-
Notifications
You must be signed in to change notification settings - Fork 143
Open
Description
I am trying to reconfigure a Optane DC PM from devdax to system-ram.
On the machine with problem, the command logs are:
root@server:~# ndctl disable-namespace all
disabled 4 namespaces
root@server:~# ndctl disable-region all
disabled 2 regions
root@server:~# ndctl zero-labels all
zeroed 2 nmems
root@server:~# ndctl enable-region all
enabled 2 regions
root@server:~# ndctl create-namespace --region region0 --mode=devdax
{
"dev":"namespace0.0",
"mode":"devdax",
"map":"dev",
"size":"124.03 GiB (133.18 GB)",
"uuid":"2b1a66fa-525b-4658-a216-5372b36036a6",
"daxregion":{
"id":0,
"size":"124.03 GiB (133.18 GB)",
"align":2097152,
"devices":[
{
"chardev":"dax0.0",
"size":"124.03 GiB (133.18 GB)",
"mode":"devdax"
}
]
},
"align":2097152
}
root@server:~# daxctl list --regions --devices
[
{
"path":"\/LNXSYSTM:00\/LNXSYBUS:00\/ACPI0012:00\/ndbus0\/region0\/dax0.0",
"id":0,
"size":133175443456,
"align":2097152,
"devices":[
{
"chardev":"dax0.0",
"size":133175443456,
"mode":"devdax"
}
]
}
]
root@server:~# daxctl reconfigure-device --mode=system-ram dax0.0
libdaxctl: daxctl_dev_enable: dax0.0: failed to enable
error reconfiguring devices: No such device
reconfigured 0 devices
And I have noticed that the dmesg shows:
[ 714.196176] kmem dax0.0: rejecting DAX region [??? 0x60fe200000-0x7fffffffff flags 0x0] with invalid node: -1
[ 714.196177] kmem: probe of dax0.0 failed with error -22
But on a normal machine, the same command outputs:
root@normalserver:~# ndctl disable-namespace all
disabled 2 namespaces
root@normalserver:~# ndctl disable-region all
disabled 1 region
root@normalserver:~# ndctl zero-labels all
zeroed 1 nmem
root@normalserver:~# ndctl enable-region all
enabled 1 region
root@normalserver:~# ndctl create-namespace --region region0 --mode=devdax
{
"dev":"namespace0.0",
"mode":"devdax",
"map":"dev",
"size":"30.88 GiB (33.16 GB)",
"uuid":"6cf2cb24-29c9-41b7-aae6-7aa0f96daeea",
"daxregion":{
"id":0,
"size":"30.88 GiB (33.16 GB)",
"align":2097152,
"devices":[
{
"chardev":"dax0.0",
"size":"30.88 GiB (33.16 GB)",
"target_node":1,
"mode":"devdax"
}
]
},
"align":2097152
}
root@normalserver:~# daxctl migrate-device-model
Error: failed to install /etc/modprobe.d/daxctl.conf: File exists
root@normalserver:~# daxctl list --regions --devices
[
{
"path":"\/LNXSYSTM:00\/LNXSYBUS:00\/ACPI0012:00\/ndbus0\/region0\/dax0.0",
"id":0,
"size":33160167424,
"align":2097152,
"devices":[
{
"chardev":"dax0.0",
"size":33160167424,
"target_node":1,
"mode":"devdax"
}
]
}
]
root@normalserver:~# daxctl reconfigure-device --mode=system-ram dax0.0
dax0.0:
WARNING: detected a race while onlining memory
Some memory may not be in the expected zone. It is
recommended to disable any other onlining mechanisms,
and retry. If onlining is to be left to other agents,
use the --no-online option to suppress this warning
dax0.0: all memory sections (247) already online
[
{
"chardev":"dax0.0",
"size":33160167424,
"target_node":1,
"mode":"system-ram",
"movable":false
}
]
reconfigured 1 device
The parameter "target_node":1, is also missing in the problem machine.
I am confused about this missing parameter. And I have no idea about how ndctl works with NUMA.
Metadata
Metadata
Assignees
Labels
No labels