March 20, 20251 yr Dear Support Team, I am currently running a Parity-Check on my Unraid server and I am surprised by how long it is taking, considering my setup and the relatively small amount of data in use. Quote My setup: Parity Drive: 4TB Data Drives: 4TB (XFS) + 1TB (XFS) Data Usage: Less than 500GB in total Current Speed: ~49.3 MB/s Elapsed Time: Over 2 days and still not finished Everything appears to be working correctly, but I was expecting a faster check given the current data usage. Is this speed considered normal for my configuration, or could there be an underlying issue affecting performance? I have attached the diagnostics file in case it helps with troubleshooting. Any insights or recommendations would be greatly appreciated. Thank you in advance for your support! takezo tower-diagnostics-20250320-1726.zip
March 20, 20251 yr Community Expert Data usage has no effect. Only size of parity drive. Parity doesn't know anything about data usage. Normally 4TB parity check would complete in about 8 hours if it were allowed to run without pausing. Mar 18 15:53:25 Tower kernel: mdcmd (36): check correct Mar 18 15:53:25 Tower kernel: md: recovery thread: check P ... Mar 18 15:53:49 Tower kernel: mdcmd (37): nocheck pause Mar 18 15:53:49 Tower kernel: md: recovery thread: exit status: -4 Mar 18 18:38:01 Tower kernel: mdcmd (38): check resume Mar 18 18:38:01 Tower kernel: md: recovery thread: check P ... Mar 18 22:10:39 Tower kernel: mdcmd (39): nocheck pause Mar 20 17:18:20 Tower kernel: mdcmd (40): check resume Mar 20 17:18:20 Tower kernel: md: recovery thread: check P ... Mar 20 17:21:54 Tower kernel: mdcmd (41): nocheck pause Mar 20 17:22:07 Tower kernel: md: recovery thread: exit status: -4 11 minutes ago, takezo said: Over 2 days and still not finished Looks like it was paused for most of that time. And was still paused when the diagnostics were taken. Perhaps you forgot to restart it after pausing.
March 20, 20251 yr Author thanks for your support! I paused the parity check to run diagnostics and check the disks. I uploaded another diagnostic with the parity running. tower-diagnostics-20250320-1804.zip
March 20, 20251 yr Community Expert Mar 20 17:37:03 Tower kernel: mdcmd (42): check resume Mar 20 17:37:03 Tower kernel: md: recovery thread: check P ... What do you mean by this? 3 minutes ago, takezo said: check the disks
March 20, 20251 yr Author 8 minutes ago, trurl said: Mar 20 17:37:03 Tower kernel: mdcmd (42): check resume Mar 20 17:37:03 Tower kernel: md: recovery thread: check P ... What do you mean by this? physical visual inspection only :-) and also run SMART short self-test in each disk that looks are ok
March 20, 20251 yr Community Expert 1 minute ago, takezo said: also run SMART short self-test That would slow it down. The fact that is was paused for nearly 2 days is the main thing obviously.
March 20, 20251 yr Author 25 minutes ago, trurl said: That would slow it down. The fact that is was paused for nearly 2 days is the main thing obviously. Yes ,but right now it's checking parity and Current Speed: ~49.3 MB/s ,so question here is why so slow?
March 20, 20251 yr Community Expert Solution 37 minutes ago, takezo said: Yes ,but right now it's checking parity and Current Speed: ~49.3 MB/s ,so question here is why so slow? Its as fast as the slowest disk is, if my memory serves me well, the 1 TB you use should absolutely fall in that speed that you observe, especially once you reach the end of the platter. If im not mistaken now, you should be about to reach the 1TB threshold, once that it is cleared, it should get faster since that drive does not contain any data to be tracked (if nothing is using the array at the same time) Edited March 20, 20251 yr by Mainfrezzer
March 20, 20251 yr Author 33 minutes ago, JorgeB said: Post new diags while the parity check is running. its already attached, the last one, here it is tower-diagnostics-20250320-1804.zip
March 20, 20251 yr Author 39 minutes ago, Mainfrezzer said: Its as fast as the slowest disk is, if my memory serves me well, the 1 TB you use should absolutely fall in that speed that you observe, especially once you reach the end of the platter. If im not mistaken now, you should be about to reach the 1TB threshold, once that it is cleared, it should get faster since that drive does not contain any data to be tracked (if nothing is using the array at the same time) you were right, once past 1TB the speed has increased considerably : Quote Total size:4 TB Elapsed time:2 days, 4 hours, 12 minutes Current position:1.50 TB (37.4 %) Estimated speed:136.5 MB/sec Estimated finish:5 hours, 6 minutes so the problem is as you say the 1TB disk has a slow speed compared to the 4TB... well I'll look for a new replacement :-/ thank you all very much for your support, you guys are awesome.
March 20, 20251 yr Community Expert Typically, older smaller disks will not perform as well as newer larger disks simply due to data density. RPM may be the same, but more data per revolution so faster data. Also, longer outer tracks have more data than shorter inner tracks simply because of the length. Data density is the same so longer track has more data at the same RPM. And parity checks start with the outer and progress to the inner, so will start out fast and slow down. Since it is done now with disk2 and its slower inner tracks, it is going faster. When first writing to a disk it will usually start with the outer tracks also for just this reason.
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.