Drives don't connect when plugging in the T4 cables

@henry-in-florida 

I will try to set up your system as best I can. I cannot match everything.

Did you change your AJA settings and get the same result of a slowdown? I can run AJA for days and never see a slowdown, for a counter example.

Also, did you ever run Samsung Magician? there is a Mac version now. Older drives like your could have a firmware update.

Speed (Throughput) Issue:

My AJA test of the 2.5TB SATA RAID, to your standard...

Writing: Starts @210MB/sec (after initially showing 400-ish for very short period), after 60 frames goes down significantly from there. Write speed decreases almost linearly. I;m writing to the root level of the drive. Last speed as shown - on first pass! With the jumper (daisy-chain) in place. Note the error readings! At these sizes the pass takes a while to complete of course. While test ran I looked at Activity Monitor (AM) and grabbed a screenshot there as well. The writing data rate is varying in AM at @43-108MB/s! All six drive lights are lit up. 

Reading: Stays pretty consistently 300-400MBs. AVG = 360MBs. 

May I see what you came up with? Something badly out of whack, it seems to me. Will look into Samsung Magician, clunky though it may be,

@henry-in-florida 

My initial test was a M1 laptop, 2 Thunderbay Mini's with really old SSD's and a OWC Thunderbolt 3 dock. It should get close enough to your setup for testing. I will modify it later.

No dropouts, first.

AJA is getting far better numbers, even with the 10 year old drives. 1200MB/s writes/2200MB/s reads on RAID 4. Writes will sometimes drop to 1,000MB/s, but that is all.

So this points clearly to your issue being a hardware problem somewhere.

First, connect only the two enclosures, no dock, one to each side, then run AJA. Are you getting anywhere near my numbers? Does performance stay steady?
if not, then l would make 2 RAID 0 volumes, if you can, one in each enclosure, and use AJA to make sure you are sustaining very good numbers (up to 2800M/s with those drives, which are fast). One might be faulty.

If you are getting good numbers, daisy chain them, test. then Connect through the dock, and test again. Where do you see the performance drop? That may be the fastest way to isolate where your issue is arising from.

Split cables, but not the way you last described, may try that next. The way you describe, I will have to remove my data to reconfigure the RAID into 2x RAID 0, one in each frame. There's about 1.2TB to move out (I may have a backup already, but I'm gonna have to move all back in. That's gonna take awhile to get to a usable system...

Meanwhile note that the read speed came way up. I did have to completely restart the system to get any read speed after reconfiguring.

@henry-in-florida 

Still, for an 8 drive RAID 4, not good.

You can at least do the split Thunderbolt connections without reconfiguring. Just remove the dock temporarily, connect one to the left, one to the right and run AJA, see what you get.

@henry-in-florida 

Same setup, daisy chained enclosures off the dock, 2200MB/s writes, 2400 MB/s reads on modern SSD drives. So you have a reference.

Posted by: @softraid-support

↑

@henry-in-florida 

Same setup, daisy chained enclosures off the dock, 2200MB/s writes, 2400 MB/s reads on modern SSD drives. So you have a reference.

@softRAID Support

Near as I can come RAID 0 drives reformatted using single Volume on EACH enclosure, 2 DRIVES 1TB. Split cabling. Submitting Report and AJA pics. Wrtite Speed kaput, x2! I left the existing erased drives in the enclosure but set up 2 in each frame as you indicated. So all drives dead? Both enclosures dead. Don't make me compare with Apple formatted drives. 

So I am now out of one Array and nothing pointing at a specific HDW problem. . 

Same results on each!

Tried Test #2 same conditions, just a complete restart, unplug cables at the computer end, replug, start up and retest. Results were better on Enclosure 2 than Enclosure 1, but both much better. the previously. So think the differences are from drive differences, tried to eliminate the daisy chain issue and placed the daisy-chain cable in place of the independent cables. Results appear the same. So I conclude the issue is drive related rather than enclosure related. Also tried a different pair of drives in Enclosure 1. Lo! The issues there are much worse. So, very likely a drive issue! Tried to Certify in SoftRAID app to see if they would certify. Initialize before building a volume made no difference. So conclude that the drives may be aging out or wearing out. Thoughts? Is Samsung's remapping worthwhile to try? Magician?

Forgot to include Screenshots from later test.I am verifying all the drives and finding improvement, but nowhere near your 2000MBs benchmark. 

@henry-in-florida 

I do not think the drives are "worn out". With TRIM enabled (it is), the drives won't get slowdowns on writes. Writing 6TB on 2TB drives is far from excessive. So those drives should be performing to their specs. The other drives have 70,000 hours, but still have 98%, so should be OK also.

Note: Apple's RAID driver does not support TRIM, so it is not good to use for RAID on flash media in any case. And there is virtually no performance difference between Apple's and SoftRAID's driver, as they both function the same.

I would next look to see if there is a firmware update for these drives. That could be it. At least we are narrowing this down. IMO, you should be getting at least close to the numbers I was seeing. also, there was no slowdown in AJA, even after hours of continuing running (at the settings I provided)

What is the exact model of the two enclosures you have? That is the last variable, I am not sure I have the exact ones, but might. I really doubt it is the enclosures being slow, unless there is an issue with these specific drives.

@henry-in-florida 
Note: A quick check with ChatGPT (its useful for some things) I don't think this is your issue, but it did have some issues on the 840 series drive. Your writes are awful, reads poor, this note only covers 50% of your issue. The 870 is stable on the Mac and 850 should be fine also. Only the 840 showed issues.
The 870's in the 6TB RAID 4 volume should show optimal performance, they are new-ish and should be in prime condition. the fact that volume is not showing great AJA numbers is also concerning. This is why your system is so hard to pin down, your throughput numbers don't make much sense.

Samsung 840 EVO (your drive, EXT0BB6Q)

• Problematic history:

  • The 840 EVO was notorious for a “read speed slowdown over time” bug.

  • Old, rarely accessed files would slow to ~50 MB/s due to charge leakage in TLC NAND.

  • Samsung issued multiple firmware patches (yours, EXT0BB6Q, is the last). It fixed most of the slowdown, but reports persisted of partial regression months later.

• On macOS:

  • Same issue applied — macOS didn’t escape the bug.

  • With EXT0BB6Q, you should be okay for casual use, but read speeds can still dip compared to newer EVOs.

  • Reliability-wise: safe, but slower with age.

Posted by: @softraid-support

↑

@henry-in-florida 
Note: A quick check with ChatGPT (its useful for some things) I don't think this is your issue, but it did have some issues on the 840 series drive. Your writes are awful, reads poor, this note only covers 50% of your issue. The 870 is stable on the Mac and 850 should be fine also. Only the 840 showed issues.
The 870's in the 6TB RAID 4 volume should show optimal performance, they are new-ish and should be in prime condition. the fact that volume is not showing great AJA numbers is also concerning. This is why your system is so hard to pin down, your throughput numbers don't make much sense.

Right, not my issue. 

Checking the two drives with slow-downs tested in pair with those that work and in combination with the with slow-downs in a single RAID0 combo:

• 840EVO
• 850EVO ENCL1 slot 3, disk 13 in my system. 

Snapshots from SoftRAID:

They both slow down by themselves, in any combo with other drives and are unaffected by split cabling. Regardless of the history THESE SLOW DOWN drastically. What helped bring up the speed? Using your Verify function on the other drives seems to speed things up for me. Also found single drive (850EVO) that has serious slowdowns. 

Posted by: @softraid-support

↑

@henry-in-florida 

I do not think the drives are "worn out". With TRIM enabled (it is), the drives won't get slowdowns on writes. Writing 6TB on 2TB drives is far from excessive. So those drives should be performing to their specs. The other drives have 70,000 hours, but still have 98%, so should be OK also.

Note: Apple's RAID driver does not support TRIM, so it is not good to use for RAID on flash media in any case. And there is virtually no performance difference between Apple's and SoftRAID's driver, as they both function the same.

I would next look to see if there is a firmware update for these drives. That could be it. At least we are narrowing this down. IMO, you should be getting at least close to the numbers I was seeing. also, there was no slowdown in AJA, even after hours of continuing running (at the settings I provided)

What is the exact model of the two enclosures you have? That is the last variable, I am not sure I have the exact ones, but might. I really doubt it is the enclosures being slow, unless there is an issue with these specific drives.

Samsung with Mac is a bummer! Magician doesn't run well update options don't work on MacOS 15.7 Sequoia. I did find drivers but no way to install except by Terminal maybe, do you know how? A step-by-step would be appreciated (I found tctcl command, but know nothing about it). Latest version drivers here:

@henry-in-florida 

Only a "certify disk" can try to refresh the media. verify does not do anything for flash media, it requires a write/ then a 0 write to clear the cells. A "zero disk" only marks the "directory" on the disk that it is empty, nothing is actually erased. 
I have never used the Magician software, some users have since they made a Mac version. I would have to find a samsung drive and try it to know how it works.

Did you determine however, that only using the other drives give better performance?  (no slowdowns). But you also discovered that no combination gave you the numbers I get with AJA?

So having done what we could do to replicate the problem and eliminate causes of the enclosure(s) and cables in the slowdown issue and found related causes of poorly responding drives and/or FW updates that can't be installed on Samsung easily, I would like to propose a solution based on your testing and comparison with your equipment there.

The reason? I NEED MY RAID STORAGE TO FUNCTION. 

Address the Hardware issue:

1. Put this solution on the "system" going through the Hub and retest the existing Test RAID 0. If all goes well, and the speed is still the same as stand-alone, proceed with next steps. 
2. Replace one or two drives with non-Samsung devices @500GB.
3. Retest with a RAID 4 Array containing 6 DRIVES with the slowest (worst as the Parity). NB, I have already tried this with fewer drives (5) in a 2TB Array and sent you pictures.
4. While I agree that the overall Read/Write Speed is sub-par to your set up and I how no funds to completely replace all media at this time, I would accept the 1300-1500 stable performance the test system until there is a better solution. 
5. It will be a time consuming exercise to restore it (8+ H via NAS backup and CCC, I figure after recreating the config I had in place.