in an actual project we are trying to set up a copy of a database on the same host using rman with duplicate from active database. Everything works fine up to the forth datafile to be copied: suddenly the copy hangs and there is absolutely no chance to kill it. The database has a size of approx. 400 GB and the problem occured after 100 GB (each datafile is 32 GB) being copied successfully.
More than that: I tried do use an ordinary copy into the same filesystem and it hangs as well after some few Gigabyte.
It looks as if one Oracle process is blocking the entire filesystem and as the process is in "uninterruptible sleep" I don't have a chance to kill it other than to reboot the system.
Server is Oracle Enterprise Linux 6.3,
Filesystem is ext4 with filesystemio_options=DIRECTIO
Both databases do have their own filesystems.
Interesting though is that the FRA filesytem is blocked as well (ther's only the controlfile currently as I'm in the middle of the copy process).
We already raised an SR but the problem is that we always Need to reboot the Server to get rid of the processes (cumbersome and critical as there are other databases running on the same machine).
Any help will be appreciated