Description of problem: While running Anaconda under Kadischi, RPM transactions aren't possible on the host. lsof shows that Anaconda grabs all required locks in the target system __db.00{1,2,3} but __db.000 is locked on the host, rather than the target. Not exactly sure this is supposed to be this way, it could be perhaps. It could also be a falling out between Anaconda and any changes between it and the RPM Python modules. I didn't look into it too deep. It is my understanding Anaconda should only grab those locks in which are intended under the --rootpath specified. This is how Kadischi invokes Anaconda also. Thanks. Version-Release number of selected component (if applicable): anaconda-11.0.5-1 How reproducible: Run Anaconda under Kadischi, try an RPM transaction on the host. Steps to Reproduce: 1. RPM transactions 2. 3. Actual results: RPM transaction on the host were not possible during running of Anaconda and Kadischi. Expected results: Should perhaps be able to install something, anything say with rpm or yum while using Kadischi with Anaconda. Additional info: I can provide a brief lsof text if needed, Anaconda grabs the locks mentioned above in the target, but grabs the __db.000 lock on the host.
I have a separate issue with anaconda holding locks on /var/lib/rpm/__db.00{1,2,3}. In %post I would like to briefly unmount /var but cannot due to these locks. I think this started around core 4 or core 5.
This issue seems to have been fixed. While using the latest RAWHIDE as of today September 6th 2006, I can successfully run Kadischi and install some package with Yum at the same time. Thanks.