pctine wrote:
1 & 3 很容易...(恕刪)
感謝P大的回復
單純覺得可惜罷了
畢竟裝了SSD，就是希望能提升I/O表現
而使用BTRFS就是因為其中的檔案系統亮點 "自我修復"、"快照"等等
希望官方能想到開啟SSD讀寫快取時的自我修復的方法囉!!
目前改成唯讀快取，系統變慢了 不習慣阿
個人積分：6分
文章編號：68461778
個人積分：1199分
文章編號：68513756
個人積分：1169分
文章編號：68591823
個人積分：4958分
文章編號：68979835
個人積分：4958分
文章編號：69532335
前日在 momo 出現了 intel 企業級 SSD DC S3500 480GB $3899 的特價, 經請示師父後建議這雖然是前幾代的產品, 但全新品仍能提供五年保固, 沒有理由不買, 所以就一次訂了四顆.
這次在 ds3018xs 上面裝了三顆建立 raid0 & btrfs, 試了一下讀寫效能.
以 fio 測試 NAS 內部的循序讀寫效能, 三顆組成 raid0 達到 1100MB/s 以上, 這樣也剛好可以吃滿 ds3018xs 10G 網卡的頻寬.
root@DS13:/volume2/public# fio fio2.conf
read: (g=0): rw=read, bs=64K-64K/64K-64K/64K-64K, ioengine=libaio, iodepth=256
write: (g=1): rw=write, bs=64K-64K/64K-64K/64K-64K, ioengine=libaio, iodepth=256
fio-2.13
Starting 2 processes
read: Laying out IO file(s) (1 file(s) / 16384MB)
write: Laying out IO file(s) (1 file(s) / 16384MB)
Jobs: 1 (f=1): [_(1),W(1)] [100.0% done] [0KB/1131MB/0KB /s] [0/18.1K/0 iops] [eta 00m:00s]
read: (groupid=0, jobs=1): err= 0: pid=22905: Wed Aug 1 09:22:03 2018
read : io=16384MB, bw=1166.2MB/s, iops=18657, runt= 14050msec
slat (usec): min=9, max=611, avg=18.02, stdev=17.98
clat (usec): min=290, max=40878, avg=13697.84, stdev=10272.71
lat (usec): min=305, max=40889, avg=13715.93, stdev=10271.51
clat percentiles (usec):
| 1.00th=[ 596], 5.00th=[ 1012], 10.00th=[ 1320], 20.00th=[ 1928],
| 30.00th=[ 2736], 40.00th=[10944], 50.00th=[12992], 60.00th=[14912],
| 70.00th=[22400], 80.00th=[25216], 90.00th=[28288], 95.00th=[30080],
| 99.00th=[31616], 99.50th=[32128], 99.90th=[34560], 99.95th=[36096],
| 99.99th=[39168]
bw (MB /s): min= 1154, max= 1177, per=100.00%, avg=1166.45, stdev= 5.54
lat (usec) : 500=0.59%, 750=1.62%, 1000=2.59%
lat (msec) : 2=16.24%, 4=11.89%, 10=2.80%, 20=31.19%, 50=33.07%
cpu : usr=2.18%, sys=35.67%, ctx=70668, majf=0, minf=4106
IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%
submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.1%
issued : total=r=262144/w=0/d=0, short=r=0/w=0/d=0, drop=r=0/w=0/d=0
latency : target=0, window=0, percentile=100.00%, depth=256
write: (groupid=1, jobs=1): err= 0: pid=23072: Wed Aug 1 09:22:03 2018
write: io=16384MB, bw=1184.3MB/s, iops=18947, runt= 13835msec
slat (usec): min=11, max=4732, avg=20.14, stdev=20.14
clat (usec): min=155, max=48912, avg=13469.45, stdev=9315.89
lat (usec): min=173, max=48928, avg=13489.70, stdev=9314.72
clat percentiles (usec):
| 1.00th=[ 516], 5.00th=[ 812], 10.00th=[ 1064], 20.00th=[ 1624],
| 30.00th=[ 3152], 40.00th=[12992], 50.00th=[16192], 60.00th=[17792],
| 70.00th=[19584], 80.00th=[21376], 90.00th=[24448], 95.00th=[27776],
| 99.00th=[31872], 99.50th=[33024], 99.90th=[35072], 99.95th=[36608],
| 99.99th=[45312]
bw (MB /s): min= 1122, max= 1236, per=100.00%, avg=1186.18, stdev=43.53
lat (usec) : 250=0.28%, 500=0.58%, 750=3.07%, 1000=4.67%
lat (msec) : 2=15.63%, 4=7.24%, 10=3.66%, 20=37.13%, 50=27.73%
cpu : usr=31.37%, sys=39.04%, ctx=95970, majf=0, minf=9
IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%
submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.1%
issued : total=r=0/w=262144/d=0, short=r=0/w=0/d=0, drop=r=0/w=0/d=0
latency : target=0, window=0, percentile=100.00%, depth=256
Run status group 0 (all jobs):
READ: io=16384MB, aggrb=1166.2MB/s, minb=1166.2MB/s, maxb=1166.2MB/s, mint=14050msec, maxt=14050msec
Run status group 1 (all jobs):
WRITE: io=16384MB, aggrb=1184.3MB/s, minb=1184.3MB/s, maxb=1184.3MB/s, mint=13835msec, maxt=13835msec
AJA 測試的結果:
ps: ds3018xs 使用 E10G17-F2 10G card, pc 端為 Mellanox ConnectX-2 10G NIC.
FB: Pctine
個人積分：4958分
文章編號：71117430
三顆組 RAID5, 使用 btrfs
windows pc i5-8400 使用 Mellanox ConnectX-2 10G NIC.
DS3018xs 為 Mellanox ConnectX-3 10G NIC,
大檔讀取已近 10G NIC 極速. 寫入約在 700MB/s
MX500 和 Intel SSD 溫度相差較大, 不過這和 ssd sensor 所偵測位置也有關.
SSD 大降價, 要組 all flash NAS 的人會愈來愈多吧.
FB: Pctine
個人積分：4958分
文章編號：71117503
read: 13xxMB/sec
write: 6xxMB/sec
root@DS13:/volume2/data# fio fio2.conf
read: (g=0): rw=read, bs=64K-64K/64K-64K/64K-64K, ioengine=libaio, iodepth=256
write: (g=1): rw=write, bs=64K-64K/64K-64K/64K-64K, ioengine=libaio, iodepth=256
fio-2.13
Starting 2 processes
Jobs: 1 (f=1): [_(1),W(1)] [100.0% done] [0KB/530.7MB/0KB /s] [0/8490/0 iops] [eta 00m:00s]
read: (groupid=0, jobs=1): err= 0: pid=18217: Mon Nov 26 09:50:35 2018
read : io=16384MB, bw=1389.8MB/s, iops=22236, runt= 11789msec
slat (usec): min=11, max=336, avg=18.80, stdev= 8.84
clat (usec): min=109, max=57242, avg=11488.43, stdev=12924.71
lat (usec): min=157, max=57256, avg=11507.31, stdev=12923.84
clat percentiles (usec):
| 1.00th=[ 145], 5.00th=[ 171], 10.00th=[ 249], 20.00th=[ 828],
| 30.00th=[ 1464], 40.00th=[ 2128], 50.00th=[ 3536], 60.00th=[ 6688],
| 70.00th=[22656], 80.00th=[28544], 90.00th=[30848], 95.00th=[34560],
| 99.00th=[35584], 99.50th=[36096], 99.90th=[36608], 99.95th=[40192],
| 99.99th=[53504]
bw (MB /s): min= 1007, max= 1378, per=79.46%, avg=1104.34, stdev=74.75
lat (usec) : 250=10.04%, 500=3.26%, 750=3.51%, 1000=6.18%
lat (msec) : 2=15.45%, 4=15.44%, 10=9.14%, 20=5.65%, 50=31.30%
lat (msec) : 100=0.02%
cpu : usr=3.78%, sys=43.76%, ctx=134894, majf=0, minf=4106
IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%
submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.1%
issued : total=r=262144/w=0/d=0, short=r=0/w=0/d=0, drop=r=0/w=0/d=0
latency : target=0, window=0, percentile=100.00%, depth=256
write: (groupid=1, jobs=1): err= 0: pid=18342: Mon Nov 26 09:50:35 2018
write: io=16384MB, bw=614303KB/s, iops=9598, runt= 27311msec
slat (usec): min=19, max=17833, avg=47.29, stdev=103.96
clat (msec): min=1, max=234, avg=26.60, stdev=19.29
lat (msec): min=2, max=234, avg=26.64, stdev=19.29
clat percentiles (msec):
| 1.00th=[ 10], 5.00th=[ 13], 10.00th=[ 15], 20.00th=[ 17],
| 30.00th=[ 19], 40.00th=[ 20], 50.00th=[ 22], 60.00th=[ 23],
| 70.00th=[ 26], 80.00th=[ 31], 90.00th=[ 44], 95.00th=[ 63],
| 99.00th=[ 111], 99.50th=[ 143], 99.90th=[ 202], 99.95th=[ 217],
| 99.99th=[ 229]
bw (KB /s): min=459264, max=669696, per=100.00%, avg=614645.91, stdev=35015.15
lat (msec) : 2=0.01%, 4=0.01%, 10=1.74%, 20=40.33%, 50=50.18%
lat (msec) : 100=6.48%, 250=1.26%
cpu : usr=16.36%, sys=44.19%, ctx=48627, majf=0, minf=9
IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=0.1%, 32=0.1%, >=64=100.0%
submit : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
complete : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.1%
issued : total=r=0/w=262144/d=0, short=r=0/w=0/d=0, drop=r=0/w=0/d=0
latency : target=0, window=0, percentile=100.00%, depth=256
Run status group 0 (all jobs):
READ: io=16384MB, aggrb=1389.8MB/s, minb=1389.8MB/s, maxb=1389.8MB/s, mint=11789msec, maxt=11789msec
Run status group 1 (all jobs):
WRITE: io=16384MB, aggrb=614302KB/s, minb=614302KB/s, maxb=614302KB/s, mint=27311msec, maxt=27311msec
FB: Pctine
個人積分：4958分
文章編號：72205713
當 share folder 很多時, 透過 DSM web ui 查詢各個 share folder 的權限設定就有點費時, 透過 command line 會方便些.
root@nas01:~# synoshare --enum ALL
Share Enum Arguments: [0xF0F] ALL ENC DEC
9 Listed:
backup
database
homes
music
photo
scan
software
temp
web
root@nas01:~# synoshare --get music
SYNOSHARE data dump:
Name .......[music]
Comment ....[]
Path .......[/volume1/music]
Deny list ..[]
RW list ....[@[email protected],@[email protected],admin]
RO list ....[@administrators]
fType ......[2]
fBrowseable [yes]
FTPPrivilege[0]
Status .....[0x2100]
WinShare .....[no]
ACL ..........[no]
Skip smb perm.[no]
Permit .......[yes]
RecycleBin....[yes]
RecycleBinAdminOnly....[no]
HideUnreadable ........[no]
Snapshot browsing .....[yes]
synoshare --enum ALL > sharefolder.txt
while read LINE; do synoshare --get "$LINE" | cut -f1 -d":"; done < ./sharefolder.txt > dump.txt
FB: Pctine
個人積分：4958分
文章編號：72603414
常常有網友詢問查看系統資源時, 常常發現 NAS CPU loading 很高, 尤其是在較大的 NAS 版本更新時, 在前幾次較大的版更, 或者是系統初次建置, 將大量檔案 copy 到 NAS, 此時 NAS 常常需要處理相片的縮圖以及檔案的索引, 這些都是相當耗 NAS 系統資源的.
DS3018xs NAS 的硬體效能較好, 在處理這些索引及縮圖相對快多了, 但有些入門級的 NAS, 建議在索引部份還是再做些細項的調校, 對於不需要索引的資料夾, 建議將其排除在外.
在 Synology Drive 套件安裝後, 系統預設只要是 team folder(團隊資料夾) 就一定會加入索引, 這設計立意良好, 既是團隊資料夾, 當然希望透過 Universal search 可以很快的查詢到檔案, 甚至檔案內容及 tags 等, 但有時候建立 team folder 只是為了同步/備份的需求. 此時讓 NAS 去做 index 意義不大. 這時就可以將其全文檢索的功能取消掉.
進入 universal search 設定. 逐一檢視 '索引資料夾清單'
對於#1 檔名的索引這部份預設是啟用, 它並不會太耗資源, 但 #2 檔案內容的索引, 此時就可以視需求設定, 例如不需要全文檢索, 就把 '文件' 取消打勾.
同樣的, 對於 'Drive team folder' 也能做如此細項的微調.
經過調校, 相信會更貼近各人使用 NAS 的習慣.
FB: Pctine
關閉廣告