►
From YouTube: New Sidekiq Cluster configuration with GitLab 13.0
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
Everyone
so
with
collab
13,
we're
switching
from
defaulting.
Do
a
single
psychic
process
to
psychic
clusters,
which
allows
you
to
have
multiple
psychic
processes.
For,
however,
you
define
keys
so
I
want
to
go
into.
You
know
what
that
means.
What
it
looks
like
if
you
use
the
old
config
and
what
are
the
use
cases
we're
actually
using
multiple
as
like
processes?
Alright.
So
let's
get
started
so
here.
A
A
A
A
So
you
here's
our
original
psychic
process.
Here
is
our
psychic
cluster
process,
and
here
are
the
second
cluster.
This
is
for
the
extra
cues
I've
defined
in
my
setup.
So
if
we
look
in
me,
the
old
you
live
are
beating.
So
from
the
1210
see
you
it's
like
a
cluster
enabled
and
then
here
my
key
groups
process
commit
and
post
receive.
Is
one
group,
so
that's
one
process
and
then
get
up
shell,
the
second
process
and
that
lines
up
with
what
we
see
here.
A
Process
of
post
receive
and
collab
shell,
and
then
you
can
see
this
is
your
of
3
is
a
number
of
worker
threads,
so
you
can
tell
that
we
have
its
number
of
Q's
plus
1
up
to
limit.
So
we
have
two
queues
here.
So
we
up
to
three
threads
one
queue
here
so
to
go
up
to
two
threads
back
on
13.
We,
you
know
I,
think
we
just
run
out
of
space
here
if
it's
a
pretty
tall,
the
queues,
but
presumably
actually,
let's
check
this.
A
13
threads
yeah,
so
there's
always
some
amount
of
overhead
of
extra
threads
that
aren't
actually
doing
all
the
work.
That's
good
to
know
so,
presumably
don't
actually
have
60
worker
threads.
Maybe
it's
next
to
relevant
threads
up,
speak
and
check
the
other.
We
write
4,
3,
3,
14,
ok,
yeah,
so
maybe
it's
an
extra
11
threads.
A
A
I'll
make
this
one
kind
of
a
catch-all,
so
they've
got
one
process
for
Caleb
jump
post
receive
two
more
that
just
to
get
up
shell
and
then
this
one
that
does
everything
else.
Just
as
no,
you
can
have
multiple
workers
handling
a
single
queue
if
that
queue
gets
out
of
control.
So
if
you
have
some
kind
of
edge
case,
the
customer,
where
they
have
a
job,
that's
taking
five
minutes
to
run
it's
just
sandbagging
their
psychic
instance.
A
A
A
A
A
A
A
A
A
A
You
through
seven
it's
only
seven,
it
is
interesting.
Maybe
that's
the
maximum
concurrency
I,
don't
know
that
works
in
psychic,
I
know
that
and
you
live
our
B.
You
can
set
minimum
of
max
concurrency.
We
talked
about
that
in
the
docs
or
previously
would
be
here
is
your
concurrency
and
it
was
just
25
or
50
or
whatever,
and
now
we
can
say
down
here,
you
can
have
either
n
plus
1
if
it's
in
that
range
or
they
made
our
max
value.
A
A
A
A
A
A
A
A
A
A
Let
me
now
get
a
nice
message
here
using
psychic
clusters
now
default.
Please
move
your
settings
rid
of
psychic
asterisk
config.
Configuring
cyclist
will
be
removed
in
fourteen
and
then
we
also
somewhat
confusingly
get
a
message
that
running
psychic
directly
is
deprecated
and
we
removed
and
get
get
lab
14
please
consider
using
sexy
cluster,
and
then
we
tell
it
and
say
that
I
don't
again,
psychic
clusters
been
deprecated
and
then
we
check
the
killer,
CTL
status.
A
A
So
now
we
have
what
we
saw
on
the
other
one
which
is
sidekick
just
with
25
threads
there,
and
then
we
also
so
I
could
cluster
as
we
configured,
but
user
should
get
a
nice
warning
when
that
happens.
So
hopefully
they'll
notice
that
just
slightly
confusing
that
you
also
get
told
you
rank
sidekick
but
other
than
that,
it's
basically
the
same
as
running
normal
sidekick
or
as
it
was
before.
So,
if
you
look
at
this
config
and
the
compared
to
what
we
have
here,.