March 16, 201115 yr I'm having a bit of trouble trying to determine why my read speeds are so slow compared to others here on the board, and I was hoping someone might know how to proceed in diagnosing the problem. For lack of a better method, I'm using UnMenu's network monitor to determine my read throughput from the array. It looks like I average around 20MB/s (note MegaByte) read speeds. Strangely, my write speeds are around 18MB/s. Some other posts on the board have mentioned that your write speed is typically a fraction of your read, so that seems like an indicator something is wrong. I'm using a trio of 2TB WDEARS drives, and during preclear on all three drives saw Read/write speeds between 60 and 100MB/s, depending upon where in the platter the access was occuring. This seems to be right around the average from what I've seen, so I think the SATA controller and drives are probably OK. I'm thinking my read speed should be at least double what I'm achieving, and maybe there's a little headroom for increased write speed but probably not much. My syslog, and outputs from ethtool and ifconfig are attached. It appears I'm connected at Gigabit speeds, but just barely. I tried to use "dd" from my Mac, but for some reason the transfers slowed to 2MB/s, when I'd get the above 20MB/s if I transfer the same file using the Finder. Strange. I've gone through the FAQs and Troubleshooting sections, but the parts that apply don't seem to help. I'm on UnRaid 4.7. Any ideas where to go from here, or how to more accurately measure throughput? My system: Foxconn AMD NF4UK8AA (Yes I know this isn't a recommended board due to disk issues, but after thorough testing the issues that plague many of the NF4 boards are not present on this example) It looks like the chipset of its Gigabit NIC is a Cicada 8201 2GB DDR RAM AMD Athlon X2 4200+ NIC info (from ethtool) Settings for eth0: Supported ports: [ MII ] Supported link modes: 10baseT/Half 10baseT/Full 100baseT/Half 100baseT/Full 1000baseT/Full Supports auto-negotiation: Yes Advertised link modes: 10baseT/Half 10baseT/Full 100baseT/Half 100baseT/Full 1000baseT/Full Advertised auto-negotiation: Yes Speed: 1000Mb/s Duplex: Full Port: MII PHYAD: 8 Transceiver: external Auto-negotiation: on Supports Wake-on: g Wake-on: d Link detected: yes NIC driver info (from ethtool -i) driver: forcedeth version: 0.64 firmware-version: bus-info: 0000:00:0a.0 Ethernet config info (from ifconfig) eth0 Link encap:Ethernet HWaddr 00:15:58:49:a2:9c inet addr:10.0.1.100 Bcast:10.0.1.255 Mask:255.255.255.0 UP BROADCAST NOTRAILERS RUNNING MULTICAST MTU:1500 Metric:1 RX packets:369641 errors:0 dropped:0 overruns:0 frame:0 TX packets:280013 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:541652477 (516.5 MiB) TX bytes:23509453 (22.4 MiB) Interrupt:23 Base address:0xa000 syslog-2011-03-15.txt
March 16, 201115 yr Author This kind of dropped off the page, and I was really hoping someone might have an idea what's going on. Any help at all would be appreciated. Doing some further reading for potential problems, I verified that I have pre-assembled Cat-5e between my computer and the router, and the router and the UnRaid server. My link speed is recognized as 1000Mb. I'm not sure what's going on.
March 16, 201115 yr Author Gigabit: ifconfig en0 en0: flags=8963<UP,BROADCAST,SMART,RUNNING,PROMISC,SIMPLEX,MULTICAST> mtu 1500 ether d4:9a:20:d0:48:bf inet6 fe80::d69a:20ff:fed0:48bf%en0 prefixlen 64 scopeid 0x4 inet 10.0.1.2 netmask 0xffffff00 broadcast 10.0.1.255 media: autoselect (1000baseT <full-duplex,flow-control>) status: active Some testing I just did shows that writing directly to the disk share instead of using a user share increases my throughput to about 41MB/s. I guess user shares are really slow for some reason?
March 17, 201115 yr User shares are only slightly slower then disk shares. Are you writing to the same disk that the user share was writing?
March 17, 201115 yr Author Yes, it was the same disk. Although, even if it wasn't, all three drives are the same model on the same controller. I did manage to find a nice tool for measuring throughput: the Intel NAS Performance Toolkit. It looks like it's giving me much more reliable numbers than just watching the NIC throughput. So, it's showing 33MB/s from user shares, and 41MB/s from disk shares. That's a little bit closer to what I was expecting, and much better than 20MB/s. Write speeds in both cases is 20MB/s. That seems to be a little closer to what others are getting. I do wonder if the remaining performance gap is my NIC chipset. I'd probably upgrade the motherboard/processor before replacing the NIC, so if it is that I'm stuck with it.
Archived
This topic is now archived and is closed to further replies.