
french
translation
effort
led
by
angelique,
jard
and
duchess
france,
I
think
that
we've
got
good
candidates
to
help
us
do
a
really
good
job
of
making
the
translation
experience
much
better
for
jenkins
contributors.
A
Thanks
a
lot
just
a
few
new
items
to
add
to
this
milestone,
we
had
a
request
from
someone
proposing
to
add
the
jenkins
mirror
in
singapore.
So
thanks
harvey
for
adding
the
issue,
so
that's
something
that
will
be
treated
if
there
are.
If
there
is
someone
motivated,
I
will
add
it
to
the
milestone.
Anyone
interested
assign
yourself
to
the
issue-
and
I
will
bear
with
you
them.
B
A
Okay
but
nice
thing
to
reminding
us
about
alibaba
it
was
it.
Was
it
left
my
mind
to
ask
them?
What
is
the
status.
A
Of
their
mirror,
there
has
been
a
request
from
lick
about
replacing
the
default
display
url
from
the
jenkins
check
notification
for
ci
jenkins
io.
So
that's
the
developer
are
not
redirected
to
blue
ocean,
but
to
the
the
usual
ui.
Instead,
it's
not
the
request
to
remove
blue
ocean
from
ci
gents
io,
I'm
not
sure
if
it's
technically
possible
to.
I
assume
there
is
a
github
check
setup
to
be
done
somewhere.
A
C
There's
a
java
property
that
you
could
set.
If
I
remember
from
reading
the
ticket
that
would
determine
which,
which
ui
you
would
get
by
default,.