On 10/30/2017 at 7:48 PM, Reckless Maker said:

I just started with unRAID and for some reason my system won't stay up for more than an hour. This is what posts to the screen:

And this is what posts in safe mode:

This is on a Ryzen 1600X with nothing attached but the ram and an AMD RX 550 graphics card.

tower-diagnostics-20171030-1937.zip

 

We are following very closely a linux kernel bug report re: Ryzen Freezes:

https://bugzilla.kernel.org/show_bug.cgi?id=196683

 

I didn't notice this until today but the kernel config RCU_NOCB_CPU_ALL was removed in kernel 4.13.  It was one of the important options I found to prevent Ryzen cpus from locking up.

 

I'm looking at patching support back in to the kernel but in the meantime, if you have Ryzen, you can add this to your Syslinux configuration's 'append' lines,

 

For 16-threads Ryzen (1800X, 1700X, 1700):

rcu_nocbs=0-15

 

For 12-threads Ryzen (1600X, 1600):

rcu_nocbs=0-11

 

For 8-threads Ryzen (1500X, 1500, 1400):

rcu_nocbs=0-7

 

For 4-threads Ryzen (1300X, 1200):

rcu_nocbs=0-3

 

Edited November 1, 20178 yr by limetech
Add link to kernel bug report

Interesting, thanks for the possible workaround !

 

FYI I've been running RC10b since it came out, no crashes so far. In the first 24 hours I ran the stock kernel. I then rebooted and added the DVB update once it was released. Been running since.

 

Uptime of 3d 10hours. 

Very interesting! I never would have figured that out in a million years. Just out of curiosity, why does Ryzen have trouble with RCU callbacks? Also what are RCU callbacks? Asking for a friend who doesn't know the first thing about Linux kernel.

Edited October 31, 20178 yr by Reckless Maker

19 hours ago, eschultz said:

I'm looking at patching support back in to the kernel but in the meantime, if you have Ryzen, you can add this to your Syslinux configuration's 'append' lines,

 

For 12-threads Ryzen (1600X, 1600):

rcu_nocbs=0-11

 

I'm not sure how to do this. I searched for syslinux config and found several files named like that in the /boot directory, should I edit one of those?

30 minutes ago, RonUSMC said:

 

I'm not sure how to do this. I searched for syslinux config and found several files named like that in the /boot directory, should I edit one of those?

 

On the main page click on your flash drive and then scroll down to Syslinux Configuration

4 hours ago, johnnie.black said:

 

On the main page click on your flash drive and then scroll down to Syslinux Configuration

 

thanks Johnnie.

 

I see this, should I just add a line at the end "APPEND rcu_nocbs=0-11"

 

 

default menu.c32
menu title Lime Technology, Inc.
prompt 0
timeout 50
label unRAID OS
  menu default
  kernel /bzimage
  append initrd=/bzroot
label unRAID OS GUI Mode
  kernel /bzimage
  append initrd=/bzroot,/bzroot-gui
label unRAID OS Safe Mode (no plugins, no GUI)
  kernel /bzimage
  append initrd=/bzroot unraidsafemode
label unRAID OS GUI Safe Mode (no plugins)
  kernel /bzimage
  append initrd=/bzroot,/bzroot-gui unraidsafemode
label Memtest86+
  kernel /memtest

or should it go here?

 

label unRAID OS
  menu default
  kernel /bzimage
  append initrd=/bzroot rcu_nocbs=0-11

like that?

Edited November 1, 20178 yr by RonUSMC

9 minutes ago, RonUSMC said:

label unRAID OS
  menu default
  kernel /bzimage
  append initrd=/bzroot rcu_nocbs=0-11

like that?

 

This should be OK but since I'm not sure it can go in the end, do it like so:

 

label unRAID OS
  menu default
  kernel /bzimage
  append rcu_nocbs=0-11 initrd=/bzroot 

Also note that if you plan to use other boot mode, e.g. GUI mode, you'll need to add it there also.

 

Thanks. Just put together my new Ryzen storage rig and just managed to request a trial before the cursor stopped blinking and the system froze. Will try this fix.

EDIT: This solved it. Thanks a lot!

Edited November 19, 20178 yr by Tech

Is this fix still needed for Ryzen systems? I had this in my syslinux at one point but just looked and it's not longer there. Not sure if the upgrade wiped it out but I haven't had any freezes so I'm not sure if I need to add it back? 

2 hours ago, ffhelllskjdje said:

Is this fix still needed for Ryzen systems? I had this in my syslinux at one point but just looked and it's not longer there. Not sure if the upgrade wiped it out but I haven't had any freezes so I'm not sure if I need to add it back? 

Yes, but make sure you're using it correctly:

 

 

On 12.2.2018 at 11:54 PM, johnnie.black said:

Yes, but make sure you're using it correctly:

 

 

So you mean in 6.4.1 its still needed generally or only for a problem to solve?

Because i dont really get, what it does and which problem it tries to fix, could you please explain more in detail?

i dont have any "rcu_" message in my syslog, dmesg or messages logs.

Im using a Ryzen 5 1600x with 24G on a ASUS PRIME X370 Pro

Edited March 12, 20188 yr by jammsen

6 hours ago, jammsen said:

So you mean in 6.4.1 its still needed generally or only for a problem to solve?

Because i dont really get, what it does and which problem it tries to fix, could you please explain more in detail?

i dont have any "rcu_" message in my syslog, dmesg or messages logs.

Im using a Ryzen 5 1600x with 24G on a ASUS PRIME X370 Pro

You need to use if your Ryzen based system crashes.

6 hours ago, johnnie.black said:

You need to use if your Ryzen based system crashes.

I only have stutters up to hangs for seconds when im writing data to a non ssd disk of the array. Doesnt really count, does it?

Just now, jammsen said:

I only have stutters up to hangs for seconds when im writing data to a non ssd disk of the array. Doesnt really count, does it?

No, it won't help with that.

Just now, johnnie.black said:

No, it won't help with that.

Any idea, if thats related to old mechanical disks?

9 minutes ago, jammsen said:

Any idea, if thats related to old mechanical disks?

Start a thread on the General support forum and post the diagnostics grabbed after that happens.

On 12.3.2018 at 5:40 PM, johnnie.black said:

Start a thread on the General support forum and post the diagnostics grabbed after that happens.

I will do that, when it happens again, thanks.