doc: Update configs

Signed-off-by: Stéphane Graber <stgraber@stgraber.org>
Sponsored-by: https://webdock.io

doc/config_options.txt

@@ -2278,6 +2278,22 @@ The events can be any combination of `lifecycle`, `logging`, and `network-acl`.
 Possible values are `bzip2`, `gzip`, `lzma`, `xz`, or `none`.
 ```
 
+```{config:option} instances.lxcfs.per_instance server-miscellaneous
+:defaultdesc: "`false`"
+:scope: "global"
+:shortdesc: "Whether to run LXCFS on a per-instance basis"
+:type: "bool"
+LXCFS is used to provide overlays for common `/proc` and `/sys`
+files which reflect the resource limits applied to the container.
+
+It normally operates through a single filesystem mount on the host which is then shared by all containers.
+This is very efficient but comes with the downside that a crash of LXCFS will break all containers.
+
+With this option, it's now possible to run a LXCFS instance per
+container instead, using more system resources but reducing the impact
+of a crash.
+```
+
 ```{config:option} instances.nic.host_name server-miscellaneous
 :defaultdesc: "`random`"
 :scope: "global"

internal/server/metadata/configuration.json

@@ -2528,6 +2528,15 @@
 							"type": "string"
 						}
 					},
+					{
+						"instances.lxcfs.per_instance": {
+							"defaultdesc": "`false`",
+							"longdesc": "LXCFS is used to provide overlays for common `/proc` and `/sys`\nfiles which reflect the resource limits applied to the container.\n\nIt normally operates through a single filesystem mount on the host which is then shared by all containers.\nThis is very efficient but comes with the downside that a crash of LXCFS will break all containers.\n\nWith this option, it's now possible to run a LXCFS instance per\ncontainer instead, using more system resources but reducing the impact\nof a crash.",
+							"scope": "global",
+							"shortdesc": "Whether to run LXCFS on a per-instance basis",
+							"type": "bool"
+						}
+					},
 					{
 						"instances.nic.host_name": {
 							"defaultdesc": "`random`",