Effects of failing Gump runs

[Stefan Bodewig <bo...@apache.org>]

Hi,

last week we had failing JDK 1.5, Kaffe and test builds for different
reasons, yesterday the non-official public build failed.  I haven't
investigated why they failed so far, since I've been fighting the
side-effects.

If Gump fails it will now

(1) Not remove the lock files in the .../workspace/cvs or .../staging
    directories if there are any.

(2) Keep some integrate.py process hanging around.

(3) May even leave .../gump/gump.lock in place.

(1) leads to new processes of new Gump runs hanging forever and
probably happens if either an update or sync process was active when
Gump failed.

It is possible that (2) doesn't happen all the time, but there've been
about twenty processes from yesterday when I logged in to Brutus to
see why the official run didn't seem to be active.

(3) happenend last week, it did not happen yesterday for the public
run.  I manually removed the locks yesterday.

Stefan

---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org

Re: Effects of failing Gump runs

[Leo Simons <ma...@leosimons.com>]

On 10-05-2005 10:32, "Stefan Bodewig" <bo...@apache.org> wrote:
> last week we had failing JDK 1.5, Kaffe and test builds for different
> reasons, yesterday the non-official public build failed.  I haven't
> investigated why they failed so far, since I've been fighting the
> side-effects.

I have tried to figure it out in several cases. Failure so far. A lot of our
errors are "transient". Perhaps we could have a try/finally in the right
location to remove lock files. Adam, any pointers?

Cheers,

Leo



---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@gump.apache.org
For additional commands, e-mail: general-help@gump.apache.org