April 1, 201214 yr Firstly I would like to say thanks for such cool product Secondly, I have spent hours reading through the forums and trying various things but I am not getting the performance I would think I should. My setup is Mobo=GA-945GCM-S2C CPU=C2D 2.8Ghz Ram=4GB Parity=ST32000641AS Datax2=ST2000DM001 Cache=WD1500HLS I am only getting about 20-22MB/s write speeds with this setup, from both my mac and my windows machine, the windows is slightly faster around 24MB/s Network is 1Gbps I did the SMB config file change for unix. And also the ifconfig eth0 mtu change, no changes. During the parity build I saw it was running around 120MB/s. Included is my syslog which I don't see anything that should looks wrong, however I am a bit of a unix noob. Thanks in advance I realized I wasn't using the cache drive, my fault, speeds are much faster with the cache drive, but then they look slow watching disk performance to the 2TB drives from the cache drive. I know the 2TB drives i have should be faster than 20-25MB/s they are 7200rpm 64MB Cache drives. I looked at the hardware configuration wiki, and I noticed that the main drive that is getting used for storage and the parity drive were on the same bus controller for the sata controller, so I switched them and it seems to have helped, I am not sustaining about 30MB/s give or take.
April 1, 201214 yr I'm a noob at this as well, but I think I've read that 20MB/s - 40MB/s is normal (that's in the FAQ I believe). This is because it has to update the parity drive as you are writing data. Reads should be much faster.
April 2, 201214 yr 20MB/sec to 35MB/sec is about all you are going to get, even with the fastest hard drives and best performing hardware you can buy. Unraid is not made for performance and is expected to be slower then most RAID type systems. If you are or wanting to video edit and / or stuff like that then don't think unraid will work.
April 2, 201214 yr Author Thanks for the replies gents, I didn't realize the limitations on speed, I don't really need super fast, just using it as a backup storage for various media. The cache drive really helps On a side note if I want my cache drive to dump over every 30 minutes is this the right config? */30 * * * *
April 2, 201214 yr I'm a noob at this as well, but I think I've read that 20MB/s - 40MB/s is normal (that's in the FAQ I believe). This is because it has to update the parity drive as you are writing data. Reads should be much faster. Mostly correct, but it has to read and then write the parity disk and the disk being written with a revolution of the platter in between. The physical rotation of the platters is the limiting factor. Faster spinning/higher areal density drives can be written faster.
April 2, 201214 yr Thanks for the replies gents, I didn't realize the limitations on speed, I don't really need super fast, just using it as a backup storage for various media. The cache drive really helps On a side note if I want my cache drive to dump over every 30 minutes is this the right config? */30 * * * * It should work just fine. I personally used when I had a cache drive. (unRAID is a bit dumb and puts the whole string in the cron file) */5 * * * * /boot/are_disks_idle.sh && The /boot/are_disks_idle.sh script lets the mover start only if the disks are idle and the array is online. It wakes up every five minutes and if the drives are in use, it does not move anything. If all disks are idle, it moves whatever is in cache. I don't usually run a cache drive. I found it to not help me in how I use unRAID. I prefer to write to the /mnt/diskX drives, and read from /mnt/user-shares the are_disks_idle.sh script looks like this: #!/bin/bash ############################################################## # are_disks_idle.sh # exit status = 0 if all disks idle # exit status = 1 if any disk spinning # exit status = 2 if array is not started or cache not present # # April 2010 Joe L. # Check if the array is started if [ -d /mnt/cache0 ] then # rdevLastIO will be non-zero if a disk is spinning last=`/root/mdcmd status | grep -a rdevLastIO | grep -v '=0'` if [ "$last}" = "" ] then # all disks are idle exit 0 else # all disks are not idle exit 1 fi else # array is not started exit 2 fi Since my older server is a slower PCI based machine, I did not want to use the PCI bus bandwidth for moving files off of the cache if I was actively using the array. Joe L.
Archived
This topic is now archived and is closed to further replies.