2.1.111 VM and low memory machines

Alan Cox (alan@lxorguk.ukuu.org.uk)
Sun, 26 Jul 98 20:56 BST


I've been NFS destruction testing 2.1.111 and compared to 2.1.109 its much
better. The environment is entirely local except the kernel build tree
NFS mounted on top via NFS

16Mb 8KNFS kernel builds work reliably. If I really attack
the machine I can get it retrying NFS due to low
memory but no more
4K/1KNFS Attacking the machine makes no odds. Reliably builds

8Mb 8KNFS kernel builds work. A deliberate hard attack can
make the machine fail. With retry counts higher
it always succeeded in the end
4KNFS Attacking the machine makes no odds
1KNFS Attacking the mahine makes no odds

4Mb 8KNFS kernel builds generally fail
4KNFS Except when being attacked or very heavily loaded
kernel builds work fine. Under extreme attack
I can about force I/O errors and NFS deadlocks
1K NFS Builds fine

In comparison I could reliably kill 16Mb machines under attack with 2.1.109
and 4K NFS even, with 8Mb boxes they tended to die over time during the build
and 4Mb hardly got into make dep.

On a 8 and 16Mb machine its now pretty much the same reliability as 2.0 on
4Mb it feels subjectively better but I've not had time to qualify the
measurements (it takes SOOO long to build a kernel).

Alan

-
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.altern.org/andrebalsa/doc/lkml-faq.html