With the upstream addition of HERM (Hardware Events Report Method ) patches1 at the EDAC Kernel drivers, edac-utils became unsupported. A new toolset rasdaemon is since then available to replace it and allow collecting information about memory errors in userspace. It is bundled with most distros, otherwise available from source here https://github.com/mchehab/rasdaemon .
As most of these tools require contribution from the community to populate their database, they are not always up to date with less common hardware. It is however easy to add new database entries for each device by adding a file to the /etc/ras/dimm_labels.d/ folder.
A new entry has the following structure
Vendor: [System vendor name]
Model: [System model name]
[label1]: [Memory controller]:[Channel]:[Slot];
[label2]: [Memory controller]:[Channel]:[Slot];- Vendor and model can be accessed through dmidecode or sysfs under /sys/devices/virtual/dmi/id/board_vendor and /sys/devices/virtual/dmi/id/board_name.
- Labels can be put on a single line, one per line, broken down as needed.
My VEP4600 has the following topology
| Motherboard label | Memory controller | Channel | Slot |
|---|---|---|---|
| DIMM0 | MC0 | 0 | 0 |
| DIMM1 | MC0 | 1 | 0 |
| DIMM2 | MC1 | 0 | 0 |
| DIMM3 | MC1 | 1 | 0 |
As the CPU supports quad-channel memory, it is better to populate all slots. Memory is limited to 2133MT/s however, but with the current ram price surge, this could not be a problem since slow DDR4 RDIMMs are still a bit cheaper than their faster counterparts
Under /etc/ras/dimm_labels.d/ create file edge3400.txt
Vendor: DELL
Model: EDGE3400
DIMM0: 0.0.0; DIMM2: 0.1.0;
DIMM1: 1.0.0; DIMM3: 1.1.0;Alternatively, create file vep4600.txt
Vendor: DELL
Model: VEP4600
DIMM0: 0.0.0; DIMM2: 0.1.0;
DIMM1: 1.0.0; DIMM3: 1.1.0;
- Log in to post comments