Hello, my ryzen server is constantly crashing I have replaced every hard and cache drive. Thanks to SpaceInvaderOne I have added the script to my startup and that stopped for some time. It looks like it is a cache problem and I have replaced the drives, the ports and the cables. So I decide to do a fresh install on the usb and started from scratch. Party and Data Drive where fine no errors. Added the cache drives no error. Suddenly I start having errors on the cache drives and I replaced them too and still an issue. I don't know but my server keeps locking up. I really don't have much experience and anyone tell me why my system keep crashing. Thank you for all your help. 

zeus-diagnostics-20191122-2108.zip

I’ve got the same MSI B450 tomahawk motherboard albeit with a 2400g instead of a 2700x like you.  I’ve been very stable on 1A0 firmware....that 1C0 you have is quite new.   Might be worth trying the lower version unless you’ve already been down that road?  Just a shot in the dark as you exhaust other options

Edited November 23, 20196 yr by danull

Diagnostics are just after a reboot so not much to see, you can try mirroring the syslog to flash, then post it after a crash: Settings -> Syslog Server.

 

Also, and though there's some debate if this is still needed for 2nd gen Ryzen, check your bios for "Power Supply Idle Control" (or similar) and set it to "typical current idle" (or similar), and make sure the RAM is not overclocked, it's known to cause issues with some Ryzen servers.

 

12 hours ago, danull said:

I’ve got the same MSI B450 tomahawk motherboard albeit with a 2400g instead of a 2700x like you.  I’ve been very stable on 1A0 firmware....that 1C0 you have is quite new.   Might be worth trying the lower version unless you’ve already been down that road?  Just a shot in the dark as you exhaust other options

Thank you for your help. Let you know of any changes

6 hours ago, johnnie.black said:

Diagnostics are just after a reboot so not much to see, you can try mirroring the syslog to flash, then post it after a crash: Settings -> Syslog Server.

 

Also, and though there's some debate if this is still needed for 2nd gen Ryzen, check your bios for "Power Supply Idle Control" (or similar) and set it to "typical current idle" (or similar), and make sure the RAM is not overclocked, it's known to cause issues with some Ryzen servers.

 

First of thank you. Can you tell what speed setting to use. I have 64 gb Ram installed. Please see attached the picture of the box. In addition I have overclocked the processor at the max speed. I will run server for a bit so I can collect more data in the syslog as I took your recommendation.

zeus-diagnostics-20191123-1435.zip

Edited November 23, 20196 yr by Howard Callender
missing attachments

After the parity rebuilt now i have 9 errors. I have rebuilt it 5 times already and errors keep coming back after a system hangs. i have tried everything to get this of the ground it has been over a month. All help is welcome. Thank you!

41 minutes ago, Howard Callender said:

After the parity rebuilt now i have 9 errors. I have rebuilt it 5 times already and errors keep coming back after a system hangs. i have tried everything to get this of the ground it has been over a month. All help is welcome. Thank you!

I liked johnnie’s comments about overclocking, although he only mentioned RAM.   Would undo any of that you did for either RAM or CPU before trying BIOS downgrades.  If you don’t remember what previous settings in BIOS were may be worth a reset of that config and starting from scratch.   I am not OCing anything in my stable Ryzen config, just using the default for that.

Edited November 23, 20196 yr by danull

Should be noted ( *IF* I've got the model number of your RAM right ) that it is listed as being incompatible with your board in a 4 dimm configuration on MSI's QVL

 

(But there are other 3000MHz 16GB dimms that MSI does list as being compatible in 4 DIMM configs)

 

Oh, and a side note for the diagnostic pros out there.  system/meminfo.txt does include the actual info on the sticks, including part numbers when running 6.8+

36 minutes ago, Squid said:

Should be noted ( *IF* I've got the model number of your RAM right ) that it is listed as being incompatible with your board in a 4 dimm configuration on MSI's QVL

 

(But there are other 3000MHz 16GB dimms that MSI does list as being compatible in 4 DIMM configs)

 

I didn't think a lack of checkmark necessarily meant incompatible, I thought a lack of checkmark means they simply hadn't tested it...i.e. they may have only had 2 DIMMs available for testing.   I agree it is probably worth going through MEMTEST and/or removing two and retrying as long as he's trying to get to the bottom of this.

17 hours ago, Howard Callender said:

First of thank you. Can you tell what speed setting to use. I have 64 gb Ram installed. Please see attached the picture of the box. In addition I have overclocked the processor at the max speed. I will run server for a bit so I can collect more data in the syslog as I took your recommendation.

With 4 DIMMs they should be set 1866 or 2133Mhz depending if they are single or dual rank, doesn't mean it can't be stable at higher speeds, but you should start at non overclocked speeds (same for CPU) and only if stable try overclocking, though IMHO file servers are not for overclocking.