New ram, unraid won't boot

[beire]

Hey guys, 

 

I recentyl bought a second hand server with following hardware

Supermicro x7dbe

2x Intel xeon E5405

2Gb of ram

22 1tb hitachi drives

 

I actully bought it for the superchassis CSE-846 it came in and thought i could mabye temperarily salvage the hardware. The amount of ram thoug was problematic for VM's though.

So i went ahead and bought 32gb of KVR667D2D4F5/4G replacement RAM ( https://www.lambda-tek.eu/-~sh/B42676959 ).

 

After installation of this ram the server boots, but unraid halts druring boot, no errors seem to be given, it just stalls.

I did a memtest overnight of around 24h, no errors there.

 

This ram should be compatible no?

It still boots with the old 2gb ram. I'm planning on doing a bios update when i got the time (it's offsite).

 

Any suggestions on this case? Can i even recover logs of the booting process (flash drive?), i certainly can't find any.

 

Thx for any help,

Bart

[trurl]

Have you tried booting anything other than Unraid with that RAM installed?

[beire]

On 12/18/2018 at 1:24 AM, trurl said:

Have you tried booting anything other than Unraid with that RAM installed?

I tried booting ubuntu live. Same result.

 

I now updated bios which gave no other result.

Then i started fiddling with bios settings regarding ram memory, and YES is boots !! Don't know which setting exactly it was though.

 

Next problem is 2 of 3 new WD red 4tb drives fail to be recognised by the old areca 1280ml controller, i'll swap in a sas2lp to test if it's really the drives.

Parity check speeds with the areca is 7mb/s, so i'm not sure that is normal either.

[Frank1940]

Did you check to see if the RAM you purchased is on the Manufacturer's list of supported RAM.  Some MB are particular about RAM.  What size modules did you get?

 

You also might need to reset the BIOS to the Factory Defaults since you purchased this MB used.  

 

By the way, I don't believe that that sas2lp controller is still being recommended.  I believe it uses a Marvell controller which has given issues for some Unraid users.  The current recommendation is for LSI controllers. 

[beire]

5 hours ago, Frank1940 said:

Did you check to see if the RAM you purchased is on the Manufacturer's list of supported RAM.  Some MB are particular about RAM.  What size modules did you get?

 

You also might need to reset the BIOS to the Factory Defaults since you purchased this MB used.  

 

By the way, I don't believe that that sas2lp controller is still being recommended.  I believe it uses a Marvell controller which has given issues for some Unraid users.  The current recommendation is for LSI controllers. 

Frank,

I did check the supported RAM list, the KVR667D2D4F5 4GB modules are on the list, although it's a replacement module.

I did also reset the bios to default settings after the update.

The sas2lp is one i have laying around from my home server which i replaced with a m1015 a while back. It's just to test. Performance is way better with it then the areca though! (12mB/s vs 85mB/s).

 

The setting that allows the server to boot now is "Cache Extended Memory", if set to anything other than write through it won't boot.

However, my syslog is now spammed with following errors:

 

Dec 20 20:17:41 Tower kernel: EDAC MC0: 1 CE Read error on mc#0branch#1channel#0slot#1 (branch:1 channel:0 slot:1 page:0x0 offset:0x0 grain:8 syndrome:0x0 - Rank=1 Bank=5 RDWR=Read RAS=16208 CAS=2908, CE Err=0x2000 (Correctable Non-Mirrored Demand Data ECC)))
Dec 20 20:17:42 Tower kernel: EDAC MC0: 1 CE Read error on unknown memory (branch:1 channel:0 slot:3 page:0x0 offset:0x0 grain:8 syndrome:0x0 - Rank=1 Bank=1 RDWR=Read RAS=2484 CAS=4, CE Err=0x2000 (Correctable Non-Mirrored Demand Data ECC)))
Dec 20 20:17:43 Tower kernel: EDAC MC0: 1 CE Read error on unknown memory (branch:1 channel:0 slot:3 page:0x0 offset:0x0 grain:8 syndrome:0x0 - Rank=1 Bank=1 RDWR=Read RAS=2484 CAS=4, CE Err=0x2000 (Correctable Non-Mirrored Demand Data ECC)))
Dec 20 20:17:45 Tower kernel: EDAC MC0: 1 CE Read error on mc#0branch#1channel#0slot#1 (branch:1 channel:0 slot:1 page:0x0 offset:0x0 grain:8 syndrome:0x0 - Rank=1 Bank=5 RDWR=Read RAS=16208 CAS=2908, CE Err=0x2000 (Correctable Non-Mirrored Demand Data ECC)))
Dec 20 20:17:46 Tower kernel: EDAC MC0: 1 CE Read error on unknown memory (branch:1 channel:0 slot:3 page:0x0 offset:0x0 grain:8 syndrome:0x0 - Rank=1 Bank=1 RDWR=Read RAS=2484 CAS=4, CE Err=0x2000 (Correctable Non-Mirrored Demand Data ECC)))
Dec 20 20:17:47 Tower kernel: EDAC MC0: 1 CE Read error on unknown memory (branch:1 channel:0 slot:3 page:0x0 offset:0x0 grain:8 syndrome:0x0 - Rank=1 Bank=4 RDWR=Read RAS=5628 CAS=692, CE Err=0x2000 (Correctable Non-Mirrored Demand Data ECC)))
Dec 20 20:17:48 Tower kernel: EDAC MC0: 1 CE Read error on unknown memory (branch:1 channel:0 slot:3 page:0x0 offset:0x0 grain:8 syndrome:0x0 - Rank=1 Bank=0 RDWR=Read RAS=1330 CAS=4, CE Err=0x2000 (Correctable Non-Mirrored Demand Data ECC)))
Dec 20 20:17:49 Tower kernel: EDAC MC0: 1 CE Read error on unknown memory (branch:1 channel:0 slot:3 page:0x0 offset:0x0 grain:8 syndrome:0x0 - Rank=1 Bank=

So this is not very good i think... It does however function, i have VM's working now too. I'm not sure what the drawbacks of the error's are.

Edited December 20, 2018 by beire

[Frank1940]

I am not a true Guru in reading these types of problems but it sure looks to be that you have some memory issues!   I would be readin the MB Manual and see what physical slots on the MB are represented by these errors.  Then I would double check that the memory is fully seated in the MB and the locking tabs are in the locked state.  If those checks indicate that everything is OK, I would pull the modules for those positions and see if that clears the problem.  If it does, I would then relocate some memory from the other slots to those slots and see if the problem reappears.