►
From YouTube: Keynotes Día 2 - Alvin Estrada, GuateGeeks; Chen Siyu, PingCAP & Feynman Zhou, QingCloud
Description
Estas fueron las palabras de apertura de Sergio Méndez y Keynotes donde Alvin Estrada nos hablo sobre "DevOps: Mucho mas que k8s, contenedores y pipelines" , Chen Siyu sobre "Disrupt your systems with Chaos Mesh (constructively!)" y Feynman sobre "Processing logs in a Kubernetes-native way with Fluent Bit Operator". También tuvimos un super ameno momento de preguntas. Y fue una sesión donde hablamos español, inglés y nuestros speakers nos dijeron adios en Chino.
Contacta a Alvin en:
- https://www.linkedin.com/in/alvin-estrada-868b22a5/
- https://twitter.com/adawolfs
Contacta a Siyu en:
- https://github.com/iguoyr
Y Feynman en:
- https://www.linkedin.com/in/feynman-zhou-569461178/
- https://twitter.com/FeynmanZhou
A
A
B
B
B
B
B
B
B
B
B
B
C
C
A
D
D
So
how
much
confidence
we
have
in
our
systems?
Here
is
the
example
that
I'm
sure
you
are
all
familiar
with
and
mostly
have
been
affected
by
its
global
outage
facebook
caused
by
the
loss
of
ip
wrote
to
the
dna
servers.
It
happened
just
last
month
and
this
global
outage
lasted
for
longer
than
5
hours,
which
resulted
in
many
means
across
the
internet.
D
D
D
It's
the
discipline
of
experimenting
on
a
system
in
order
to
build
confidence
in
the
system
capability
to
withstand
turbulent
conditions
in
production,
and
it's
quite
difficult
to
understand
how
we
can
actually
implement
chaos
engineering.
Just
by
reading
the
definition,
so
I'm
going
to
introduce
the
phases
of
chaos
engineering.
First,
we
need
to
define
your
steady
state.
For
example,
it
could
be
the
regular
behavior
of
your
service
metrics
after
we
have
the
steady
state
it's
time
to
start
asking
questions
like
what,
if
a
container
stops
working,
what
might
go
wrong
in
our
systems.
D
D
How
much
time
do
we
need
to
recovery
to
a
steady
state
or
how
much
time
do
we
need
to
detect
the
failure
and
the
last
step
is
fix
the
problems,
so
we
can
find
that
the
harder
it
is
disrupts
the
steady
state,
the
more
confidence
we
have
in
the
behavior
of
the
system.
If
our
weakness
is
uncovered,
we
now
have
a
target
for
improvement
before
that
behavior
manifests
in
the
system
at
large.
D
D
It's
easy
to
use
and
it
is
designed
for
kubernetes
natively
and
it
is
open
source.
Currently,
cal
smash
is
cncf.
Sandbox
project-
and
we
are
in
the
process
of
onboarding
with
cncf
incubating
here-
are
some
kel's
experiments
provided
by
chaos
mesh.
The
experiments
are
designed
to
simulate
40
situations,
enabling
you
to
run
tests
such
as
drawing
drawing
for
lost
network
connection
or
file
system
for
it
before
your
system
runs
in
production.
D
D
The
interface
of
kels
mesh
is
similar
to
other
resources
such
as
port
or
deployment,
so
you
can
create
or
delete
cal's
experiment.
The
same
way
we
support
using
the
cube
control
of
other
tools,
and
if
you
would
like
to
operate,
experiment
or
observe
more
details,
you
can
use
the
built-in
kels
dashboard
onto
the
architecture
of
chaos.
Mesh.
The
over
architecture
can
be
divided
into
three
parts.
D
First,
user
input
and
observation
users
can
use
cube
con
cube
control,
command
or
calc
dashboard
to
submit
a
chaos,
experiment
or
workflow
to
kubernetes
api
server,
and
then
cal's
control
manager
will
accept
events
from
the
kubernetes
api
server.
This
event
describes
the
change
of
a
certain
a
certain
chaos
resource,
such
as
the
creation
of
the
chaos
object.
D
The
third
part
is
the
execution
part
kelsmash
will
deploy
a
demon
set
to
set
up
demon
on
every
node.
This
demon
will
receive
requests
from
control
manager
and
inject
failure
on
the
target
and
in
the
next
part
I
will
give
you
an
example
of
how
to
use
chaos.
Mesh
I'd
and
I
think
natural
chaos
is
a
good
start.
D
D
As
you
can
see,
the
right
part
is
the
definition
of
network
health
and
the
cal's
experiment.
Simulates,
the
network
delay
calc
mesh
use
different
selectors
to
define
the
scope
of
photo
injection
here.
Labels
lectures
will
select
port
with
fdb
label
and
the
mode
that
represents
that
z-scale
experiment
will
select
one
randomly
from
all
ports
that
meet
the
defined
selectors.
D
So
now
we
have
our
example,
applications
and
the
chaos
experiment
manifest
how
to
actually
run
the
experiment.
First,
we
need
to
check
the
steady
state
without
the
calf
mesh
experiment
in
if,
in
this
example,
let's
assume
there
are
two
apis,
the
get
api
for
acquiring
data
and
the
port
api
for
modifying
data
before
the
injection
starts.
The
steady
state
of
the
application
should
be
that
both
of
two
apis
responds
properly
and
next
we
will
apply
the
chaos
mesh
experiment
to
kubernetes
cluster.
D
D
D
D
D
D
Besides
latency
calc
mesh
can
also
inject
package
laws
and
package
duplication
with
the
help
of
network
emulator.
Queueing
discipline
and
calc
mesh
can
also
shape
network
bandwidth
with
tbf
traffic
control.
This
can
help
you
emulate
different
connection
quality
between
multiple
machines
or
multiple
data,
centers.
D
We
will
continue
to
focus
on
ease
of
use
and
observability,
such
as
we
are
planning
to
manage
and
schedule
chaos
on
both
kubernetes
and
non-kubernetes
targets
on
the
unified
dashboard
and
providing
a
more
comprehensive
status,
inspection
mechanism
and
reports
and
improve
visibility
of
cal's
experiment
via
events
logs
and
metrics,
and
regarding
to
security.
We
plan
to
improve
the
authorization
system
and
the
support
users
to
log
into
cals
dashboard,
using
their
gcp
or
aws
accounts.
D
Types
user
can
use
this
plugin
to
develop
their
own
chaos
types
such
as
rabbit
mq,
kels
and
the
radish
mq,
and
a
hub
is
also
planted
for
users
to
share
their
own
health
types
and
workflow,
which
can
reduce
the
cost
for
new
users
to
get
started
with
scales,
experiments
and
you
can
follow
along
our
journey
at
the
chaos
mesh
roadmap
and
we
recently
released
kel's
mesh
version
2.0
and
we
plan
to
release
version
2.1
shortly.
You
can
experiment
experience
the
new
version
with
pre-configured
environments
on
the
cloud
and
you
can
try
out
chaos.
D
Mesh
version,
2.0
interactive,
send
rows
from
our
your
browser.
There
is
a
link
on
the
cal
smash
website
and
if
you
are
interested
in
cal
smash,
but
don't
know
how
to
get
started,
go
to
the
cal
smash
github
page
and
we
have
tagged
a
number
of
issues
suitable
for
beginners
with
a
table
good
first
issue,
and
it
can
help
you
to
get
familiar
with
scale
smash
and
if
you
have
any
questions
or
suggestions,
feel
free
to
discuss
with
us
via
the
following
methods,
and
we
will
have
monthly
cal
smash
community
meetings
to
share
recent
activities.
D
A
A
A
D
A
E
D
D
D
D
A
A
D
D
A
Yeah,
that's
that's
really
really
really
really
nice.
I
think
that
there's
a
lot
of
people
that
are
involved
behind
chaos,
mesh
and
yeah,
for
example.
Maybe
I
can
tell
my
story
behind
that.
Well,
we
were
working
with
some
student
about
chaos,
engineering
and
I
remember
that
I
started
to
talk
with
the
people
of
chaos
mesh
in
the
slack
channels
of
cncf
and
I
was
like
well.
A
A
A
D
D
D
A
Chao,
spanish:
I
think
that
is
a
break
that
is
growing
really
fast.
I
remembered
like
a
few
months
ago
right
now
you
are
improving
the
things
adding
more
experiments
and
yes,
it's
really
nice.
I
remember
that
you
have
a
feature
to
create,
like
your
custom
experiments
using
the
kernel
and
that
things,
so
it's
pretty
interesting
to
to
play
a
little
bit
with
the
kernel
to
have
some
or
explode
some
problems
inside
your
house
and
everything
so
yeah.
That's
that's
really
really
nice.
A
Well,
let's
say
that
we're
adding
this
thing,
if
you
tweet
with
the
hashtags
already,
you
can
win
a
book,
maybe
you
can
do
it
with
the
hashtags
kcdgt
and
already,
and
maybe
you
can
win
a
book,
we
are
going
to
contact
you
and
you
can
get
whatever
book.
You
want
from
irely
media
and
we
are
going
to
send
you
the
book
on
digital,
so,
okay,
yeah,
great
yeah.
So
thank
you
very
much
and
see
you
soon
and
thanks
for
your
participation
in
this
keynote.
A
A
So,
let's,
let's
remember
well
to
mention
something:
well,
we
have
a
final
concert:
music
show
at
the
end
of
the
event
at
6,
30
p.m,
gmt,
six,
the
guatemala
time
it's
more
or
less!
In
china,
it's
like
nine
or
or
or
10
a.m,
more
or
less,
and
you
want
to
join
us
from
other
countries
from
spain
from
united
kingdom
and
yeah.
We
are
going
to
wait
a
little
bit
for
our
presenter
fame
and
shoe
from
cube
sphere
and
yeah.
We
have
like
some
technical
problems
just
giving
some
time
there
yeah.
A
A
A
A
Yeah
so
fame
and
just
giving
some
seconds
more
well,
so
we
have
a
lot
of
sessions
today.
Let
me
read
about
these
sessions.
Well
right
now
we
are
in
parallel
in
separate,
separate
tracks.
You
want
to
wait
for
feyman.
We
can
stay
on
this
track.
If
you
want
to
move
to
another
session,
we
have
right
now.
Let
me
check
we
have
the
introduction
to
container
d
is
a
session
in
spanish.
A
A
Really
nice,
then
we
are
going
to
talk
about
the
people
from
restaurants.
Red
hat
is
going
to
well
welcome
to
fame
and
it's
right
now.
That's
really
nice!
Well,
hey!
You
know
right
now
we
are
live,
so
maybe
we
can
experiment
some
problems
so
welcome
to
fame,
and
maybe
you
can
share
your
screen
now,
it's
time
to
another,
really
really
awesome
project
from
china
cubesphere.
Let's
say
that
is
the
openshift
from
china
and
the
rancher
of
china.
So
really
really
nice,
this
topic.
A
E
E
E
In
today's
session,
I'm
gonna
show
show
you
the
challenges
of
the
the
general
challenges
of
building
the
login
layer
in
kubernetes,
and
next
we
are
going
to
introduce
to
the
front
bit
and
next
we
are
going
to
showcase
the
general
solution
that
we
built
on
front
bit
of
front
bait
yeah.
The
name
is
from
base
operator
and
finally,
we're
going
to
show
you
the
deep
dive
into
the
front
base
operator,
and
it
is
use
case
and
our
contribution
to
this
project.
E
Well,
I
think
I
believe
you
guys
must
has
known
this
drama
television
series,
yeah
game
of
thrones
yeah,
you
maybe
know
the
scared,
white
workers
and
yeah.
I
think
the
challenges
of
kubernetes
logging
is
very
similar
to
the
game
of
thrones
and
you
might
hear
a
lot
of
complaints
from
the
white
walkers
yeah.
E
The
some
guys
complained
that
we
have
different
data
sources
and
we
need
lightweight
security
and
resilience
on
the
logging
in
kubernetes
and
we
still
have
different
data
formats
and
we
have
we
need
to
send
the
data
and
logs
to
different
outputs
and
destinations
yeah.
All
of
those
complaints
we
heard
about
our
users
and
customers
and
from
the
industry.
E
E
E
E
E
Last
but
not
least
from
bit
is
it
has
high
support
with
low
cpu
and
memory
usage,
so
this
is
the
general
workflow
of
from
bit.
It
can
run
as
the
demo
set
on
each
node
and
to
collect
and
gather
the
blocks
from
each
kubernetes
node
and
then
send
the
logs
and
the
data
to
multiple
outputs
and
destinations
yeah.
We
have
listed
the
splunk
kafka,
elasticsearch,
s3
and
so
on.
E
E
E
E
E
E
So
why
we
create
the
front
bait
operator
so
initially
we
need
to
control
the
front
bait
while
the
native
kubernetes
apis
and
we
would
like
to
deploy
and
destroy
the
front
bait
demo
set
automatically
while
the
front
bit
operator.
So
this
is
the
general
and
initi
initiate
purpose
for
this
project
yeah,
as
we
mentioned
earlier.
You
know.
The
dynamic
reloading
is
not
supported
in
the
front
bait
itself,
so
we
would
like
to
support
the
dynamic
reloading
in
this
project
and
and
enable
it
to
update
the
configuration
without
rebooting
from
bait
parts.
E
E
E
You
know
the
front
bit
from
bit
config
and
the
four
crds
that
that
includes
the
input
password
filter
and
the
output
yeah
for
the
first
one.
The
front
bit
is
used
to
define
the
frame
based
demo
set
and
config
and
front
page
config
is
used
to
select
the
input,
filter,
output
plugin
and
generate
the
final
configuration
into
a
secret.
E
And
for
those
plugins,
which
is
used
to
define
the
respective
config
sections
that
can
be
selected
by
the
label
selectors
in
the
front
bait
and
finally,
you
can
use
the
outputs
plugins
to
transfer
those
logs
blocks
and
datas
to
different
destinations.
You
know
the
elastic
search,
kafka,
graph,
analogy
and
s3.
E
All
of
these
crds
were
watched
by
the
front
base
operator
and
as
soon
as
the
configuration
has
any
change,
the
frame
bit
operator
can
create
a
secret
that
include
the
changes
and
the
circuit
can
be
mounted
to
the
front
base
thermostat.
Finally,
so
in
this
workflow,
if
anything
has
changed
in
the
crds
or
the
configuration
this,
this
operator
can
automatically
pick
it
up
and
mount
them
into
the
front
page
demo
set
and
port.
E
Yeah,
let's
give
that
a
little
bit
more
about
the
wrapper
of
the
front
bit
operator
so
how
we
implement
the
front
bit
operator
to
to
make
it
dynamically
dynamically
reconfiguration
yeah.
We
have
the
from.
We
have
written
the
frame
bait
watcher,
which
is
used
to
watch
the
front
bit
part
and
the
secret
and
to
watch
the
secret,
and
this
configuration.
E
E
Last
but
not
least,
I
would
like
to
demonstrate
how
we
build
a
kubernetes
login
layer
with
the
phloem
bait
operator
project,
so
yeah.
This
is
a
general
use
case
in
cubesphere,
so
cubesphere
is
an
open
source
container
platform
and,
as
sergio
mentioned
before,
cubesphere
is
a
little
bit
similar
to
rancher
and
red
hat,
open
shift,
which
is
used
to
deploy
and
package
the
containerized
application
to
the
kubernetes
yeah.
E
So
cubesphere
has
its
own
login
console
and
the
logging
system,
which
is
built
on
the
front
bait
and
from
bit
operator
as
we
can
see
from
this
graph
cubestream,
has
its
login
console
on
the
top
layer.
So
users
can
query
the
logs
and
the
data
from
the
cubesphere
console
and
this-
and
this
is
implemented
by
the
cubesphere
api
server.
So
the
login
console
will
call
the
rest
api
to
from
the
api
server
and
the
crd
will
manipulate
the
front
bit
running
on
each
node.
E
So
this
is
the
we
can
say.
The
front
bait
operator
is
running
as
a
log
agent
and
it
it
can
run
from
bit
as
a
demo
set
on
each
node.
Instead
of
running
the
front
d
on
each.
Only
single
node,
so
it
can
avoid
the
single
node
failure
risk
at
same
time.
Cubesphere
provides
this
centralized
load
query
so
it
and
it
enables
you
to
build
these
customized
log
logging
console
and
it
allows
you
to
query
the
logs
while
different
dimensions.
So
that
means
you
can
search.
E
E
E
E
E
So
let's
go
back
to
the
cubesphere
console
here
we
have
those
crds
and
each
id
has
its
own
configuration.
We
can
enter
into
the
typical
crd
from
page
config
yeah,
as
we
mentioned
before,
those
plugins
are
controlled
by
the
front
bait
config
and
you
can
select
those
plugins
while
the
label
selectors.
E
E
E
E
A
E
Can
go
through
and
drill
into
a
container
yeah.
We
can.
Oh,
we
can
enable
the
real-time
refresh
and
we
can
see
the
real-time
blocks
that
generated
by
the
google
container,
since
we
have
enabled
the
systemd
logs
and
convert
it
into
this
standout
container
logs
format.
So
we
can
see
those
container
logs
in
this
console.
E
E
E
Yeah,
finally,
I
would
like
to
share
addition.
Some
I
would
like
to
share
some
additional
resources
which
helps
you
to
learn
more
about
the
operator
pattern
and
the
front
base
operator
yeah.
Actually,
we
write
the
front
bait
operator
with
the
cube
builder,
so
I
would
like
to
share
the
computer
documentation
and
the
basic
introduction
to
the
kubernetes
operator
and
his
use
case,
and
this
is
the
address
of
the
front
bit
operator,
so
you
can
check
it
out
after
you
get
this
slide.
E
E
A
Hi,
I
am
back
well,
so
you
have
any
question
for
trainman
sue.
You
can
write
in
the
chat.
We
can
read
the
questions
in
advance.
I
I
really
appreciate
your
participation
here
in
this
event
and
well,
as
we
know.
Well,
sometimes
we
have
think
our
problems.
I
want
to
add
see
you
sometime
here
see
you.
Can
you
share
your
your
camera
situation?
A
Nice
well
yeah.
So
I
really
appreciate
that
people
from
china
is
sharing
their
products.
I
really
like
the
diversity
and
everything.
So
I
really
appreciate
this.
I
I
got
I
have
to
say
thank
you
for
your
participation
cyu.
Thank
you
to
you
feyman.
So
for
me,
it's
like
really
really
nice
to
have
you
here
guys.
So
I
really
appreciate
that
part.
I
think
that
there
are
no
questions,
maybe
because
for
the
time
and
well
so
in
general.
Thank
you
very
much.
I
I
I
want
to
improvise
something
right
now.
A
A
A
That's
nice
yeah,
so
thank
you
very
much
guys
for
your
presentation.
Yo
pablo
pacheco
is
like
giving
thank
yous
here
so
yeah.
Thank
you
very
much.
I
really
appreciate
this
presentation
and
well
see
you
in
another
event.
Maybe
why
not
in
a
cubecon,
china
or
something
so
yeah,
really
nice
guys.
So
thank
you
very
much
for
participation.