Hi,
I'm debugging something, and I'd like some help to understand what our
CREAMCE server is doing. If I do voms-proxy-destroy, then
voms-proxy-init, and then I submit a job, then that job goes off to my
CREAMCE server as it should. If, soon after, I again do
voms-proxy-destroy, then voms-proxy-init, and then I submit another job,
then new new job should go off to my CREAMCE server with a NEW proxy.
However, I've written jobs that query their own proxy, and both get the
same proxy. I.e. although I'm destroying the proxy from my submission
host, by the time the second (or third...) job gets to the worker node,
it has somehow acquired the old proxy, that I got rid of with
voms-proxy-destroy. What can explain this? Here's the proof:
:--- SOME PROOF ---
[sjones@hepgrid1 creamtests]$ voms-proxy-destroy ; voms-proxy-init
--voms dteam; voms-proxy-info -all ; ./doit.sh
.... Your proxy is valid until Fri Feb 11 06:46:32 2011
But in the job output, the previous (destroyed!) proxy got used:
[sjones@hepgrid1 creamtests]$ cat kHUHEJ0rCGEl7iz7VYoz3w/* | grep -i after
Time after the gzip operation
Not After : Feb 11 06:42:22 2011 GMT
:---
It's all working OK, but I'm curious - anybody know about this? Now and
again, the new proxy does get used, but not for a while.
Steve
|