Messages in this thread | | | Date | Wed, 16 Feb 2000 20:12:44 +0100 | From | François Désarménien <> | Subject | Oracle data corruption: feed back |
| |
Hello,
A few months ago, I post the list about heavy Oracle databases corruption and loss of data. Today, I just want to give you some feedback.
I specially want to thanks Allan Coax for his help and suggestions, as I don't have anymore trouble, while running about 20 Linux/Oracle servers at our customers.
I was running Oracle 8.0.5.0 on a 2.2.9 patched kernel (by the distro) with a bug in rc.sysinit where '/sbin/update' was commented out :-(.
So, here is what I've done:
- I promptly uncommented '/sbin/update', which corrected the data losses when servers were shutdown, but '/' couldn't be unmounted.
- As suggested Alan, I downloaded a *fresh* (unpatched) 2.2.13 kernel and compiled it with plain gcc2.7.2, which became my standard kernel on all servers. It solved various problems, such as unkillable wait I/O processes which prevented to cleanly unmount filesystems.
- As suggested people from both the Oracle-Linux mailing list and Oracle french support, I disabled the Oracle server asynchronous I/O, which seems quite buggy (they told me it's not Linux specific, they had troubles with other Unices too) and upgraded to 8.0.5.1.
Since then, all thoses servers (applicative, not web) run really smoothly.
Again, thank you to all of you, folks, I really appreciate your work and your help
François
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |