►
From YouTube: Managing community backlog with the help of metrics | Ray & Alberto | CHAOSScon EU 2020
Description
Managing community backlog with the help of metrics
One of the ways you can measure responsiveness to community contribution is by looking at the size of the backlog periodically. If the backlog of community contribution keeps growing without contributions being closed/merged in a timely manner, this will lead to dissatisfaction among community members. In this talk, Ray will discuss why “Community contribution” Merge Request (MR) backlog has become important at GitLab and how he worked with Alberto at Bitergia to develop visualizations for MR backlogs across GitLab projects. Alberto will discuss why current GrimoireLab dashboards needed to evolve to support GitLab’s use case (beyond GrimoireLab’s existing backlog focused dashboards), technical challenges and limitations when developing the visualization (e.g. calculating the number of contributions that were in opened state on the fly within a given time range that can be selected by the user), and plans for upstreaming this work to CHAOSS.
Slides: https://chaoss.github.io/website/CHAOSScon/2020EU/slides/Managingbacklogs.pdf
-----
Ray Paik
Community Manager - GitLab
@rspaik
Ray is a Community Manager at GitLab where he is helping to grow the community of contributors to GitLab. Prior to GitLab, Ray was responsible for the day-to-day operation of the Open Platform for NFV (OPNFV) community since its launch in 2014. He has over 15 years of experience in the high-tech industry in roles ranging from software engineer, product manager, program manager, account manager, and team lead at companies such as EDS, Intel, Linux Foundation, and Medallia. Ray lives in Sunnyvale, CA with his wife and daughter and all three are loyal season ticket holders of the San Jose Earthquakes soccer team.
Alberto Pérez García-Plaza
Software Engineer - Bitergia
@alpgarcia
Alberto Pérez García-Plaza is currently working as part of support and consulting teams at Bitergia. He is involved in helping customers to deal with data and compute metrics on top of them. He has over 10 years of experience working with data in fields from automatic document organization to cyber security intelligence tasks. He has also a Ph.D. in Artificial Intelligence related to web page representation by means of fuzzy logic oriented to clustering tasks.
-----
About CHAOSScon
Learn about open source project health metrics and tools used by open source projects, communities, and engineering teams to track and analyze their community work. This conference will provide a venue for discussing open source project health, CHAOSS updates, use cases, and hands-on workshops for developers, community managers, project managers, and anyone interested in measuring open source project health. We will also share insights from the CHAOSS working groups on Diversity and Inclusion, Evolution, Risk, Value, and Common Metrics.
https://chaoss.community/chaosscon-2020-eu/
A
So
I'm
very
Paik
I'm,
a
community
manager
at
gitlab
and
I,
have
the
honor
of
speaking
would
call
zorrito
I.
Think
I
mean
I
worked
with
the
turgid
folks
for,
like
five
years
now
like
it
was
Charlie.
Remember
if
I've
ever
done
a
co
presentation
with
the
treasurer
I
think
this
is
the
first
time
so
yeah
I,
don't
know
what
what
it
took
me
so
long.
But
here
we
are
so
I'm
gonna
talk
about
being
responsive
to
community
contributions
and
one
of
the
areas
that
I
started
looking
at
last
year
was
backlog
and
I'd
ping.
A
Alberto
I
thought
it
was
a
simple
question
and
usually
when
I
ask
a
question
Alberto,
he
says
here's
your
visualization
I'm
done
right,
but
this
one
comparison
came
back
and
said:
I
have
to
think
about
this,
so
I
thought
it
somehow
stumbled
onto
a
more
challenging
visualization
than
I
anticipated.
So
we'll
talk
about
you
know,
I'll
talk
about
seven
minutes
about
our
use
case
at
gate
lab
and
then
I'll
turn
things
over
to
Alberto.
A
You'll
talk
more
about
the
technical
implementations,
I'm
think
I'm,
probably
preaching
to
the
choir,
because
a
lot
of
you
folks
are
very
passionate
about
open
source.
I
mean
responsiveness
to
contributions
for
community
members
is
extremely
important,
especially
for
projects
that
are
based
on
led
by
a
single
company
like
yet
lab,
because
if
especially
the
new
contributors
are
just
doing
this
after
they
submit
their
pull
requests
or
merger
quest
like
this
is
not
a
good
experience
and
really
easy
way
for
people
to
be
discouraged
and
then
leave
and
not
come
back.
A
So
obviously
that's
not
something
you
want
to.
You
want
to
see
happen
and
if
you've
been
using
grimoire
laugh
or
for
a
number
of
years
I
mean
you,
you've
probably
seen
similar
like
visualization,
so
it
basically
gives
you
like
a
median
or
average
time
it
takes
for
pull
requests
or
merge
request
be
merged
into
an
actual
product.
So
I
got
a
couple
of
views
here
for
gitlab
and
the
one
on
your
left.
Sorry.
A
So
this
gives
you
like
a
media
and
open
days
for
it's
called
milestone
here.
But
this
is
basically
our
releases.
We
have
monthly
release
cycles
so
for
each
release.
It
tells
you
how
many
take
how
many
days
it
took
for
merger
quest
to
be
merged,
and
fortunately,
ever
since
I've
been
here
for
about
18
months,
it
fluctuates
between
like
a
three
and
a
half
to
five
days,
which
I
think
it's
pretty
healthy.
So
hopefully
we
can
sort
of
maintain
that
good
standard,
but
that's
sort
of
the
medium
view
day
and
one
on
your
right.
A
A
So
these
visualization
is
very
helpful
because
I
can
compare
like
between
releases
right,
for
example,
this
hasn't
happened.
Yet
if,
if
it's
been
averaging
three
to
four
days
and
all
of
a
sudden
forward
the
new
release,
it
goes
to
like
ten
days
and
we
may
have
a
problem
right.
So
it
gives
me
a
little
more
that
I
should
I
should
probably
look
at
what's
going
on
and
even
amongst
projects
I
mean
there's
a
project
like
runner.
The
challenge
there
has
been,
we
pretty
much
had
like
two
full-time
engineers
supporting
community
members
on
the
runner
team.
A
So
that's
why
the
numbers
off
compared
to
some
of
the
others
that
gives
you
a
chance
to
sort
of
you
know,
compare
and
contrast
like
between
projects
like
how
things
are
how
things
are
going.
So
this
is
great.
This
is
an
awesome
tool
that
I
that
are
probably
uses
more
than
other
ways
to
look
at
backlog,
but
the
challenge
is
for
the
gitlab
project.
The
average
open
days
have
been
so
far
has
been
like
14
15
days,
but
what
was
it
like
about
a
year
ago?
A
The
reason
why
this
was
important
last
year,
a
couple
of
things
have
happened.
I
mean
they're
a
couple
of
different
ways
for
your
back
laws
to
grow
I
mean
one
is
because
you
were
able
to
attract
a
lot
of
contributors,
your
backlog.
This
grows
because
there's
a
lot
of
inbound
contributions
coming
in
and
that
was
sort
of
reflected
on
the
chart
on
the
right
we
were
about.
We
doubled
the
number
of
contributors
like
last
year
from
twenty
to
twenty
to
twenty
nineteen,
which
is
great
but
I
was
concerned
about.
A
Are
we
gonna
be
able
to
digest?
All
of
this
that
are
coming
in?
That's
one
to
a
couple
of
people
asked
me.
You
know
how
many
employees
we
have
gitlab
ever
since
I.
Didn't
get
lab
we
more
than
triple
the
number
of
employees,
which
is
great,
so
we're
close
to
about
1200
people
now.
But
what
happens
is
that
it's
great
that
people
are
coming
on
board,
but
it
takes
people
a
few
months
to
get
productive
right,
which
is
kind
of
natural,
and
roughly
I
mean
it
depends
on
your
experience
for
a
lot
of
new
engineers.
A
A
He
would
assign
it
to
a
more
junior
person
that
came
on
board
so
that
that
person
had
can
have
experience
managing
community
contribution,
but
it
it
took
more
than
a
day
which
is
natural
like
for
them
to
review
and
assign
it
back
to
Evan
to
do
a
final
March
right
so,
rather
than
taking
it
about
a
day,
it
would
take
about
two
to
three
days.
So
it's
getting
concerned
about
our
our
ability,
you
know
ability
to
service
all
the
community
contributions
that
come
in,
because
this
is
happening
across
like
different
product.
A
You
can
start
for
all
the
engineering
themes,
we're
getting
a
lot
of
new
people
and
they're
not
able
to
like
process
these
or
review
these
as
quickly
as
more
experienced
people.
So
I
wanted
to
see
you
know
what's
happening
with
our
backlog
is
too
steep.
These
two
things
are
happening
and
just
one
final
slide
before
I
turn
things
over
to
Alberto.
So
this
this
gives
you
a
cumulative
view
of
how
what's
what's
happening
with
the
backlog.
A
So
I'll
I'll
explain
the
different
colors,
the
the
ones
in
red
at
the
bottom
are
those
are
showing
like
clothes,
merger
quests,
for
whatever
reason
we
decided
to
close
them
in
this
position
and
the
one
in
the
middle.
We
have
two
colors
of
merger
quests
that
have
emerged.
The
one
in
blue
I
didn't
even
know
about
this.
Apparently,
at
some
point
we
didn't
keep
track
of
dates
of
when
the
merge
request
were
married,
so
the
ones
in
blue
are
showing
the
requests
that
are
merged
without
dates
and
the
ones
in
yellow
are
ones
with
days.
A
So
basically,
these
two
show
the
merger
quests
that
have
emerged.
So
what
I've
been
focusing
on
is
sort
of
that
green
band.
You
see
at
the
top,
so
I
mean.
Fortunately,
there
hasn't
been
a
whole
lot
of
fluctuation,
but
it
gives
me
a
sense
us
to
you
know
is
that
is
it
the
height
of
the
green
green
area?
Is
that
like
increasing
or
decreasing
over
time?
A
So
this
has
been
pretty
helpful
in
terms
of
sort
of
visual
as
I
visualizing
like
how
we've
been
doing
this
again
is
an
example
for
runner,
because
we
were
so
resource
constrained.
I
was
I
was
concerned
about
this
project.
Typically
like
over
time.
We
average
between
like
a
250
or
260
s
and
Mars
that
are
still
like
open
and
then
I
think
at
some
point,
like
a
keyboard
last
year,
we're
starting
to
keep
about
like
300.
A
B
It's
intended
to
be
me,
you
can
say
so.
Basically,
this
is
not
my
first
talk
here
in
chaos
cone,
but
is
the
first
one
I
share
with
someone
which
is
not
part
of
Victoria.
It's
a
no
for
me
because
we
have
been
working
out
of
time
with
with
ray,
and
now
we
did
love
what
the
problem
with
the
backlog
we
have
in
ramallah.
Right
now
is
this.
B
Does
works
are
basing
on
what
remains
open
today,
so
you
can
see
a
lot
of
things
here,
but
the
problem
is
the
question:
you're
answering
is
how
many
items
created
ring
whatever
period
of
time
you
select
here.
Art
is
still
open
today,
and
this
is
not
the
real
backlog.
Because
do
you
want
to
analyze
the
past?
Do
you
want
to
analyze
all
the
backlog
was
the
previous
moments
of
the
previous
year
or
whatever?
In
other
words,
what
we
have
here,
at
least
these
items
here
were
created
on
these
days.
B
This
means
this
amount
of
items
are
still
open
and
were
created
here,
but
this
is
not
the
real
Babylon
and
that
was
a
problem
I
faced
when
Ray
talked
to
me
to
try
to
get
a
back
log
out
of
the
you
more
lab
data
and
it's
not
an
easy
problem
to
solve
at
least
not
easy
with
the
current
data.
But
the
problem
is:
the
limitations
are
very
strong
here,
because
we
cannot
analyze
the
evolution
over
time
and
we
cannot
analyze
also
the
status
of
the
backlog
at
a
specific
period
of
time.
B
So
what
do
you
have
a
vision
of
you
of
your
backlog
right
now,
but
anything
else
and
again,
this
is
the
question
we
are
answering
with
the
current
use
case.
We
have
in
more
lab
and
now
I'm
I
want
to
tell
you
about
the
human
story,
about
behind
the
scenes
of
all
of
these
process.
The
process
there
was
this:
a
couple
of
good
lab
users,
probably
talking
between
them
and
realizing
that
the
Battle
of
is
but
I
said
and
not
what
they
expected,
and
the
next
step,
as
Ray
said,
was
exactly
this.
B
Have
to
say
that
I'd
like
to
thank
you
to
me
Alison
for
further
out
so
forward,
and
what
the
next
question
was.
What
can
you
do
and,
of
course
everyone
know
about
this
tip,
don't
know
travel
time,
but
I
try
I
asked
my
boss,
man,
Ricky
I,
say
hey
man,
Ricky
I
want
to
travel
in
time
to
get
the
bottle
of
give
it
one
bigger
Tommy.
Yes,
you
can
do
it,
but
you
need
to
send
a
talk.
I
didn't
have
time
to
Senna
talk
to
the
class,
so
I
directly
try
to
find
on
the
solution.
B
B
The
second
one
was
also
what
waited,
and
now
it's
under
review
in
a
non
issue
in
Ramallah,
because
another
juicer
followed
that
the
second
one
is
valid,
but
it's
based
it
on
when
you
study
that
means
a
new
index
new
data
and
the
product
that
is,
if
I
have
to
say
to
get
lab
hey
we
can
go
for
this.
The
next
thing,
I
have
to
say,
is
what
we
need
to
maintain
a
new
index.
We
need
to
create
a
new
code
and
we
need
money
for
all
of
these.
B
This
one
is
just
doing
thing
things
on
top
of
the
data
we
already
have
so
going
quick
on
about
the
data.
What
we
have
is
forex
with
Merck
us.
We
have
the
time
the
creation
time
of
these
molecules
and
the
date
on
which
the
merger
was
was
merged
or
closed
with
this
data.
What
we
can
do
is
the
cumulative
zoom
of
the
items
as
they
are
created
and
from
from
this
one
we
can
subtract
the
items
that
are
close
or
met
on
each
specific
specific
time
range.
The
idea
of
doing
this
is
early
on
timeline.
B
Timeline
is
a
tool
with
half
in
Havana,
and
you
can
do
this
kind
of
expression.
Basically,
this
expression
says:
hey
you
last
except
took
take
this
in
this
little
at
Mercy,
West
uncounted,
eight
items
depending
on
they
created,
add
field.
So
what
you
have
here
is
the
count
of
items.
Items
that
were
created
here-
and
this
is
pressing
here-
is
the
cumulative
sum
of
all
of
this.
That
means
we
have
one
here
two
here
the
total
is
3.
B
This
is
the
typical
saved
you
can
expect
if
you
keep
adding
things
from
the
beginning
till
the
end
top
of
this,
we
can
subtract
thing
well,
I
think
this
is
obvious,
and
it's
the
same.
Hey
last,
except
from
this
index,
you
can
take
all
the
items
that
were
closer
tormenta,
a
specific
time
frame
and
subtract
them
from
the
creative
ones,
and
on
top
of
these
again
the
cumulative
sum
well
remember.
This
means
that
if
we
open
four
items-
and
we
don't
close
anything,
we
have
four.
B
If
in
t2
you
have
three
and
we
close
to,
we
have
one
new
item
to
zoom
to
the
poor.
We
have
so
we
have
now
five
five
times
and
if
these
three
before
is
just
to
let
you
know
that
you
can
close
more
things
that
you
are
opening,
so
in
this
case,
you're,
improving
your
backlog
and
when
you
have
less
than
before,
and
you
can
just
get
espera
in
this
okay-
and
this
will
be
the
final
say-
oh
you're
back
just
the
last
is
like
Feliz
to
subtract.
B
The
the
corner
day
is
a
very
mentioned
before
for
those
items
that
work
it
created
even
before
the
Murtagh
feel
even
existed
in
gitlab.
So
this
is
the
same.
We
have
right
now
and
the
same
I
saw
before
with
only
counting
opening
things.
Here
is
the
same
because
we
are
stacking,
the
closer
to
merger
than
the
current
is
here
and
if
we
just
hide
all
of
the
clothes
and
merge
items
we
have
this
back.
This
is
just
a
double
check
of
the
data
to
make
sure
that
all
the
calculations
I
having
are
correct.
B
And
finally,
just
as
a
recap
of
all
of
this,
we
started
this
by
solving
the
use
case.
That
Ray
is
playing
it
with
custom
visualizations.
From
that
all
the
information
is
ProPublica
be
available,
so
you
can
go
and
read
all
the
thread
with
a
quick
grid
lab
you
can
go
to
the
more
life
and
read
the
issues
and
in
the
future
you
can
go
and
read
the
pull
request
for
a
panel,
which
is
the
work-in-progress
I
mention
here,
because
we
need
to
build
a
specific
that
were
around
this
visualization.