June 1, 201115 yr My Setup Motherboard - Gigabyte X48-DQ6 CPU - Intel C2D E8400 RAM - 2GB DDR2-800 Parity Disk - Hitachi 7K3000 7200rpm 2TB (on Channel 0 of 1st Sil3132 PCI-e 1x card) Disk 1 to 8 - Hitachi 5K3000 5900rpm 2TB x 8 (on board ICH9 (AHCI) x 6 + on board JMicorn (AHCI) x 2) Cache Disk - WD 7200rpm 1TB (on Channel 0 of 2nd Sil3132 PCI-e 1x card) UnRaid 4.7 AIO For parity check, I just got 10 - 30 MB/s (Average about 20MB/s), it take me nearly 24 hours to check the 2TB parity disk. Which is much slower than those speed stated on User Benchmark Page (50-90MB/s for new system with PCIe like mine) http://lime-technology.com/wiki/index.php?title=User_Benchmarks Any hints for me to improve the speed?
June 1, 201115 yr Author I think your bottleneck will be the slower RPM drives. Really? if rpm is related to speed 5900/7200 = 82% According to the test here. http://www.digitalversus.com/hdd-reviews-hitachi-deskstar-5k3000-7k3000-news-17909.html The read/write test result of 5K3000 (5900rpm) is about 90% of 7K3000 (7200rpm) But my case is much less than that, (25MB/s vs 50-90MB/s)
June 1, 201115 yr Author Your problem is not drive speed. Post a syslog. Is AHCI set in BIOS? Yes! I set the AHCI mode in BIOS Attached please find the syslog at the booting stage syslog.txt
June 1, 201115 yr That log does not show enough. Collect one after a parity check or some other too slow operation.
June 1, 201115 yr Author That log does not show enough. Collect one after a parity check or some other too slow operation. I cannot found any line related to the parity check process in syslog. Do you know which line is related to parity check? (like the name of the command)
June 1, 201115 yr That log does not show enough. Collect one after a parity check or some other too slow operation. I cannot found any line related to the parity check process in syslog. Do you know which line is related to parity check? (like the name of the command) it would look something like this at the start. (It may, or may not have the "NOCORRECT" option): Jun 1 00:00:01 Tower2 kernel: mdcmd (49): check NOCORRECT Jun 1 00:00:01 Tower2 kernel: Jun 1 00:00:01 Tower2 kernel: md: recovery thread woken up ... Jun 1 00:00:01 Tower2 kernel: md: recovery thread checking parity... Jun 1 00:00:01 Tower2 kernel: md: using 1152k window, over a total of 1953514552 blocks. And it will look like this at the end of the parity check: Jun 1 09:52:16 Tower2 kernel: md: sync done. time=35534sec Jun 1 09:52:16 Tower2 kernel: md: recovery thread sync completion status: 0
June 1, 201115 yr Author Thanks a lot for your information. I will re-run the test and post the result here.
June 1, 201115 yr You can try to move the two HDs from the on board Jmicron to the other channel of the SIL3132 cards and then disable the Jmicron controller all together. (including the IDE)
June 2, 201115 yr Author You can try to move the two HDs from the on board Jmicron to the other channel of the SIL3132 cards and then disable the Jmicron controller all together. (including the IDE) Thanks a lot for your advice, I will try it.
June 2, 201115 yr Author Just finish another parity check. This time only need 15 hours instead of nearly 24 hours in last run. Syslog for Parity Check. Jun 2 07:28:33 UR kernel: mdcmd (50): check CORRECT (unRAID engine)Jun 2 07:28:33 UR kernel: md: recovery thread woken up ... (unRAID engine) Jun 2 07:28:33 UR kernel: md: recovery thread checking parity... (unRAID engine) Jun 2 07:28:33 UR kernel: md: using 1152k window, over a total of 1953514552 blocks. (unRAID engine) Jun 2 22:38:32 UR kernel: md: sync done. time=54594sec rate=35782K/sec (unRAID engine) Jun 2 22:38:33 UR kernel: md: recovery thread sync completion status: 0 (unRAID engine) Still just got 35MB/s, any idea?
June 2, 201115 yr Just finish another parity check. This time only need 15 hours instead of nearly 24 hours in last run. Syslog for Parity Check. Jun 2 07:28:33 UR kernel: mdcmd (50): check CORRECT (unRAID engine)Jun 2 07:28:33 UR kernel: md: recovery thread woken up ... (unRAID engine) Jun 2 07:28:33 UR kernel: md: recovery thread checking parity... (unRAID engine) Jun 2 07:28:33 UR kernel: md: using 1152k window, over a total of 1953514552 blocks. (unRAID engine) Jun 2 22:38:32 UR kernel: md: sync done. time=54594sec rate=35782K/sec (unRAID engine) Jun 2 22:38:33 UR kernel: md: recovery thread sync completion status: 0 (unRAID engine) Still just got 35MB/s, any idea? Were you doing anything else on the server last time? Other than that, and in the absence of a syslog with errors, can't offer any advice.
June 3, 201115 yr Author Were you doing anything else on the server last time? Other than that, and in the absence of a syslog with errors, can't offer any advice. Nothing else except parity check. My hardware configuration, syslog for bootup and parity check were posted in previous post. I cannot found any error in syslog. I just bootup the system and start the parity check. I just want to know, with my configuration, is 35MB/s parity check speed normal or not. Cause I found some much "slower" and "older" setup can have a speed of 50-90MB/s. Will I use a supermicro 8 ports SATA card to host all my 8 data disk will improve the problem?
June 3, 201115 yr I cannot found any error in syslog. I just bootup the system and start the parity check. I just want to know, with my configuration, is 35MB/s parity check speed normal or not. They are some quirks in your syslog: May 31 08:38:50 UR kernel: pci 0000:00:1f.0: quirk: region 0400-047f claimed by ICH6 ACPI/GPIO/TCO May 31 08:38:50 UR kernel: pci 0000:00:1f.0: quirk: region 0480-04bf claimed by ICH6 GPIO May 31 08:38:50 UR kernel: pci 0000:00:1f.0: ICH7 LPC Generic IO decode 1 PIO at 0800 (mask 000f) May 31 08:38:50 UR kernel: pci 0000:00:1f.0: ICH7 LPC Generic IO decode 2 PIO at 0290 (mask 000f) And then you have onboard SATA 6 ports controller, onboard JMicron SATA 2 ports & IDE controller and two additional SIL controllers and that combination together with the fact that the Hitachi HD are SATA3 may not like each other that much (and in addition they do share IRQs): May 31 08:38:50 UR kernel: sata_sil24 0000:02:00.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16 May 31 08:38:50 UR kernel: sata_sil24 0000:04:00.0: PCI INT A -> GSI 18 (level, low) -> IRQ 18 May 31 08:38:50 UR kernel: ahci 0000:00:1f.2: PCI INT B -> GSI 19 (level, low) -> IRQ 19 May 31 08:38:50 UR kernel: ahci 0000:05:00.0: PCI INT A -> GSI 19 (level, low) -> IRQ 19 May 31 08:38:50 UR kernel: JMicron IDE 0000:05:00.1: PCI INT B -> GSI 16 (level, low) -> IRQ 16 How to fix that: 1. Make sure you have the latest BIOS for the motherboard. Then load the default values, save and reboot - watch out here for Gigabyte famous HPA. Go again into the BIOS, make sure that the HDs are in AHCI mode and then disable any unused hardware features (serial and par. ports, audio, firewire, floppy, IDE controller, etc. and if you have choice you can disable a few of the USB ports too). Save this configuration. 2. Make sure you flash the two SIL cards with the latest nonRAID BIOS as they may come with an older one that does not like the SATA3. Then you can try again and do not forget to update us with the status.
June 3, 201115 yr Author Thanks a lot for all your advices. 1) My Gigabyte X48-DQ6 is using the latest BIOS (except there is a newer beta version) 2) All Sil3132 cards were flashed with latest non-RAID version BIOS. Just have time to move those 2 HD on JMicron to Sil3132 card. I started another round of parity check and found that the initial speed is nearly double. About 50-60MB/s now! Just order a supermicro 8 ports SATA card and I will move all my 8 data disks to this card. Hopefully it can improve the speed.
June 3, 201115 yr I have slow Parity as well. 6 drives on my ASUS motherboard and others on a supermicro card. I will check out the bios settings. If I can't fix it, I will just buy another Supermicro card and move the drives to that.
Archived
This topic is now archived and is closed to further replies.