Raid5 intermittent ...
 
Notifications
Clear all

Raid5 intermittent hang - despite workarounds on Catalina 15.4

24 Posts
3 Users
0 Likes
2,574 Views
(@fantafly)
Posts: 13
Member
Topic starter
 

Hello

I have followed the advice found in this forum (disable secure boot, set nvram to "dart=0") and have successfully avoided the dreaded finder crash since then.

However, when editing video (regardless of software used - Premiere, Avid or DaVinci), every 5 mins or so the drive will hang for up to 20 seconds.
This happens during playback, export or any other operation accessing the drive for a sustained period. Data throughput seems not to be at issue here, it happens with standard HD material well below 20MB/s as well.

It also happens regardless of write cache option enabled or disabled in Softraid.
(driver and software both version 5.8.3)

I'm considering enrolling as Mac delevoper to install the latest MacOs beta version, but considering the cost would like to know if this would fix the issue.

Any other thoughts / insights?

Many thanks

Using
Macbook Pro 13inch 2018 16GB
Catalina 15.4
AKiTiO Thunder3 Quad X
Seagate BarraCuda 3.5" 6TB x4

 
Posted : 20/05/2020 4:36 am
(@softraid-support)
Posts: 8049
Member Admin
 

Can you define "hang" better?
a beachball where everything is hanging? Or is it more like a pause in the video application?

If you immediately Finder copy a file to the same volume, does it hang until this is done?

If you keep Activity Monitor open, are any processes going "red", i.e, hanging, during this time?

I assume you still have mouse and other access?

I have not seen this reported before. It might be worth seeing if you get this in the public beta of 10.15.5, without the NVRAM setting.

 
Posted : 20/05/2020 11:26 am
(@fantafly)
Posts: 13
Member
Topic starter
 

Thanks for the quick reply.

Yes, I have access to all peripherals and the system is responsive.
But no access to the Raid5 volume: Just tried exactly what you suggested, the moment the render frame rate in DaVinci Resolve dropped to zero I pasted a 407MB file onto the Raid volume. It says zero KB copied exactly until the "hang" is over, at which point it speedily copies it across. (meant to attach screen shots but seems I can't in a reply?)

By the way, none of this happens on other drives attached to the same system.

Is the public beta available without enrolling in developer program?

 
Posted : 20/05/2020 11:51 am
(@softraid-support)
Posts: 8049
Member Admin
 

It requires a form, but not the developer program (i.e, it is free)

 
Posted : 20/05/2020 10:31 pm
(@fantafly)
Posts: 13
Member
Topic starter
 

I'm sad to report that despite upgrading to 10.15.5, the problem persists. (The macos beta still installs the old softraid driver btw, also updated that)
Is there any other information I could give you or diagnostics I could run?

 
Posted : 21/05/2020 12:32 am
(@softraid-support)
Posts: 8049
Member Admin
 

Send me the panic report when the system restarts. lets see if it is the same kernel panic.

 
Posted : 21/05/2020 1:53 am
(@fantafly)
Posts: 13
Member
Topic starter
 

Ok thanks.
But since the system doesn't crash, is there a panic report anyway?

For clarity I am attaching screen shots of data throughput as monitored by iStat Menus

One is while rendering out from DaVinci Resolve (source and destination both on the Raid5)
Another one is of just the Aja Speed test, continuously writing and reading a 4GB file

As you can see, in irregular intervals disk throughput drops to zero for around 10 seconds.
It is not related to bandwidth as clearly the raid is coping with the Aja test just fine, except for the usual hangs.

Any settings in SoftRaid that might affect this?

 
Posted : 21/05/2020 3:02 am
(@fantafly)
Posts: 13
Member
Topic starter
 

Sorry really can't figure out how to attach files to a reply so edited my previous post :-/

 
Posted : 21/05/2020 3:06 am
(@softraid-support)
Posts: 8049
Member Admin
 

What if you run Activity Monitor. Set it to CPU. Run Davinci and when you get the stall/pause, do you see any processes take over CPU time?

I assume this is over Thunderbolt?

There are no disks predicted to fail, also?

 
Posted : 21/05/2020 11:53 am
(@fantafly)
Posts: 13
Member
Topic starter
 

I appreciate your patience.
Yes, it's on Thunderbolt 3, no disks predicted to fail, no error reports whatsoever.

Did what you suggested, but no process comes up, DaVinci just goes idle until the drive comes back.
Same happens with Aja Disk test by the way, and also when I copy a large file (tried with 100 gig, got stuck twice)
Even tried it in safe mode, same thing!

In desperation I also tried hooking the Raid up to a 2014 MBPro running Mojave.
And, unfortunately the same thing happens when copying a large file.
Does all this mean it could be a hardware issue with the enclosure? Or the drives (4x Seagate Barracuda ST6000DM003)?

Did a few more tests trying the following settings in SoftRaid:
* optimized drive for digital video - no change
* disabled write cache - no change
* submit reads and writes one at a time - no change

No obvious error reports in the console either, although I'm not sure what I would be looking for.

 
Posted : 22/05/2020 5:29 am
(@fantafly)
Posts: 13
Member
Topic starter
 

Also, I've updated the Akitio firmware, to no avail.

 
Posted : 22/05/2020 6:32 am
(@softraid-support)
Posts: 8049
Member Admin
 

The fact this happens on another computer tells you this is not the computer, or OS. Two things eliminated.
The SoftRAID driver is just a pass-through, so it is not it. (it is just telling OS X where to put things, it does not actually perform the reaads or writes)

Do you have anything like istat menus installed on both? Applications that check SMART constantly can cause this kind of issue, iStat used to do that, I think that is fixed now.

If you have BBEdit (now free), try this:
run the kextstat command in terminal.
(no modifiers needed)
You will get a huge list of text, all extensions. Copy that text and paste it into a BEdit file.
In BBEdit, text menu, "process lines containing" and enter .apple in the search box. select "delete lines".
This should output only third party extensions.
Post those here. I do this all the time for users with problems, to see what is loaded and look for issues.

This does not seem like a enclosure issue, either, disks disappearing, yes, but 10 second dropouts, no.

Possibly the drives, but I am not aware of this symptom, unless something is constantly checking SMART.

 
Posted : 22/05/2020 2:31 pm
(@fantafly)
Posts: 13
Member
Topic starter
 

Hello

took a while to find some time to try your suggestions.

Since non-apple kernel extensions aren't loaded in safe mode, I booted into safe mode and tried copying a large file onto the raid, with the same dropout happening after roughly 5mins and again 3mins later (see screenshots).

Ran kextstat and purged all .apple lines, this is the result:

116 0 0xffffff7f81293000 0x3c000 0x3c000 com.softraid.driver.SoftRAID (5.8.3) 2EB72467-5496-3863-AF61-9F7EDAADADAE

Mystery...

 
Posted : 26/05/2020 2:51 am
(@softraid-support)
Posts: 8049
Member Admin
 

This is a temporary pause, not a hang with needing to force restart.

Did you try this with a clean install?
(You can create a volume inside your boot disk with Catalina - Run Disk Utility, select your boot disk, Partition, +, add volume, then boot into internet recovery and install Catalina)
See if this pause happens with a clean install.

This pause is not related to SoftRAID's driver, it is a pass-through driver that simply points OS X where to write the data.

Since you do not have any other extensions installed, it has to be something in Catalina doing this. A clean install may at least point in the right direction.

 
Posted : 26/05/2020 9:32 am
 Nubo
(@nubo)
Posts: 20
Member
 

Out of curiosity, in Energy Saver preferences, do you have selected "put hard drives to sleep when possible"?

 
Posted : 27/05/2020 11:04 pm
Page 1 / 2
Share:
close
open