475461 - LDAP account request for try-server access for [e-mail]

Status: RESOLVED FIXED

(Infrastructure & Operations Graveyard :: Account Requests, task)

Description

[Andreas Gal :gal]

We need access to our try server for [e-mail]. I am not sure what exactly is needed for this to happen. Intel is working on a parallel compiler thread, which is a Q1 goal and this would help with that effort. Currently intel can only test with trace-tests and sunspider. Running SS on our tinderboxes would give us a direct performance comparison and help getting the patch ready to go into the TM tree. If it is absolutely impossible to grant separate try-server access, then maybe I can get try server access and then push for intel to the server (currently I can't use the try server myself either, mrbkap pushes for me).

Comment 1

[Mike Shaver (:shaver -- probably not reading bugmail closely)]

You should have access to the try server, I think?  File another ticket for that, we'll get to the bottom of it.

(Also, seconding the request for Shengnan -- it'll help us a bunch with getting better perf data in a bunch of ways over time.)

Comment 2

[Andreas Gal :gal]

I don't have access to mozilla-central, only TM. Try-server seems to use the same access control bits and pieces.

Comment 3

[Mike Shaver (:shaver -- probably not reading bugmail closely)]

File
another
ticket
for
that,
we'll
get
to
the
bottom
of
it.

Comment 4

[Andreas Gal :gal]

D
o
n
e

https://bugzilla.mozilla.org/show_bug.cgi?id=475462

Comment 5

[Reed Loden [:reed]]

So, we haven't accepted account requests just for try server access in the past (in fact, I can think of multiple bugs for this same type of request that have been WONTFIX'd instead). However, if Shengnan could find some reason to get an LDAP account, he/she could then use the try server web interface to test his/her patches. Access to push to the try server via Hg is actually tied directly to the default Hg permission bit, so have that bit would allow Shengnan to push changes to most of the repos in Hg (save mozilla-central, comm-central, and maybe a few more), which probably isn't what's wanted at this time.

Comment 6

[Mike Shaver (:shaver -- probably not reading bugmail closely)]

Can someone point me to the policy rationale that exceeds the leverage we get from adding specific people to the try server?  I'd like to understand what we're gaining from not letting contributors do their work better, and help us improve our software with less load on our existing contributors.

(I guess I could reuse the one-off account I'm getting for shark-via-try-server automation and set up a web app that lets select people into it, but that seems like a lot of work compared to us just giving try server access to the people that we want to have this access. :-( )

Comment 7

[Reed Loden [:reed]]

(In reply to comment #6)
> Can someone point me to the policy rationale that exceeds the leverage we get
> from adding specific people to the try server?

What policy? That's the problem. We have no such policy for something like this that isn't for commit access, and granting such access (to push via Hg) would in effect grant commit access at the same time. If web interface access is acceptable, we probably could just create an LDAP account for Shengnan.

Comment 8

[Mike Shaver (:shaver -- probably not reading bugmail closely)]

The policy under which we've been WONTFIXing requests to add accounts just for try server access, I mean.  I suspect that the LDAP route is fine for Shengnan, since AIUI the web interface will suffice.

Is there a bug for making push-to-try use the same credential requirement as web-to-try?  Seems pretty odd that they're different, and with bugs in the web interface WRT applying patches to non-default repos, the push interface can be pretty handy.

Comment 9

[Reed Loden [:reed]]

(In reply to comment #8)
> The policy under which we've been WONTFIXing requests to add accounts just for
> try server access, I mean.

Bug 393298, comment #5 is the first thing I find.

> Is there a bug for making push-to-try use the same credential requirement as
> web-to-try?

No.

Comment 10

[Andreas Gal :gal]

Whats the next step here to get Shengnan access?

Comment 11

[Mike Shaver (:shaver -- probably not reading bugmail closely)]

(In reply to comment #7)
> If web interface access is
> acceptable, we probably could just create an LDAP account for Shengnan.

Acceptable!  Should it go over to IT for acc't creation?

Comment 12

[Reed Loden [:reed]]

IT, just need a LDAP account created with no permissions for anything. Please make sure to send the password to Shengnan.

Comment 13

[chizu]

Account created.

Comment 14

[Shengnan Cong]

Thanks for all the help! 

I am able to access  https://ldap.mozilla.org/passwordreset/ to change my passwd, but still unable to access https://ldap.mozilla.org. I am keeping asked for login/passwd. Seems I still don't have the permission to access it. Any suggestions?

Comment 15

[Reed Loden [:reed]]

(In reply to comment #14)
> I am able to access  https://ldap.mozilla.org/passwordreset/ to change my
> passwd, but still unable to access https://ldap.mozilla.org. I am keeping asked
> for login/passwd. Seems I still don't have the permission to access it. Any
> suggestions?

You shouldn't need to connect to https://ldap.mozilla.org at all. Is there a reason why you think you need to?

Comment 16

[Shengnan Cong]

Is there any pointer/document describing how to use the LDAP account for testing? I was just looking for such information. Thanks.

Comment 17

[Andreas Gal :gal]

I will send Shengnan some instructions via email.