August 20, 201213 yr All, Do I have something wrong? I am only getting 13 to 20MB/s when transferring files from my workstation to the server. A lot of times it will be between 8 and 10 mb. I have 512MB on the server. I am using a HP XW4300 and the unregistered community edition. I am connected to a Gig switch, I have verified that all systems are connected and have a Gig link. Thanks for your help. Sideband Samurai
August 20, 201213 yr All, Do I have something wrong? I am only getting 13 to 20MB/s when transferring files from my workstation to the server. A lot of times it will be between 8 and 10 mb. I have 512MB on the server. I am using a HP XW4300 and the unregistered community edition. I am connected to a Gig switch, I have verified that all systems are connected and have a Gig link. Thanks for your help. Sideband Samurai Those speeds are exactly what some others get when transferring lots of small files. If you suspect something is wrong you'll need to read this post: http://lime-technology.com/forum/index.php?topic=9880.0
August 20, 201213 yr Check cables. Always start at layer one and work your way up. What's you MTU setting on your unraid server? Switch? Host your transferring from/to? It should be the same on all three. Check speed and duplex settings of unraid server, switch port and host. Each connection should be the same. Between unraid<>switch and switch<>host. Try using direct connection to unraid to host and see if you notice a difference.
August 20, 201213 yr Author all; from the workstation to the switch Windows 7 shows a 1GB connection. unraid shows the following: root@DavyJones:~# dmesg | grep -i eth0 eth0: Tigon3 [partno(BCM95752) rev 6001] (PCI Express) MAC address 00:17:a4:17:c9:4f eth0: attached PHY is 5752 (10/100/1000Base-T Ethernet) (WireSpeed[1]) eth0: RXcsums[1] LinkChgREG[0] MIirq[0] ASF[0] TSOcap[1] eth0: dma_rwctrl[76180000] dma_mask[64-bit] tg3: eth0: Link is up at 1000 Mbps, full duplex. tg3: eth0: Flow control is on for TX and on for RX. root@DavyJones:~# @Joe L. No I am not transferring small files, I am transferring LARGE files 4 to 8 GB in size. @ Seanant: All cables are brand new. What MTU should it be set at. Everything is a default install. With both the Windows 7 machine and the Unraid server. The version of the server is 4.7 right now, and its running unmenu. I do see that flow control is NOT enabled on the Windows 7 machine. I will enable flow control on windows 7 and report back. Thanks for the help. Sideband Samurai
August 20, 201213 yr The cables being brand new does not eliminate that as the problem. There could be a loose connection or a cable could be bad even though it is new.
August 20, 201213 yr Author @Lainie I will check the connections but they are the locking type. Even if the cables are bad, I am unable to replace them at this time. I changed the flow control and saw an improvement. I am now copying files between 15MB a sec and 19MB a sec. It is no longer dropping to the single digits when copying now. What about the MTA. What should that be set to? On the Windows machine, I also saw a setting for Jumbo frames. Can that improve anything? Thanks for the suggestions they are working. Sincerley, Sideband Samurai
August 20, 201213 yr All three devices need to be the same MTU. If the host, unraid server and switch support jumbo then check out the junbo thread in this forum to read about that. I wouldn't spend too much time on it, however it will improve performance if set up correctly. The flow and duplex settings are a big problem if not set correctly but it sounds like you got that under control.
August 20, 201213 yr Author All; if I enter the command ifconfig eth0 mtu 9014 i receive an Invalid Option error if I enter the command ifconfig eth0 mtu 1412 I receive no error and the MTU is set to 1412 if I enter the command ifconfig eth0 mtu 1500 I receive no error and the MTU is set to 1500. The built in Ethernet card I have is a Broadcom NetXtreme Gigabit Ethernet. it seems to me that there is not an option on these cards to set the MTU. As I look on my windows 7 machine (I have two HP xw4300) I don't see an option to set the MTU here either. Hmmm looks like I will have to get an Intel Gigabit Ethernet Card unless I find out how to set the MTU on these cards. --Sideband Samurai
August 20, 201213 yr Author ... and i was able to confirm from the Broadcom site that my Ethernet adapter does not support jumbo frames. hmmm. Any other suggestions would be welcome at this point. Sincerely, Sideband Samurai
August 21, 201213 yr Get a faster cpu... an old P4 is just not up to the task. Sadly unRAID is being advertises as a something that will run on almost any hardware you have lying around, which it basically does, but there is a difference between just running and actually working properly... i guess also today's expectations are higher. If you want to get the most out of unRAID and a GB network, get an Intel i3 minimal.
August 21, 201213 yr Those speeds are exactly what some others get when transferring lots of small files. If you suspect something is wrong you'll need to read this post: http://lime-technology.com/forum/index.php?topic=9880.0 Post a syslog and system specs. Are the drives connected to the MB SATA ports? Also show the output of the following: ifconfig ethtool eth0 hdparm -tT /dev/[hs]d?
August 21, 201213 yr Author @dgaschk thank you. Here is what you asked for: 1. ifconfig: root@DavyJones:~# ifconfig eth0 Link encap:Ethernet HWaddr 00:17:a4:17:c9:4f inet addr:10.1.1.10 Bcast:10.1.1.255 Mask:255.255.255.0 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:47268135 errors:0 dropped:0 overruns:0 frame:0 TX packets:56522497 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:329428784 (314.1 MiB) TX bytes:4172448545 (3.8 GiB) Interrupt:17 lo Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 UP LOOPBACK RUNNING MTU:16436 Metric:1 RX packets:8 errors:0 dropped:0 overruns:0 frame:0 TX packets:8 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:560 (560.0 B) TX bytes:560 (560.0 B) root@DavyJones:~# 2. ethtool eth0: root@DavyJones:~# ethtool eth0 Settings for eth0: Supported ports: [ TP ] Supported link modes: 10baseT/Half 10baseT/Full 100baseT/Half 100baseT/Full 1000baseT/Half 1000baseT/Full Supports auto-negotiation: Yes Advertised link modes: 10baseT/Half 10baseT/Full 100baseT/Half 100baseT/Full 1000baseT/Half 1000baseT/Full Advertised auto-negotiation: Yes Speed: 1000Mb/s Duplex: Full Port: Twisted Pair PHYAD: 1 Transceiver: internal Auto-negotiation: on Supports Wake-on: g Wake-on: g Current message level: 0x000000ff (255) Link detected: yes root@DavyJones:~# 3. hdparm: root@DavyJones:~# hdparm -tT /dev/[hs]d? /dev/sda: Timing cached reads: 2212 MB in 2.00 seconds = 1106.20 MB/sec Timing buffered disk reads: 350 MB in 3.02 seconds = 116.07 MB/sec /dev/sdb: Timing cached reads: 2740 MB in 2.00 seconds = 1370.86 MB/sec Timing buffered disk reads: 326 MB in 3.01 seconds = 108.39 MB/sec /dev/sdc: Timing cached reads: 2766 MB in 2.00 seconds = 1383.45 MB/sec Timing buffered disk reads: 336 MB in 3.02 seconds = 111.36 MB/sec /dev/sdd: Timing cached reads: 2784 MB in 2.00 seconds = 1392.44 MB/sec Timing buffered disk reads: 76 MB in 3.06 seconds = 24.80 MB/sec root@DavyJones:~# Attached is the syslog. This system is a HP xw4300 with 512MB ram and 3 Western digital Elements 2 TB drives. The Model number is:WDBAAU0020HBK-NESN I got these drives at the height of the Hard Drive shortage and only paid 120 a piece for them, when the going price was over $280 each. I removed the hard drives from their case and used these as the beginnings of my media server. This system comes with 6 SATA ports on the mother board. You can see the actual model at this link here: http://h18000.www1.hp.com/products/quickspecs/12260_na/12260_na.HTML The system contains a Intel Pentium 4 processor 630 supporting Hyper-Threading Technology (3.00 GHz/2 MB, 800 FSB, Single Core) This system is only temporary, as the wife has ordered me to install additional drives so that we can put our whole library of DVDS on the system, so that will also mean new system entirely. Meanwhile, as I have started moving all her horror films over (in preparation for Halloween), this is when i noticed the system was not transferring files as fast as I thought it should. With the change in flow control, I was able to achieve an improvement of 50%. That is, normally tera copy would hover between 8mb/s to 15 mb/s. Now I am seeing about 15mb/s to 25mb/s with it usually sitting between 18mb/s and 22mb/s IMHO, this is why I am thinking that my cable connections are not at fault. If they had been, I might not have seen such an improvement. Of course now thinking about it, it is possible I have slow hard drives that are causing this problem. After all the Elements drives are not known to be speed daemons. A little bit about me: I have been in computers since 1980 (were there computers way back then? Oh yes, TRS-80 Model I) and while in the NAVY I was operating a Keypro 10 running cp/m with DBASE II installed to maintain maintenance records of our ship. These records were routinely reviewed by the captain to help him determine the battle readiness of his ship. It was the first PC system aboard a Navy ship to accomplish this. Today, I just completed a project with the Orange County Airport (California) upgrading their computers and servers to bring them into the 21st century. I am open to any suggestions and will try anything if it seems reasonable enough. If I think something does not seem right, or I don't understand the request, then i will question it. Thanks again for all your help. This Forum has been excellent. --Sideband Samurai syslog-2012-08-21.txt
August 22, 201213 yr Test the network performance using iperf. It's available via unmenu. You can comment out the added lines in the go file to disable unmenu and its packages after testing, if desired. A little more RAM might help. Add another 512M or 2G.
August 23, 201213 yr Author I am actually thinking about upgrading to the version 5 RC. but I will give iperf a try.
August 23, 201213 yr Might be a dumb question but how is your source drive connected to your workstation? Last time I looked at a speed issue thread, the person was copying stuff off a USB-connected hard drive.
September 1, 201213 yr Author Hi, Not a dumb question at all. The source drive is connected directly to the system board via SATA cable. Its an internally mounted 160GB drive I use to temporarily store my DVD and Blueray rips. Once the drive gets full, then I do a bulk copy using teracopy. This was when I noticed that it appeared that I was a little speed challenged. Each file (mostly DVD's at this point) are around 4GB-8GB in size. You bring up a good point though, and I have already addressed this. This temp drive was reporting a SMART problem. Examine the smart table it showed repeated sector errors. The drive failed HP's DPS test. I have since replaced it with another 160GB drive that has no problems. I will see how this is working and post back -- also I am in the process of upgrading to 5 RC2 (or 3 what ever is the latest). so I have disabled all plug-ins and went with the stock go script in preparation for the upgrade. I will be purchasing a license and a couple of 3TB drives soon which is why I need to upgrade. I might just re-enable unmenu to perform the IPperf test. Thanks for your help. Sideband Samurai
Archived
This topic is now archived and is closed to further replies.