April 5, 201016 yr FYI, just a note as to why I bother setting up the ARECA in the SAFE33 environment. My parity drive is always spinning during the day. I access the data drives constantly over the internet from work. One of my data drives which handles torrents is spinning all the time too. Since I have two drives spinning all the time, It makes sense for me to set up a carved up SAFE environment. RAID0/RAID1 and let the two drives spin, but also do double duty. So I get faster parity writes, and a fast protected area for my torrents or other application servers. (mysql, cache drive, consolidated syslog host, etc, etc).
April 6, 201016 yr WeeboTech, great work as always. A couple more questions, mostly exploring what options are available in the Safe33 setup. On the Raid-1 portion, can it be partitioned to support cache disk and linux swap? Can it be further partitioned for a full Slackware install? If so, can one boot from there? Or does it need the hardware drivers loaded before it's able to read from there?
April 6, 201016 yr WeeboTech, great work as always. A couple more questions, mostly exploring what options are available in the Safe33 setup. On the Raid-1 portion, can it be partitioned to support cache disk and linux swap? Can it be further partitioned for a full Slackware install? If so, can one boot from there? Or does it need the hardware drivers loaded before it's able to read from there? To the OS. each of the Raid Volumes looks like a regular drive. So it can be partitioned as you mention. Tom gave us instructions a while back on the board. The First Partition (#1) must be the cache drive and needs to be #1. Any other partitions are not used. So if you set up 4 partitions as 1. cache 2. swap 3. slackware The raid1 drive can be used and booted. I have not tried it yet, but plan to. Although I will be setting up partition #4 as fat32 and putting the syslinux unraid Distro there. Then in the slackware dev system, I'll mount it on /unraid/boot so I can update it from slackware. Just so it is clear, The SAFE50/SAFE33 is a Silcon Image Chipset notation for dividing up two drives. I called it that because I have it the same way. On the ARECA bios it is called a RAIDSET (# of drives which are linked together). Then a VOLUME SET (These are the RAIDSET drives which are carved up, 1 volume RAID0, 1 Volume RAID1). I suppose if you had a 3rd drive, it could be part of the RAID Set and a hot spare for the RAID1 volume. In any case, these can be carved up in any respective sizes. I used two 1.5tb drives and had a full 2TB volume (Which was actually larger then a 2TB drive by a small margin). Then used the rest for the RAID1 volume. This came out to about 400GB.
April 6, 201016 yr Weebotech, it's good to know that's a possibility. I currently have my system setup similarly on a single drive (cache / swap / win os / slackware). What I am curious about is how does a raid-0 set of 5400/Green drives perform with a 7200 rpm (or even SSD) data drive. Is the raid0 between two slower drives enough to counteract their higher latency? At first when you mentioned you found a way to speed up parity, I thought maybe you worked something out with tweaking of filesystem/device buffers or even using multithreading for a) reading into a buffer b) calculating parity from the buffer.
April 7, 201016 yr What I am curious about is how does a raid-0 set of 5400/Green drives perform with a 7200 rpm (or even SSD) data drive. Is the raid0 between two slower drives enough to counteract their higher latency? I'll have to test it out, although I'm not sure it is the best way to go. I'm sure the caching controller helps in this situation. At first when you mentioned you found a way to speed up parity, I thought maybe you worked something out with tweaking of filesystem/device buffers or even using multithreading for a) reading into a buffer b) calculating parity from the buffer. I have not gone that route, still. Adjusting the scheduler or md stripe parameters could have an effect if we can write to the parity drive faster. although, I think where the RAID0 helps is in the reads. The samsung Spinpoint F3 1TB's benchmark at up to 149MB/s reads. really stellar numbers if you ask me. I would be curious to see how two Hitachi 7K 2TB drives perform. I would have purchased them only I read they run pretty hot.
April 7, 201016 yr FWIW, some benchmarks. This is raw write/read speed. 2 samsung spinpoint F3 1tb 7200rpm 32MB cache. This is more like a SAFE10 root@unraid ~ #df -vH Filesystem Size Used Avail Use% Mounted on /dev/sdh1 518M 491M 28M 95% /boot /dev/sdb1 150G 34M 150G 1% /mnt/sdb1 RAID1 /dev/sda1 1.7T 34M 1.7T 1% /mnt/sda1 RAID0 Areca RAID0 root@unraid ~ #writeread10gb /mnt/sda1/test.dd writing 4096000000 bytes to: /mnt/sda1/test.dd 726015+0 records in 726015+0 records out 743439360 bytes (743 MB) copied, 5.0064 s, 148 MB/s 1383311+0 records in 1383311+0 records out 1416510464 bytes (1.4 GB) copied, 10.0211 s, 141 MB/s 2090981+0 records in 2090981+0 records out 2141164544 bytes (2.1 GB) copied, 15.0351 s, 142 MB/s 2861533+0 records in 2861533+0 records out 2930209792 bytes (2.9 GB) copied, 20.125 s, 146 MB/s 3518678+0 records in 3518678+0 records out 3603126272 bytes (3.6 GB) copied, 25.0605 s, 144 MB/s 4000000+0 records in 4000000+0 records out 4096000000 bytes (4.1 GB) copied, 28.0151 s, 146 MB/s write complete, syncing reading from: /mnt/sda1/test.dd 4000000+0 records in 4000000+0 records out 4096000000 bytes (4.1 GB) copied, 9.29826 s, 441 MB/s removing: /mnt/sda1/test.dd removed `/mnt/sda1/test.dd' root@unraid ~ # Areca RAID1 volume set (inside tracks) root@unraid ~ #writeread10gb /mnt/sdb1/test.dd writing 4096000000 bytes to: /mnt/sdb1/test.dd 505597+0 records in 505597+0 records out 517731328 bytes (518 MB) copied, 6.03984 s, 85.7 MB/s 898253+0 records in 898253+0 records out 919811072 bytes (920 MB) copied, 10.3446 s, 88.9 MB/s 1354112+0 records in 1354112+0 records out 1386610688 bytes (1.4 GB) copied, 15.0395 s, 92.2 MB/s 1813101+0 records in 1813101+0 records out 1856615424 bytes (1.9 GB) copied, 20.3317 s, 91.3 MB/s 2215710+0 records in 2215710+0 records out 2268887040 bytes (2.3 GB) copied, 25.0659 s, 90.5 MB/s 2598413+0 records in 2598413+0 records out 2660774912 bytes (2.7 GB) copied, 30.3255 s, 87.7 MB/s 2992421+0 records in 2992421+0 records out 3064239104 bytes (3.1 GB) copied, 35.0916 s, 87.3 MB/s 3383725+0 records in 3383725+0 records out 3464934400 bytes (3.5 GB) copied, 40.2116 s, 86.2 MB/s 3855033+0 records in 3855033+0 records out 3947553792 bytes (3.9 GB) copied, 45.1206 s, 87.5 MB/s 4000000+0 records in 4000000+0 records out 4096000000 bytes (4.1 GB) copied, 46.8394 s, 87.4 MB/s write complete, syncing reading from: /mnt/sdb1/test.dd 4000000+0 records in 4000000+0 records out 4096000000 bytes (4.1 GB) copied, 21.1475 s, 194 MB/s removing: /mnt/sdb1/test.dd removed `/mnt/sdb1/test.dd' and s'more numbers in this configuration. 2 Seagate 1.5tb RAID0 outer tracks parity 2 Seagate 1.5tb RAID1 inner tracks /mnt/cache 1 ata-SAMSUNG_HD103SJ /mnt/disk1 1 ata-WDC_WD20EADS /mnt/disk2 root@unraid ~ #df -vH Filesystem Size Used Avail Use% Mounted on /dev/sdh1 518M 491M 28M 95% /boot /dev/sdb1 401G 34M 401G 1% /mnt/cache /dev/md1 1.1T 34M 1.1T 1% /mnt/disk1 /dev/md2 2.1T 34M 2.1T 1% /mnt/disk2 writing 4096000000 bytes to: /mnt/cache/test.dd 448525+0 records in 448525+0 records out 459289600 bytes (459 MB) copied, 5.01436 s, 91.6 MB/s 764669+0 records in 764669+0 records out 783021056 bytes (783 MB) copied, 10.811 s, 72.4 MB/s 1027789+0 records in 1027789+0 records out 1052455936 bytes (1.1 GB) copied, 15.2366 s, 69.1 MB/s 1420445+0 records in 1420445+0 records out 1454535680 bytes (1.5 GB) copied, 20.3366 s, 71.5 MB/s 1813101+0 records in 1813101+0 records out 1856615424 bytes (1.9 GB) copied, 25.5465 s, 72.7 MB/s 2145394+0 records in 2145394+0 records out 2196883456 bytes (2.2 GB) copied, 30.083 s, 73.0 MB/s 2468877+0 records in 2468877+0 records out 2528130048 bytes (2.5 GB) copied, 37.9191 s, 66.7 MB/s 2667229+0 records in 2667229+0 records out 2731242496 bytes (2.7 GB) copied, 40.1651 s, 68.0 MB/s 2991069+0 records in 2991069+0 records out 3062854656 bytes (3.1 GB) copied, 45.5686 s, 67.2 MB/s 3254189+0 records in 3254189+0 records out 3332289536 bytes (3.3 GB) copied, 50.2236 s, 66.3 MB/s 3646845+0 records in 3646845+0 records out 3734369280 bytes (3.7 GB) copied, 55.3009 s, 67.5 MB/s 4000000+0 records in 4000000+0 records out 4096000000 bytes (4.1 GB) copied, 59.481 s, 68.9 MB/s write complete, syncing reading from: /mnt/cache/test.dd 000000+0 records in 4000000+0 records out 4096000000 bytes (4.1 GB) copied, 33.197 s, 123 MB/s removing: /mnt/cache/test.dd removed `/mnt/cache/test.dd' write to parity protrected samsung 1tb drive. root@unraid ~ #writeread10gb /mnt/disk1/test.dd writing 4096000000 bytes to: /mnt/disk1/test.dd 279896+0 records in 279896+0 records out 286613504 bytes (287 MB) copied, 5.0469 s, 56.8 MB/s 497737+0 records in 497737+0 records out 509682688 bytes (510 MB) copied, 10.0713 s, 50.6 MB/s 709625+0 records in 709625+0 records out 726656000 bytes (727 MB) copied, 15.0388 s, 48.3 MB/s 931833+0 records in 931833+0 records out 954196992 bytes (954 MB) copied, 20.0544 s, 47.6 MB/s 1110009+0 records in 1110009+0 records out 1136649216 bytes (1.1 GB) copied, 25.0654 s, 45.3 MB/s 1314809+0 records in 1314809+0 records out 1346364416 bytes (1.3 GB) copied, 30.0813 s, 44.8 MB/s 1547265+0 records in 1547265+0 records out 1584399360 bytes (1.6 GB) copied, 35.2883 s, 44.9 MB/s 1754113+0 records in 1754113+0 records out 1796211712 bytes (1.8 GB) copied, 41.0101 s, 43.8 MB/s 1926145+0 records in 1926145+0 records out 1972372480 bytes (2.0 GB) copied, 45.1903 s, 43.6 MB/s 2147329+0 records in 2147329+0 records out 2198864896 bytes (2.2 GB) copied, 50.1333 s, 43.9 MB/s 2367533+0 records in 2367532+0 records out 2424352768 bytes (2.4 GB) copied, 55.1457 s, 44.0 MB/s 2595841+0 records in 2595841+0 records out 2658141184 bytes (2.7 GB) copied, 61.0117 s, 43.6 MB/s 2768830+0 records in 2768830+0 records out 2835281920 bytes (2.8 GB) copied, 65.1832 s, 43.5 MB/s 2984325+0 records in 2984324+0 records out 3055947776 bytes (3.1 GB) copied, 70.1872 s, 43.5 MB/s 3200014+0 records in 3200014+0 records out 3276814336 bytes (3.3 GB) copied, 75.1947 s, 43.6 MB/s 3425281+0 records in 3425281+0 records out 3507487744 bytes (3.5 GB) copied, 80.2551 s, 43.7 MB/s 3606537+0 records in 3606537+0 records out 3693093888 bytes (3.7 GB) copied, 85.7637 s, 43.1 MB/s 3786769+0 records in 3786769+0 records out 3877651456 bytes (3.9 GB) copied, 90.2319 s, 43.0 MB/s 3999761+0 records in 3999761+0 records out 4095755264 bytes (4.1 GB) copied, 95.3352 s, 43.0 MB/s 4000000+0 records in 4000000+0 records out 4096000000 bytes (4.1 GB) copied, 95.3366 s, 43.0 MB/s write complete, syncing reading from: /mnt/disk1/test.dd 4000000+0 records in 4000000+0 records out 4096000000 bytes (4.1 GB) copied, 22.956 s, 178 MB/s removing: /mnt/disk1/test.dd removed `/mnt/disk1/test.dd' write to parity protected eads2tb root@unraid ~ #writeread10gb /mnt/disk2/test.dd writing 4096000000 bytes to: /mnt/disk2/test.dd 236065+0 records in 236065+0 records out 241730560 bytes (242 MB) copied, 5.33615 s, 45.3 MB/s 408561+0 records in 408561+0 records out 418366464 bytes (418 MB) copied, 10.6804 s, 39.2 MB/s 582641+0 records in 582641+0 records out 596624384 bytes (597 MB) copied, 15.0393 s, 39.7 MB/s 740337+0 records in 740337+0 records out 758105088 bytes (758 MB) copied, 20.1382 s, 37.6 MB/s 898253+0 records in 898253+0 records out 919811072 bytes (920 MB) copied, 25.0746 s, 36.7 MB/s 1068525+0 records in 1068525+0 records out 1094169600 bytes (1.1 GB) copied, 30.0627 s, 36.4 MB/s 1213425+0 records in 1213425+0 records out 1242547200 bytes (1.2 GB) copied, 35.0884 s, 35.4 MB/s 1392625+0 records in 1392625+0 records out 1426048000 bytes (1.4 GB) copied, 40.095 s, 35.6 MB/s 1552369+0 records in 1552369+0 records out 1589625856 bytes (1.6 GB) copied, 45.0936 s, 35.3 MB/s 1719281+0 records in 1719281+0 records out 1760543744 bytes (1.8 GB) copied, 50.2372 s, 35.0 MB/s 1876977+0 records in 1876977+0 records out 1922024448 bytes (1.9 GB) copied, 55.1139 s, 34.9 MB/s 2045697+0 records in 2045697+0 records out 2094793728 bytes (2.1 GB) copied, 60.124 s, 34.8 MB/s 2204657+0 records in 2204657+0 records out 2257568768 bytes (2.3 GB) copied, 65.1355 s, 34.7 MB/s 2371569+0 records in 2371569+0 records out 2428486656 bytes (2.4 GB) copied, 70.1927 s, 34.6 MB/s 2534385+0 records in 2534385+0 records out 2595210240 bytes (2.6 GB) copied, 75.1587 s, 34.5 MB/s 2710513+0 records in 2710513+0 records out 2775565312 bytes (2.8 GB) copied, 80.1699 s, 34.6 MB/s 2883473+0 records in 2883473+0 records out 2952676352 bytes (3.0 GB) copied, 85.1825 s, 34.7 MB/s 3038193+0 records in 3038193+0 records out 3111109632 bytes (3.1 GB) copied, 90.1961 s, 34.5 MB/s 3203057+0 records in 3203057+0 records out 3279930368 bytes (3.3 GB) copied, 95.2295 s, 34.4 MB/s 3364849+0 records in 3364849+0 records out 3445605376 bytes (3.4 GB) copied, 100.261 s, 34.4 MB/s 3528689+0 records in 3528689+0 records out 3613377536 bytes (3.6 GB) copied, 105.337 s, 34.3 MB/s 3703793+0 records in 3703793+0 records out 3792684032 bytes (3.8 GB) copied, 110.236 s, 34.4 MB/s 3861489+0 records in 3861489+0 records out 3954164736 bytes (4.0 GB) copied, 115.245 s, 34.3 MB/s 4000000+0 records in 4000000+0 records out 4096000000 bytes (4.1 GB) copied, 119.234 s, 34.4 MB/s write complete, syncing reading from: /mnt/disk2/test.dd 4000000+0 records in 4000000+0 records out 4096000000 bytes (4.1 GB) copied, 29.4291 s, 139 MB/s removing: /mnt/disk2/test.dd removed `/mnt/disk2/test.dd'
April 7, 201016 yr Weebotech, it's good to know that's a possibility. I currently have my system setup similarly on a single drive (cache / swap / win os / slackware). BRiT, could you please share how have you done the partitioning and activation of the drive. I plan something similar with cache/swap/TimeMachine partitions on the same drive (minus the OS'es however). Perhaps also holding the SlimCenter binaries but not sure on that. This drive I don't want to be protected at all but it should be checked and spun down if there is no activity on it (perhaps cron job each hour?). Thanks!
April 8, 201016 yr I followed the steps listed here for the partitioning -- http://lime-technology.com/forum/index.php?topic=3899.0
July 2, 201016 yr Why Not add a Second parity drive simply for redundancy / additional fail safe. It seems to me that the Achilles heel of the unraid system is the single parity drive. If several data drives fail you can always reconstruct the data from the parity. But if the parity & a data drive fail all data would be lost. A dual parity setup would effectively be like a raid 6 and provide much much greater fault redundancy Why not add the option for a second parity drive?
July 2, 201016 yr You don't understand how unRAID works to make that statement. It's not true in unRAID's case. With unRAID, if you lose the parity drive and a data drive, you only lose the contents on that single data drive. All the other data drives are perfectly fine.
Archived
This topic is now archived and is closed to further replies.