Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
I was a little premature in sending this email, so let me conclude here... running the tests below, I seen a consistent degradation in the raw block device, this was when the drbd module was loaded and state disconnected. Once I removed the module by running /etc/init.d/drbd stop and re-ran the tests, the performance write was back up to speed. So there must be something in the module. Please follow below... > I have been running this command to benchmark WRITE performance to the SCSI > block device. > The funny thing is that each additional run, the WRITE performace > decreases... INTERESTING.... > > > ./dm -a 0 -b 1M -s 3G -y -m -p -o /dev/sdb > 71.87 MB/sec (3221225472 B / 00:42.745236) > > ./dm -a 0 -b 1M -s 3G -y -m -p -o /dev/sdb > 67.68 MB/sec (3221225472 B / 00:45.387656) > > ./dm -a 0 -b 1M -s 3G -y -m -p -o /dev/sdb > 60.62 MB/sec (3221225472 B / 00:50.679095) > > ./dm -a 0 -b 1M -s 3G -y -m -p -o /dev/sdb > 51.54 MB/sec (3221225472 B / 00:59.604501) > > ./dm -a 0 -b 1M -s 3G -y -m -p -o /dev/sdb > 46.71 MB/sec (3221225472 B / 01:05.771989) > > ./dm -a 0 -b 1M -s 3G -y -m -p -o /dev/sdb > 42.49 MB/sec (3221225472 B / 01:12.297245) > > ./dm -a 0 -b 1M -s 3G -y -m -p -o /dev/sdb > 39.35 MB/sec (3221225472 B / 01:18.063164) > > ./dm -a 0 -b 1M -s 3G -y -m -p -o /dev/sdb > 38.00 MB/sec (3221225472 B / 01:20.845423) > > ./dm -a 0 -b 1M -s 3G -y -m -p -o /dev/sdb > 35.44 MB/sec (3221225472 B / 01:26.670520) > > ./dm -a 0 -b 1M -s 3G -y -m -p -o /dev/sdb > 34.32 MB/sec (3221225472 B / 01:29.512314) > > ./dm -a 0 -b 1M -s 3G -y -m -p -o /dev/sdb > 33.47 MB/sec (3221225472 B / 01:31.781341) > > > (linux-kernel 2.6.20.3) > Adaptec aacraid driver (1.1-5[2423]-mh3) > ACPI: PCI Interrupt 0000:04:08.1[A] -> GSI 30 (level, low) -> IRQ 17 > AAC0: kernel 2.8-1[6098] > AAC0: monitor 2.8-1[6098] > AAC0: bios 2.8-1[6098] > AAC0: serial 80d081d3 > scsi0 : percraid > scsi 0:0:0:0: Direct-Access DELL PERC Mirror V1.0 PQ: 0 ANSI: 2 > scsi 0:0:1:0: Direct-Access DELL PERC Volume V1.0 PQ: 0 ANSI: 2 > megasas: 00.00.03.05 Mon Oct 02 11:21:32 PDT 2006 > SCSI device sda: 71091456 512-byte hdwr sectors (36399 MB) > sda: assuming Write Enabled > sda: assuming drive cache: write through > SCSI device sda: 71091456 512-byte hdwr sectors (36399 MB) > sda: assuming Write Enabled > sda: assuming drive cache: write through > sda: sda1 sda2 sda3 < sda5 > > sd 0:0:0:0: Attached scsi removable disk sda > SCSI device sdb: 71677440 512-byte hdwr sectors (36699 MB) > sdb: assuming Write Enabled > sdb: assuming drive cache: write through > SCSI device sdb: 71677440 512-byte hdwr sectors (36699 MB) > sdb: assuming Write Enabled > sdb: assuming drive cache: write through > sdb: unknown partition table > sd 0:0:1:0: Attached scsi removable disk sdb > sd 0:0:0:0: Attached scsi generic sg0 type 0 > sd 0:0:1:0: Attached scsi generic sg1 type 0 now I removed the module by running /etc/init.d/drbd stop lsmod shows drbd was indeed removed.... ./dm -a 0 -b 1M -s 3G -y -m -p -o /dev/sdb 71.99 MB/sec (3221225472 B / 00:42.672736) ./dm -a 0 -b 1M -s 3G -y -m -p -o /dev/sdb 71.39 MB/sec (3221225472 B / 00:43.029786) ./dm -a 0 -b 1M -s 3G -y -m -p -o /dev/sdb 72.55 MB/sec (3221225472 B / 00:42.341708) tried MULTIPLE times on two identical pieces of hardware, and get the same results: stable, consistent write throughput. so in summary, with the drbd module loaded (nd in disconnected state even), write performance to the raw block device degrades over time... Thanks Duane Cox