Re: Very high system load writing data to SCSI DVD-RAM

2001-02-05 Thread Oliver Feiler
Jens Axboe wrote: > > This is an old problem, and not related to the dvd-ram itself. If you > dirty lots of data and the target device is slow, kswapd/bdflush > will go crazy trying to free up memory. It should behave better on > 2.4.1, where we impose a global limit on locked buffers. Try and

Re: Very high system load writing data to SCSI DVD-RAM

2001-02-05 Thread Jens Axboe
On Mon, Feb 05 2001, Oliver Feiler wrote: > Hello, > > I have the following problem with a DVD-RAM drive. The drive is a > Panasonic LF-D101 connected to a Tekram DC395U SCSI controller. Kernel is > 2.2.18 with the patch for the Tekram controller >

Very high system load writing data to SCSI DVD-RAM

2001-02-05 Thread Oliver Feiler
Hello, I have the following problem with a DVD-RAM drive. The drive is a Panasonic LF-D101 connected to a Tekram DC395U SCSI controller. Kernel is 2.2.18 with the patch for the Tekram controller (http://www.garloff.de/kurt/linux/dc395/). When I write huge amounts of data to

Very high system load writing data to SCSI DVD-RAM

2001-02-05 Thread Oliver Feiler
Hello, I have the following problem with a DVD-RAM drive. The drive is a Panasonic LF-D101 connected to a Tekram DC395U SCSI controller. Kernel is 2.2.18 with the patch for the Tekram controller (http://www.garloff.de/kurt/linux/dc395/). When I write huge amounts of data to

Re: Very high system load writing data to SCSI DVD-RAM

2001-02-05 Thread Jens Axboe
On Mon, Feb 05 2001, Oliver Feiler wrote: Hello, I have the following problem with a DVD-RAM drive. The drive is a Panasonic LF-D101 connected to a Tekram DC395U SCSI controller. Kernel is 2.2.18 with the patch for the Tekram controller (http://www.garloff.de/kurt/linux/dc395/).

Re: Very high system load writing data to SCSI DVD-RAM

2001-02-05 Thread Oliver Feiler
Jens Axboe wrote: This is an old problem, and not related to the dvd-ram itself. If you dirty lots of data and the target device is slow, kswapd/bdflush will go crazy trying to free up memory. It should behave better on 2.4.1, where we impose a global limit on locked buffers. Try and run a