►
From YouTube: Ceph DocuBetter Meeting 2019-01-18
Description
B
B
The
problems
that
we're
facing
are
that
we
host
documentation
for
all
the
different
versions
of
Ceph
so
like
mimic
master,
but
all
the
way
back
to
like
bobtail,
it's
like
online,
and
if
you
go
to
Google
and
start
asking
questions
about
different
parts
of
stuff,
you
may
get
directed
to
master
or
you
might
get
directed
to
something
from
eight
years
ago.
It's
like
it's
a
little
bit
of
a
problem,
so
we
figure
there's,
probably
things
we
want
to
change
on
this.
B
Like
way
we
structure
the
documentation,
but
then
we
figured
there's
also
parts
of
like
metadata,
or
something
like
this-
that
we
want
to
expose
at
Google
and
other
search
engines
that
they
like
and
do
a
better
job
of
figuring
out.
What
we
want
to
give
users
and
I've
read
a
little
bit
about
it,
but
you
posted
in
on
the
mailing
list
about
something
called
canonical
and
then
so
we
invited
you
because
you
seem
to
know
way
more
than
then.
We,
though,
we
would
like
to
learn
from
you
sure.
D
C
You
know
link
in
the
head
or
in
head
section
of
the
HTML,
and
then
that
will
you
know
redirects
the
search
engines
to
to
you
know
treat
this
as
the
canonical
version
of
the
page.
So
I
was
thinking.
We
would
probably
want
to
do
that.
First
stuff.
You
know
in
order
to
keep
it
very,
very
lightweight
and
minimal.
C
You
could
do
that
just
for
master
and
have
you
know
all
the
docs
appoints
a
master
is
the
canonical
version,
because
that
branch
name
will
never
change
and
then
I
noticed
that
Apache
does
something
a
little
bit
different
to
you
where
they
have
like
they
have
like
a
current
alias.
So
they
you
know
like
if
you
go
to
slash
current
for
their
Doc's
I,
guess
it
gives
you
the
most
stable
one.
I
think
ansible
does
this
as
well
for
their
Doc's.
So
that
would
be
another
idea
if
you
wanted
to
get
more
fancy.
C
B
So
I'm
not
sure
how
much
of
the
improvements
you've
been
following.
The
the
last
round
of
improvements
we
merged
at
the
end
of
last
year
were
a
system
for
putting
banner
ads,
not
banner
ads.
There's
like
banners
on
all
the
pages.
So
if
you
visit
master,
for
example,
it'll
tell
you
that
you're
looking
at
a
development
version
or
if
you
visit
nimac
it'll,
say
this
is
a
stable
and
current
boarded
version
and
the
reason
I
mentioned.
That
is
because
it
seems
like
choosing
what
version
is
used
for.
B
C
E
B
E
D
B
E
E
B
A
B
A
C
Yeah
my
suggestion
with
that,
if,
like
you
do
end
up
using
like
a
symlink
or
whatever,
is
just
try
to
make
Spanx
do
that
or
whatever
it
is
the
Jenkins
job.
That's
running.
Do
that
for
you,
instead
of
like
asking
David
Galloway,
you
know
or
whoever's
running
the
server
like
hey.
Can
you
go
update
that
so
you
can
fake
or
nginx
configure
whatever
it
is
like?
Can
you
ssh
into
this
thing
that?
Because
then
people
will
forget
right,
because
we
only
bump
that
every
like
here
or
something.
A
C
B
C
B
C
A
B
The
other,
the
other
thing
that
we
had
mentioned
in
the
mailing
list
about
SEO,
was
using
the
I
think
it's
called
the
robots.txt
file,
which
can
be
used
to
do
things
like
tell
a
search
engine
to
completely
ignore
certain
prefixes
and
I
feel
like
that
might
be
really
powerful
like
in
some
sense.
We
could
just
tell
it
to
ignore
bobtail
and
Emperor
right.
It's
like
those
things,
I'm,
not
sure
we
even
want
to
index
them,
but
maybe
somebody
if
somebody
knows
more
about
robots
or
maybe
there's
other
ways
to
use
it.
C
That's
a
good
idea
to
put
in
bobtail
cuttlefish.
Forget
the
dumpling.
You
know
all
the
way
out,
basically
up
to
hammer
I.
Imagine
you
wouldn't
want
to
index
and
then
yeah.
You
could
probably
have
Jenkins
write
that
file
out,
for
you
just
add
it
onto
the
end
of
that
job.
That's
generating
those
it
basically
just
runs
Sphinx
or
whatever
it
is.
It's
like
make
docks
or
something,
and
then
it
does
in
our
sink
into
the
docks
server
and
it
puts
you
in
the
HTML
pages
on
the
server.
B
Happen
to
know
if
I
mean
this
is
probably
really
easy
to
just
go
look
up,
but
if
we
could
whitelist
instead
of
blacklist
and
the
reason
I'm
thinking
about
that
is
because
you
know,
Jenkins
will
also
be
building
all
the
branches
and
stuff
like
that,
and
so
it's
like
I
suspect.
Google
doesn't
pick
that
robots
up
that
fast,
like
faster
that
we
can
still
do
stuff
yeah.
E
B
C
E
C
To
write
that
down,
because,
like
this
change
hands,
even
the
docs
have
change
hands.
So
much
like
alfredo
worked
on
a
project
to
do
this
long
long
ago
with
you
know,
like
maybe
picking
the
versions
dynamically
and
that
was
pretty
under
documented
and
then
like
just
having
like
a
standard
thing
of
like
this
is
what
we
do.
Every
time
we
make
create
a
new
branch
like
for
octopus.
You
know,
for
example,
here
like
this
is
and
and
what
that
means
for
the
dockets
I'm.
B
B
E
E
D
This
person
work
for
you
from
June,
July
and
August
40
hours
a
week
on
whatever
documentation
related
stuff.
You
guys
could
deem
necessary
as
long
as
you
figure
that
out
and
the
intern
could
write
like
actual
actual
Docs
or
they
can
work
on
Doc's
generation
stuff,
either
or
whatever
upstream
feels
most
fit,
and
this
internship
also
come
to
this
meeting
every
week.
D
B
E
D
Go
ahead,
no
I
think
we
I
know
at
least
I
always
overestimate,
underestimate
the
amount
of
just
stuff
about
stuff.
That
I
know,
though,
like
what
would
take
me
and
probably
what
would
take
you
guys,
probably
two
weeks
and
probably
taken
in
turn,
maybe
three
or
four
weeks,
but
that's
just
that's
my
guess.
A
Iii
just
think
my
take
would
be
I
agree
with
Noah
that
just
building
those
tools
and
because
we're
talking
about
a
timeline
of
June,
which
is
probably
six
months
from
now,
and
we
want
to
also
ensure
that
we
don't
wait
for
that
person
to
come
and
solve
all
our
problems.
And
we
want
to
keep
making
progress
in
the
meantime.
So
I'm
not
sure
what
exactly
is
going
to
be
left
then
and
based
on
what
we
prioritize
right
now.
A
But
I
can
see
that
as
a
combination
of
the
both
the
activities
you
just
mentioned,
having
to
do
some
of
the
tooling
and
improving
the
infrastructure
behind
documentation
and
also
you
know,
talking
to
different,
leads
and
identifying
their.
You
know,
pain
points
with
respect
to
documentation,
and
you
know
that
gives
the
person
opportunity
to
learn
about
surface
well
instead
of
just
tooling,
and
you
know,
UI
and
all
that
kind
of
stuff.
Yeah.
D
You
know
I
initially
thought
of
a
project
like
this
for
the
previous
round.
I'm.
Sorry,
you
guys
don't
have
an
intern
right
now,
because
I
guess
this
stand-up
didn't
exist
in
August
right
or
did
it
I,
don't
think
so?
Yes,
but
some
of
the
actual,
like
Doc's
writers
at
Red,
Hat,
were
interested
in
mentoring
too,
and
I
said.
All
you
have
to
do
is
like
it
evaluate
the
PRS,
like
the
actual,
you
know
documentation
that
the
interns
write
for
I,
don't
know
like.
D
If
is
it
understandable,
to
read
it's
the
content
structured
right
stuff
like
that
up,
and
that
was
where
I
initially
thought
about
that?
That's
how
I
initially
thought
of
an
intern
just
like
moving
around
different
teams
and
like
helping
them
document
stuff-
that's
not
already
been
documented,
but
I
also
realize
that
there's
like
just
as
much
complexity
in
the
documentation
generation
of
process
and
that's
probably
actually
better,
because
it
sounds
like
outreach.
The
intern.
Just
writing.
Docs,
for
everyone
is
a
bit
of
a
band-aid
solution.
I
am.
A
E
B
D
B
D
Yes,
or
is
there
anything
that
you
really
wish
could
get
done
or
like
wish?
The
docs
could
look
like
but
know
that
it
would
just
take
like
way
more
effort
amongst
the
three
of
you
or
whoever
else
comes
in
the
meeting
to
like
do
and
that
way
like
it's
something
that
somebody
could
do
full-time,
but
if
it
doesn't
pan
out,
then
we're
not
like
back
a
step
or
anything
like
that.
A
E
E
E
A
A
E
A
What
I
also
want
I
think
you
can
help
us
with
this?
Is
you
interact
with
a
lot
of
students
and
people
who
want
to
contribute
to
self,
but
don't
know
where
to
start?
We
have
these.
You
know
this.
If
you
look
at
the
chat,
we
have
this
PR
that
got
merged
and
that
lets
you
just
fix
documentation
using
the
github
UI
or
even
like
report
bugs
and
there's
description
on
how
a
person
can
make
use
of
that.
So
maybe
then
some
asks
you.