RADIatiON Posted December 21, 2006 Share Posted December 21, 2006 OK, So, I just assembled my server (ASUS P5PE-VM, D820, 1GB ram, 2 x 320G Seagate SATA2 on mb connectors, 4 x 250GB WD IDE hooked up to Promise TX2 100) and hooked up a new Netgear 5 port gig switch. I am getting pretty poor transfer speed right now, copying files from my old (real old... p233 old) OpenBSD server with a 3com gig card to my new server. I'm getting on average about 10GB an hour. Now I tried through the switch, and different cables (just put in new CAT5e cables) and even tried a direct connection between the machines... still about the same. I updated the firmware on the Promise cards just in case... new IDE ribbons... everything I can think of. Does it have anything to do with unRAID creating parity while it's copying? Is it because of the old p233 machine? I put a new DLink Gig card (not a fan of DLink but I couldn't get my hands on anything else) in my desktop (WnXP, AMD XP 2600, 1GB ram, SATA1 drives) and tried copying some files across to see what it was like and I was getting about 20% utilization (according to Task Manager) after I changed some registry entries to help out the Gig NIC. I also turned off the QOS. It still was taking over 2 mins to copy a 350MB file. BTW... watched a couple shows last night on my HTPC served by the new unRAID (no copying at the time, but parity was being rebuilt) and everything was great I just want more network copying speed RADIatiON Quote Link to comment
TheMaster Posted December 21, 2006 Share Posted December 21, 2006 I have a fairly high end workstation and the unRAID server whose benchmarks I just posted in the HARDWARE section. These are connected through a Netgear Gigabit switch. I just copied an 11.2 GB file to test the total time required to transfer the file and it looks like it is going to take about 22 minutes. That's 19.6 minutes for a 10GB file. Quote Link to comment
RADIatiON Posted December 21, 2006 Author Share Posted December 21, 2006 I just copied an 11.2 GB file to test the total time required to transfer the file and it looks like it is going to take about 22 minutes. That's 19.6 minutes for a 10GB file. Not nice to brag Just kidding.. thanks for the comparison numbers. This is really just an annoyance right now. I don't plan to be doing any more 1TB+ copies anytime soon. I'm assuming it's the old p233 that is probably causing most of the problem. I'll know better on the weekend. Quote Link to comment
TheMaster Posted December 22, 2006 Share Posted December 22, 2006 Hehe.. Yea, I was just trying to give you a "best case" scenario to judge by. Frankly, I was surprised it took over 20 minutes to copy my 11GB file. I would say 10GB an hour isn't too far out of line since there is an older workstation involved. Having said that, I would have almost expected the network to be the limiting factor rather than the PC, but perhaps that isn't the case. I also backup my workstation to the unRAID server and I just looked back at the first night it backed up. In this case it had to transfer 117.1GB to begin the backup set - I don't do any compression, so the speed is limited by the transfer rate of the systems in question. This transfer took 7 hours and 9 minutes. That's only 16.37 GB per hour. The backup keeps track of performance for each backup and the best transfer rate I have achieved is 567 MB/minute. When I did these backups to a local USB 2.0 portable hard drive they performed at rates of over 40GB/hr... this is why I thought the network would be the limiting factor above. The workstation I am using is an Athlon FX-60 with a pair of WD Raptor hard drives in a RAID stripe configuration (OK, now I must be bragging)... so, I wouldn't beat yourself up for having a 10GB/hr transfer rate with the P233. Quote Link to comment
limetech Posted December 22, 2006 Share Posted December 22, 2006 RADIatiON- Use telnet and post the output of these commands: ifconfig ethtool eth0 This will let us determine what speed the GigE port is actually running at. Quote Link to comment
RADIatiON Posted December 22, 2006 Author Share Posted December 22, 2006 Hey Tom, Looks like it's running full out. I'll report back once I get the old p233 out of the mix. root@Tower:~# ifconfig eth0 Link encap:Ethernet HWaddr 00:18:F3:19:D0:50 inet addr:192.168.2.100 Bcast:192.168.2.255 Mask:255.255.255.0 UP BROADCAST NOTRAILERS RUNNING MULTICAST MTU:1500 Metric:1 RX packets:119013529 errors:0 dropped:0 overruns:0 frame:0 TX packets:80858392 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:637452652 (607.9 Mb) TX bytes:2133412504 (2034.5 Mb) Interrupt:11 Memory:fbff8000-0 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:14 errors:0 dropped:0 overruns:0 frame:0 TX packets:14 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:2036 (1.9 Kb) TX bytes:2036 (1.9 Kb) root@Tower:~# 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: 0 Transceiver: internal Auto-negotiation: on Supports Wake-on: g Wake-on: g Link detected: yes Quote Link to comment
limetech Posted December 22, 2006 Share Posted December 22, 2006 Yep, and no errors either... Start a parity-check, let it run about 5 min, then click 'Refresh' and tell me what rate is being reported. Quote Link to comment
RADIatiON Posted December 23, 2006 Author Share Posted December 23, 2006 OK, Now that all the heavy lifting is done, I seem to be getting some "OK" speed from my main machine to the new server. I'm getting about 700MB a minute on average... which I think ain't bad. Start a parity-check, let it run about 5 min, then click 'Refresh' and tell me what rate is being reported. The system is up with valid Parity now... I think I remember the speed was around 29KB but don't quote me on that. I didn't want to chance it while copying files. If you think it's important I will rebuild the parity in a day or two... but I think things are ok now. Thanks for the help.. oh and TheMaster Quote Link to comment
foo_fighter Posted December 24, 2006 Share Posted December 24, 2006 That's still not very good in my opinion(barely enough to saturate a 100Mbs link). I'm having similar issues but I can't track it down. I'm only getting 4-4.5MB/s from an old computer with a 100Mb/s connection. But what's worse is I'm only getting 5MB/s from a USB 2.0 disk(NTFS) directly attached using local rsync. I'm pretty sure it's not CPU limited and probably not ram limited. I can parity sync at 50,000-70,000KB/s. My specs are very similar to TheMaster's(see his post) and I only have 2 sata II drives. I would have expected closer to 20MB/s locally. I'm not able to access the computers now, but when I get back I'll have to try it with the parity unassigned and see if it's better. OK, Now that all the heavy lifting is done, I seem to be getting some "OK" speed from my main machine to the new server. I'm getting about 700MB a minute on average... which I think ain't bad. Quote Link to comment
foo_fighter Posted December 28, 2006 Share Posted December 28, 2006 I tried the transfers with the parity turned off and it didn't help much. I think it is ram limited, or there is a severe memory runaway/leak issue. When it boots up, I have about 350 megs of ram free. As soon as I transfer anything it goes down to 4-5megs free. I think that's what's limiting the transfers but I can't track down what's hogging the memory. Top shows: top - 17:05:23 up 8 min, 2 users, load average: 2.11, 1.31, 0.56 Tasks: 40 total, 3 running, 36 sleeping, 0 stopped, 1 zombie Cpu(s): 0.3% user, 16.0% system, 0.0% nice, 83.7% idle Mem: 507460k total, 502776k used, 4684k free, 188392k buffers Swap: 0k total, 0k used, 0k free, 286012k cached PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 1042 root 14 0 476 476 392 D 15.3 0.1 0:26.78 cp 18 root 12 0 0 0 0 R 1.0 0.0 0:02.99 usb-storage-1 1 root 8 0 240 240 212 S 0.0 0.0 0:04.38 init 2 root 9 0 0 0 0 S 0.0 0.0 0:00.01 keventd 3 root 19 19 0 0 0 S 0.0 0.0 0:01.37 ksoftirqd_CPU0 4 root 9 0 0 0 0 S 0.0 0.0 0:00.58 kswapd 5 root 9 0 0 0 0 S 0.0 0.0 0:00.00 bdflush 6 root 9 0 0 0 0 S 0.0 0.0 0:00.02 kupdated 9 root 9 0 0 0 0 S 0.0 0.0 0:00.00 khubd 15 root 9 0 0 0 0 S 0.0 0.0 0:00.00 usb-storage-0 16 root 9 0 0 0 0 S 0.0 0.0 0:00.00 scsi_eh_0 19 root 9 0 0 0 0 S 0.0 0.0 0:00.00 scsi_eh_1 68 root 9 0 676 672 572 S 0.0 0.1 0:00.00 syslogd 71 root 9 0 456 456 400 S 0.0 0.1 0:00.00 klogd 233 root 9 0 0 0 0 S 0.0 0.0 0:00.00 scsi_eh_2 234 root 9 0 0 0 0 S 0.0 0.0 0:00.00 scsi_eh_3 921 root 9 0 516 516 456 S 0.0 0.1 0:00.00 inetd 927 root 9 0 592 592 508 S 0.0 0.1 0:00.00 crond 930 root 9 0 552 552 492 S 0.0 0.1 0:00.00 acpid 942 root 9 0 1136 1136 880 S 0.0 0.2 0:00.01 emhttp 950 root 8 0 520 520 464 S 0.0 0.1 0:00.00 ifplugd 960 root 9 0 0 0 0 Z 0.0 0.0 0:00.00 mdrecover <defunct> Quote Link to comment
limetech Posted December 28, 2006 Share Posted December 28, 2006 foo- There are no memory leaks - the memory reporting of 'top' is known to be next to useless because of the way linux manages memory. A couple things to try: 1. Try writing from unRAID server to your target machine & see if any difference. 2. Do you have a windows pc you can try? 3. You can try copying a file on the locally mounted NTFS usb disk to 'null', e.g., time cp /ntfs-disk/file /dev/null where 'ntfs-disk' is the mount point and 'file' is a large file. Quote Link to comment
Joe L. Posted December 28, 2006 Share Posted December 28, 2006 foo- There are no memory leaks - the memory reporting of 'top' is known to be next to useless because of the way linux manages memory. A couple things to try: 1. Try writing from unRAID server to your target machine & see if any difference. 2. Do you have a windows pc you can try? 3. You can try copying a file on the locally mounted NTFS usb disk to 'null', e.g., time cp /ntfs-disk/file /null where 'ntfs-disk' is the mount point and 'file' is a large file. Tom, Did you intend to type "/dev/null" ?? As in time cp /ntfs-disk/file /dev/null My unRaid server does not have any existing file, directory, or device at "/null" and doing as you suggest will create a copy of the ntfs file in memory at the root of the ramdisk mounted at "/" A big enough file, and you might just run out of RAM. Not sure, but that can't be good and just might crash the unRaid server. Quote Link to comment
limetech Posted December 28, 2006 Share Posted December 28, 2006 Tom, Did you intend to type "/dev/null" ?? As in time cp /ntfs-disk/file /dev/null My unRaid server does not have any existing file, directory, or device at "/null" and doing as you suggest will create a copy of the ntfs file in memory at the root of the ramdisk mounted at "/" A big enough file, and you might just run out of RAM. Not sure, but that can't be good and just might crash the unRaid server. Yes, of course you're right I'll modify the post to show the right command, thanks. Quote Link to comment
foo_fighter Posted December 30, 2006 Share Posted December 30, 2006 Yeah, I'm limited by the target PCs, I'm seeing the CPU bounce up against 100%. And even locally on the PC itself I can only get 9MB/s from an external USB disk to an internal IDE drive. I guess I was just expecting to see those IOZONE numbers. I think those are the max possible and not really real world achievable. For instance I was syncing a bunch of small files (mp3/flac) and the drives have to seek/spin up for each file so there's no way it could achieve the rate of a continuous transfer. I was surprised that the local mounted USB drive on unraid was not much faster. I haven't tried cp'ing to /dev/null yet. I have a laptop with GigE but I can't it can't see the unraid shares. Does everything have to be on the same domain/workgroup? This is a work laptop running windows 2000. I can never figure out how windows 2000 / xp shares work. foo- There are no memory leaks - the memory reporting of 'top' is known to be next to useless because of the way linux manages memory. A couple things to try: 1. Try writing from unRAID server to your target machine & see if any difference. 2. Do you have a windows pc you can try? 3. You can try copying a file on the locally mounted NTFS usb disk to 'null', e.g., time cp /ntfs-disk/file /dev/null where 'ntfs-disk' is the mount point and 'file' is a large file. Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.