coldflame

Hello all, I've had Unraid running for quite some time now and it seems to be working fine and performance is what it should be when copying large files (30MB/s writes, 100 MB/s reads). It's a Sempron 145 with 2 GB RAM and 4 SATA disks. What I've always seen however is that when I need to do a file operation, for example delete a file from the share, it takes much longer to do the same operation from a network share on my Windows PC or on my Synology Diskstation. This seems to happen for all file access operations, like even creating a directory is slow on my Unraid box. Is this level of performance expected or this is something with my hardware or Unraid settings? I'm only running Unmenu as the addon and nothing else. Thanks!

I have filled mine with a variety of drives and I'm having trouble keeping the drives cool during parity check - I think one of the drives hit 50 C. I've added an additional fan into the front (Antec 3Cool on lowest setting) and still had to remove 2 drives. Im currently running 2 x 1.5 TB Samsungs 5400 rpm and 2 x 1 TB WD Greens and HD temps stay in 35-40 range. Great case but I'm pretty disappointed by cooling performance.

I did some speed tests with my Atom tonight and I'm seeing 40ish MB/s performance without the parity drive. I expected to see close to 80-100 MB/s. This is what I see in 'top' - I think Atom is just not able to keep up with the load and is bottlenecking the system. Thoughts? PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 1600 root 20 0 47348 1768 700 S 54 0.4 2:27.46 shfs 1579 root 20 0 13960 4004 3092 S 43 0.8 1:49.58 smbd 1258 root 20 0 1664 236 156 S 22 0.0 1:09.75 dhcpcd 5680 root 20 0 0 0 0 S 17 0.0 0:15.28 flush-9:1 1489 root 20 0 0 0 0 S 8 0.0 0:13.58 unraidd 124 root 20 0 0 0 0 S 6 0.0 0:16.91 sync_supers 321 root 20 0 0 0 0 S 5 0.0 0:10.31 kswapd0 5681 root 20 0 2116 992 788 R 0 0.2 0:00.06 top 1 root 20 0 700 304 264 S 0 0.1 0:03.41 init

I got this for a bit cheaper AR off Newegg and it's a fine PSU (no modular cables) and it's not quiet. It's not super loud or anything but I can hear it when I'm in the same room.

From what I read people say Atom is OK for WHS and Unraid. I personally however am not happy with Atom performance in either my WHS or my Unraid. It is not a whole lot slower but just a bit. Given that energy savings are minimal going from something like a 45w Sempron to the Atom I myself seriously doubt it's worth it though.

Here are the results I have obtained when I initially tested Unraid perf. Those are not using the actual cache drive but from what I saw perf of "no parity" config is the same as with the cache drive. I saw ~100 MB/s transfers for the modern SATA drives. I currently have Hitachi 500 GB as the cache drive and performance over the network is identical to what I have benchmarked this drive in Windows previously locally and remotely. This is all on commodity hardware no fancy stuff. Single disk tests with no parity drive - 6.5 GB file Hitachi 500 GB 7200 rpm write - 80-90 MB/s read - 95-100 MB/s Seagate 2TB 5900 rpm write - 95-100 MB/s read - 100-120 MB/s Hitachi 2 TB 7200 rpm write: 90-100 MB/s read: 100-120 MB/s

Here is what I think. With the cache drive your write speeds are governed by the speed of the cache drive so if you put a WD Black in (or a Hitachi 2 TB which is a lot cheaper) you can get 100-120 MB/s of sequential write on large files (like media files). Without a cache drive your write speeds are going to be max around 30 MB/s which is not very much on GbE (basically it sucks compared to a regular unprotected file share). The only issue here is the small period of time when your data is not protected by parity but really if your data is that important you should be using backups and no Unraid. Hard drive failures are not very common these days (unless it is old hard drives or your system is poorly engineered aka too running too hot) so in my opinion it's not too much of an issue. Summary: if you care about your write speeds and have a spare hard disk - yes it is very worth it.

It looks like specifically for Sempron 140 the following data exists: - going from idle to Cool'N'Quiet saves around 4 watts - undervolting saves around 3 watts Source: http://www.techpowerup.com/reviews/AMD/Sempron_140/11.html So there I go - I might as well undervolt the CPU until 5.0 comes out as savings are similar.

Are power savings considerable or it's less than 1w? Thanks!

So since I'm running pretty common to Unraid hardware and I'm running the latest production version of Unraid I'm assuming this means that it is not really working in Unraid until 5.0.

Here is the result: Linux 2.6.32.9-unRAID. root@unraid:~# modprobe acpi-cpufreq FATAL: Module acpi_cpufreq not found. root@unraid:~# modprobe powernow-k6 FATAL: Module powernow_k6 not found. root@unraid:~# modprobe powernow-k7 FATAL: Module powernow_k7 not found. root@unraid:~# modprobe powernow-k8 FATAL: Module powernow_k8 not found. Is this expected or I did something wrong? I literally typed the commands from the Putty session, from whatever default directory it put me into.

Will try - thank you very much!

I saw that thread. I don't really understand what it says and what I need to do.

Is anyone here running their Unraid system with the PicoPSU? I just picked it up at local Fry's and it is rated at 150w. I'm able to spin my AMD-based Unraid system with 4 disks (2 WD Green, 2 misc 7200 RPM drives) and it seems to start and work fine. Has anyone found a practical limit of what amount of hardware PicoPSU can support? I love the efficiency of this little thing, my system consumes 15-20w less with it than with Antec SmartPower 500w PSU.

Hello everyone, I'm in final testing stages of my Unraid build with the following hardware: Gigabyte GA-MA78LM-S2H AM3 Motherboard Sempron 140 CPU 2 GB RAM 2 x 1 TB and 2 x 500 GB hard disks I'm able to get the power consumption to to 45w when the disks are spun down with 150w PicoPSU. I wonder if my CPU is running at full 2.7 GHz at that time or if it automatically clocks itself down. I've enabled both C1E and Cool'N'Quiet in BIOS but unmenu still shows 2.7 GHz. Can someone please explain whether Unraid is supposed to clock the CPU down when idle and whether that happens automatically? I did some search in the forums and it looks like it is supported but not enabled by default and some manual steps are needed which I wasn't able to find Thanks!

From my experiments when you have a cache drive your write speed is defined purely by the performance of your cache drive (assuming your source location is as fast and assuming your cache drive isn't full). The way it works is just a plain write to the cache drive. So basically if you use a modern SATA drive you should see anywhere from 80 MB/s to 120 MB/s sequential writes on large files and that is defined by what drive you pick. I was able to achieve those speeds by transferring a couple GB file. Once I disabled the cache my transfer speeds were 30-50% of the speed of the drive in the array - 30-35 MB/s.

I agree on expectations - I was coming from the fact that a single 7200 rpm drive can achieve sustained sequential 100 MB/s transfers over GbE from what I've seen and I was hoping that Unraid can provide that level of performance since it does not look it should do anything on reads or cached writes. My WHS box can provide reads at max speed of the actual drive I'm reading from and one of the reasons I'm trying Unraid is because write speeds are inconsistent in WHS. This is what I expected - thank you for confirming that!

Hello all, I'm new to Unraid and just set up a prototype server with 3x500 GB 7200 rpm drives. It seems to work fine and I'm moving to building my "production" environment. My goal is to saturate GbE network, i.e. get to approx 100 MB/s writes and reads. Reads were fine but writes were a bit slow, around 35 MB/s with parity. I'd like to get suggestions on how to architect my Unraid server for performance. I plan to use "green" drives for storage but I don't know if I should use a 7200 rpm drive for parity or not. Initially I thought I need a fast parity drive. But then I thought if I have a cache drive then overall performance is bound by the speed of the cache drive. Is that assumption correct? When will the cache drive start writing the data to the data drives? I hope it's after the writes from the clients have finished. Thanks all and I hope to get some good info!