directory-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Emmanuel Lecharny <>
Subject Re: VMs @ oxylos down
Date Wed, 17 Dec 2008 09:38:21 GMT
Felix Knecht wrote:
> Emmanuel Lecharny schrieb:
>> Felix Knecht wrote:
>>> Does anybody has an idea why all VMs on have gone
>>> down?
>>> Felix
>> Still down this morning. I don't have access to those machine. I will
>> ask Alex what's going on.
> I could startup the VMs again, but this wouldn't explain why they are
> down/if there's a special reason for them to be down.
Please, do so if you can start the machine !
>> Otherwise, I think we should rely on ASF machines for things like
>> continuus build. Infra has set up a continuum instance on their
>> managed servers, maybe it's time to move our builds there ?
> I thought that at that time the reason to do it on oxylos was to take
> some load of the ASF machines (at least I thought this was what Alex
> mentioned).
> Do you know if the have a sandbox to test the configurations and build
> commands?
Infra has set up some machines with Continuum, as some sponsor 
(specifically, Yahoo) gave some servers to the ASF.

I will check with infra and come back to you.
>> We also have a link on Oxylos for our source xref and reports, we may
>> ask infra about a place to store those infos on some infra servers too.
> This would also make it easier/possible to have the snapshots deploy on
> the ASF snapshot repository.

There is something else I think we must discuss, it's the release 
process. ATM, Alex only is capable of release ADS. I know that it's 
mainly because I'm lazzy and don't want to jump into the process, as it 
requires a deep knowledge in maven, but I also think this is a good idea 
that any committer can do the release.

There is some page on the wiki describing the release process, I just 
don't know how accurate they are.

Last, not least, we have some modules we don't compile in apacheds, may 
be it's time to move them to he sandbox or the dormant projects if we 
don't want to get them built, and we also have some clients (kerberos 
and ldap) which are not part of the global build, making them lagging, 
as they depend on trunk. I spent a few hours this week-end trying to get 
the kerberos client compiling, but it's a bit more complicated than I 
thought, due to the JNDI removal we did.
I would rather get those guys into the main build.

Maybe we need a good discussion about all those aspects :
- CI on The ASF machines
- XRef and metrics on The ASF machines
- Build and dormant modules
- Releases and packaging (javadoc, sources...)

wdyt ?

cordialement, regards,
Emmanuel L├ęcharny

View raw message