antigravity Posted March 2, 2019 Share Posted March 2, 2019 Hi everyone New user to unRAID here, looks like a great piece of kit. I've been playing around, but can't seem to get write speeds higher than 50-60MB/s. Current setup 3x Seagate 5900rpm SATA3 in Array No parity or cache while testing Gigabit ethernet network over Cat6 N54L BIOS updated to latest modded v41 with cache writes turned onSpeeds Read speeds max out my network = 110MB/s Write speeds approx = 50-60MB/s When creating a Parity drive, speed is about 170MB/s This post below shows similar speeds to what I'm seeing, so I added these lines to the go config, but they didn't change my write speeds - just wondering if I need to do anything else but add the lines to the config file? https://forums.unraid.net/topic/28016-n54l-with-unraid-50-and-4x4-tb-1st-impressions/?do=findComment&comment=263227 echo 50 > /sys/devices/system/cpu/cpufreq/ondemand/up_threshold & logger Go Script - ondemand up_threshold set to 50 echo 50 > /sys/devices/system/cpu/cpufreq/ondemand/sampling_down_factor & logger Go Script - ondemand sampling_down_factor set to 50 Quote Link to comment
trurl Posted March 2, 2019 Share Posted March 2, 2019 Go to Tools - Diagnostics and attach the complete diagnostics zip file to your next post. 1 Quote Link to comment
antigravity Posted March 3, 2019 Author Share Posted March 3, 2019 (edited) 7 hours ago, trurl said: Go to Tools - Diagnostics and attach the complete diagnostics zip file to your next post. Sure, here is the info (see attached) My parity drive completed overnight, and got the following speeds when transferring a file this morning: Speeds start off at 110MB/s or so, but then drops off significantly. mckinley-diagnostics-20190303-1059.zip Edited March 3, 2019 by antigravity Quote Link to comment
trurl Posted March 3, 2019 Share Posted March 3, 2019 That write speed to the parity array is normal with the default setting for parity calculation. The burst of speed at the beginning is just memory caching. You should remove and ignore that stuff you got from that very old post. Here is more about the 2 different methods of writing parity: 1 Quote Link to comment
antigravity Posted March 3, 2019 Author Share Posted March 3, 2019 1 hour ago, trurl said: That write speed to the parity array is normal with the default setting for parity calculation. The burst of speed at the beginning is just memory caching. Thanks. I was more concerned about my first test, where I only got 60MB/s write to an array with no parity disk. If I get a cache drive, should it write at 110MB/s or is the fact I only got 60MB/s writing direct to the array without parity more representative of what I'll get with a cache drive? 1 hour ago, trurl said: You should remove and ignore that stuff you got from that very old post. Thanks will do. Quote Link to comment
trurl Posted March 3, 2019 Share Posted March 3, 2019 7 hours ago, antigravity said: I was more concerned about my first test, where I only got 60MB/s write to an array with no parity disk. That was the confusing part. Are you saying there was no disk assigned to the parity slot at all? Or are you just saying that parity hadn't finished building yet? 1 Quote Link to comment
antigravity Posted March 3, 2019 Author Share Posted March 3, 2019 (edited) 8 hours ago, trurl said: That was the confusing part. Are you saying there was no disk assigned to the parity slot at all? Or are you just saying that parity hadn't finished building yet? Sorry it was a bit confusing. My first test was done without any parity drive assigned. Write speeds direct to the array were 60MB/s My second test was done with a parity drive assigned and speeds were 40MB/s. My first test, getting 60MB/s without a parity drive, seemed low considering my reads were 110MB/s. My concern is if I purchase a cache drive, I'll only get 60MB/s as per my first test without parity. Or are write speeds to a cache drive quicker than when writing direct to an array without a parity drive? Edited March 3, 2019 by antigravity Quote Link to comment
JonathanM Posted March 3, 2019 Share Posted March 3, 2019 What was the source for the write tests? Quote Link to comment
antigravity Posted March 3, 2019 Author Share Posted March 3, 2019 25 minutes ago, jonathanm said: What was the source for the write tests? Desktop PC connected to the same Gigabit Ethernet. OC Ryzen 2700x, 32GB 3200mhz, 512GB NVMe PCIE SSD Desktop reads off unRAID at 110MB/s no issues. Quote Link to comment
trurl Posted March 3, 2019 Share Posted March 3, 2019 I don't have an explanation of your write speeds without parity. The syslog from the diagnostics you posted apparently don't cover the time of the test. Syslog is in RAM so is lost on reboot. Possibly something about that specific disk, but it might be difficult to pin it down without taking parity out again. There is a disk speed testing docker if you want to go down that path. What did you have in mind for cache? I recommend installing cache before you setup the docker and VM services so those get created on cache where they belong and you don't have to go to extra trouble to get them moved there. 1 Quote Link to comment
antigravity Posted March 4, 2019 Author Share Posted March 4, 2019 Hi @trurl Thanks for your help - its very much appreciated. I'll remove the parity, run the test again and re-post the syslog. Disk speeds seem to be OK when creating the Parity (150MB/s). If I used the same disk, should I expect the same write speeds to a non-parity array and a cache disk? I'm thinking of getting a 1TB SSD for the cache drive, but I'm concerned that there is another bottleneck (CPU?) causing the slow write speeds. I don't want to buy an SSD for cache, only to find I still get 60MB/s writes to the array. Quote Link to comment
antigravity Posted March 4, 2019 Author Share Posted March 4, 2019 Interesting. I deleted the Parity, reboot the system and now it is writing at 110MB/s Does the GO file load at startup, or only on reboot? Quote Link to comment
trurl Posted March 4, 2019 Share Posted March 4, 2019 53 minutes ago, antigravity said: only on reboot this^ Why do you ask. If you need a script you should probably take a look at the User Scripts plugin and leave go alone. 1 hour ago, antigravity said: If I used the same disk, should I expect the same write speeds to a non-parity array and a cache disk? All other things equal, yes. There could be controller bottlenecks, for example if you use port multipliers. 1 Quote Link to comment
antigravity Posted March 4, 2019 Author Share Posted March 4, 2019 32 minutes ago, trurl said: Why do you ask. If you need a script you should probably take a look at the User Scripts plugin and leave go alone. Because when I cold boot, I get 60MB/s writes, when I reboot I get 110MB/s writes. I still have the following lines in my GO config - which could explain why my writes are quicker upon reboot. echo 50 > /sys/devices/system/cpu/cpufreq/ondemand/up_threshold & logger Go Script - ondemand up_threshold set to 50 echo 50 > /sys/devices/system/cpu/cpufreq/ondemand/sampling_down_factor & logger Go Script - ondemand sampling_down_factor set to 50 34 minutes ago, trurl said: All other things equal, yes. There could be controller bottlenecks, for example if you use port multipliers. Ok thanks. Now that I know I can get 110MB/s write speeds upon reboot, I'm comfortable to spend money on a cache drive. Thanks for your help 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.