Upgrade to Ubuntu 22.04, and custom 6.1.0-rc5 kernel

Change-Id: I7f49901478ed0085710e7df574d3b24132cd4b42

Author: benjamin-maynard

.gitignore

@@ -19,6 +19,9 @@
 *.pkrvars.json
 *.auto.pkrvars.json
 
+# Exclude Specific Packer Variables File
+!image/internal-build/image.pkrvars.hcl
+
 # Service Account
 service-account-key.json
 

deployment/README.md

@@ -5,7 +5,7 @@ This directory contains a [Terraform Module](https://www.terraform.io/docs/modul
 The `main` branch may be updated at any time with the latest changes which could be breaking. You should always configure your module to use a release. This can be configured in the modules Terraform Configuration block.
 
 ```
-source = "github.com/GoogleCloudPlatform/knfsd-cache-utils//deployment/terraform-module-knfsd?ref=v0.10.0"
+source = "github.com/GoogleCloudPlatform/knfsd-cache-utils//deployment/terraform-module-knfsd?ref=v1.0.0-beta1"
 ```
 
 ## Prerequisites
@@ -28,7 +28,7 @@ Basic usage of this module is as follows:
 ```terraform
 module "nfs_proxy" {
 
-    source = "github.com/GoogleCloudPlatform/knfsd-cache-utils//deployment/terraform-module-knfsd?ref=v0.10.0"
+    source = "github.com/GoogleCloudPlatform/knfsd-cache-utils//deployment/terraform-module-knfsd?ref=v1.0.0-beta1"
 
     # Google Cloud Project Configuration
     PROJECT                        = "my-gcp-project"
@@ -182,26 +182,6 @@ The service account will need the following project level IAM permissions:
 | LOCAL_SSDS                          | (Only used if `CACHEFILESD_DISK_TYPE` = `local-ssd`) The number of Local SSDs to assign to each cache instance. This can be either 0 to 8, 16, or 24 local SSDs for up to 9TB of capacity ([see here](https://cloud.google.com/compute/docs/disks/local-ssd#choosing_a_valid_number_of_local_ssds)). If you are setting this to 24 Local SSDs you should also change the `MACHINE_TYPE` variable to an instance with 32 CPU's, for example `n1-highmem-32`. | False    | `4`         |
 | CACHEFILESD_PERSISTENT_DISK_SIZE_GB | (Only used if `CACHEFILESD_DISK_TYPE` = `pd-standard`, `pd-balanced` or `pd-ssd`), what size should the persistent disk be in GB? Can be set between `100` and `64000`. For large volumes, consider larger instance types (see [here](https://cloud.google.com/compute/docs/disks/performance)).                                                                                                                                                            | False    | `1500`      |
 
-### Culling Options
-
-| Variable             | Description                                                                                                  | Required | Default                                                                                    |
-| -------------------- | ------------------------------------------------------------------------------------------------------------ | -------- | ------------------------------------------------------------------------------------------ |
-| CULLING              | Culling method to use (`default`, `custom`, or `none`).                                                      | False    | `custom`                                                                                   |
-| CULLING_LAST_ACCESS  | (custom only) Remove files from the cache that were last accessed over `CULLING_LAST_ACCESS` ago.            | False    | If using Local SSD: 1 hour for each `LOCAL_SSD`.<br><br>If using Persistent Disk, 6 hours. |
-| CULLING_THRESHOLD    | (custom only) Cull when the percentage of remaining disk space (or inodes) is less than `CULLING_THRESHOLD`. | False    | `20`                                                                                       |
-| CULLING_INTERVAL     | (custom only) How often to check if the remaining disk space is less than the `CULLING_THRESHOLD`            | False    | `1m`                                                                                       |
-| CULLING_QUIET_PERIOD | (custom only) After culling, how long to wait before resuming checks.                                        | False    | `CULLING_LAST_ACCESS / 4`                                                                  |
-
-The culling `CULLING` mode supports the following options:
-
-* `default` - This uses the standard cachefilesd to perform culling.
-* `custom` - This uses the custom knfsd-cull service to perform culling.
-* `none` - Disables culling completely
-
-The purpose of the custom culling agent is to workaround a known issue where cachefilesd may stop culling files in the cache. See [culling](../docs/culling.md) for more information on how to configure cachefilesd, and the known issue where cachefilesd may stop culling.
-
-The `none` option supports special cases where the cache rarely fills up.
-
 ### Mount Options
 
 These mount options are for the proxy to the source server.
@@ -232,51 +212,6 @@ These mount options are for the proxy to the source server.
 | NOHIDE         | When `true`, adds the `nohide` option to all the exports.                         | False    | `true`  |
 | EXPORT_OPTIONS | Any custom NFS exports options. These options will be applied to all NFS exports. | False    | `""`    |
 
-#### Custom culling options
-
-The durations, `CULLING_LAST_ACCESS`, `CULLING_INTERVAL`, and
-`CULLING_QUIET_PERIOD` support `h`, `m`, and `s` (hours, minutes, seconds)
-For example `5m`, `2.5h`, or `1h30m`.
-
-To avoid deleting files unnecessarily the culling process will wait until the
-remaining percentage of free space is less than `CULLING_THRESHOLD`. The
-remaining free space will be checked every `CULLING_INTERVAL`.
-
-Any file with a last access time older than `CULLING_LAST_ACCESS` will be
-deleted. Because files are deleted based on their last access, this might not
-remove enough files (or any files) to bring the free space above the threshold
-if most of the files in the cache have been used within the last access period.
-It can also remove more files than required (possibly even all the files).
-
-Once a culling attempt has been completed (even if no files were culled),
-culling will wait for `CULLING_QUIET_PERIOD` before resuming culling checks.
-This avoids repeatedly scanning the full file tree (costing IOPS) while most
-files are in use.
-
-**IMPORTANT:** `CULLING_THRESHOLD` *MUST* be greater than `bstop` and `fstop` in
-`/etc/cachefilesd.conf`. Otherwise cachefilesd will stop caching data before the
-custom culling threshold is reached so culling will never run.
-
-#### Example
-
-```terraform
-CULLING              = "custom"
-CULLING_LAST_ACCESS  = "4h"
-CULLING_THRESHOLD    = 20
-CULLING_INTERVAL     = "1m"
-CULLING_QUIET_PERIOD = "1h"
-```
-
-The culling agent will check every minute (`CULLING_INTERVAL`) to see if the
-remaining disk space (or inodes) is less than 20% (`CULLING_THRESHOLD`) of the
-total disk space.
-
-When the remaining disk space is below the 20% threshold the culling agent will
-then remove any files that were last accessed over four hours ago (`CULLING_LAST_ACCESS`)
-from the cache.
-
-The culling agent will then wait for at least one hour (`CULLING_QUIET_PERIOD`)
-before resuming culling checks.
 
 ### Autoscaling Configuration
 

deployment/fanout.md

@@ -37,7 +37,7 @@ There is no special logic in the Knfsd Terraform Module to handle the fanout arc
 ```terraform
 module "nfs_proxy_fanout" {
 
-  source = "github.com/GoogleCloudPlatform/knfsd-cache-utils//deployment/terraform-module-knfsd?ref=v0.10.0"
+  source = "github.com/GoogleCloudPlatform/knfsd-cache-utils//deployment/terraform-module-knfsd?ref=v1.0.0-beta1"
 
   # Google Cloud Project Configuration
   PROJECT = "my-gcp-project"
@@ -67,7 +67,7 @@ module "nfs_proxy_fanout" {
 
 module "nfs_proxy_cluster" {
 
-  source = "github.com/GoogleCloudPlatform/knfsd-cache-utils//deployment/terraform-module-knfsd?ref=v0.10.0"
+  source = "github.com/GoogleCloudPlatform/knfsd-cache-utils//deployment/terraform-module-knfsd?ref=v1.0.0-beta1"
 
   # Google Cloud Project Configuration
   PROJECT = "my-gcp-project"

deployment/metrics.md

@@ -59,7 +59,7 @@ Providing the metrics config from a file:
 
 ```terraform
 module "nfs_proxy" {
-  source = "github.com/GoogleCloudPlatform/knfsd-cache-utils//deployment/terraform-module-knfsd?ref=v0.10.0"
+  source = "github.com/GoogleCloudPlatform/knfsd-cache-utils//deployment/terraform-module-knfsd?ref=v1.0.0-beta1"
 
   METRICS_AGENT_CONFIG = file("metrics-config.yaml")
 }
@@ -69,7 +69,7 @@ Providing the metrics config inline using heredoc syntax:
 
 ```terraform
 module "nfs_proxy" {
-  source = "github.com/GoogleCloudPlatform/knfsd-cache-utils//deployment/terraform-module-knfsd?ref=v0.10.0"
+  source = "github.com/GoogleCloudPlatform/knfsd-cache-utils//deployment/terraform-module-knfsd?ref=v1.0.0-beta1"
 
   METRICS_AGENT_CONFIG = <<-EOT
     receivers:

deployment/terraform-module-knfsd/compute.tf

@@ -108,13 +108,6 @@ resource "google_compute_instance_template" "nfsproxy-template" {
     EXPORT_OPTIONS    = var.EXPORT_OPTIONS
     NFS_MOUNT_VERSION = var.NFS_MOUNT_VERSION
 
-    CULLING = var.CULLING
-
-    CULLING_LAST_ACCESS  = coalesce(var.CULLING_LAST_ACCESS, local.CULLING_LAST_ACCESS_DEFAULT)
-    CULLING_THRESHOLD    = var.CULLING_THRESHOLD
-    CULLING_INTERVAL     = var.CULLING_INTERVAL
-    CULLING_QUIET_PERIOD = var.CULLING_QUIET_PERIOD
-
     # system
     NFS_KERNEL_SERVER_CONF = file("${path.module}/resources/nfs-kernel-server.conf")
     NUM_NFS_THREADS        = var.NUM_NFS_THREADS

deployment/terraform-module-knfsd/resources/proxy-startup.sh

@@ -216,12 +216,6 @@ function init() {
 	DISABLED_NFS_VERSIONS=$(get_attribute DISABLED_NFS_VERSIONS)
 	READ_AHEAD_KB=$(get_attribute READ_AHEAD_KB)
 
-	CULLING="$(get_attribute CULLING)"
-	CULLING_LAST_ACCESS="$(get_attribute CULLING_LAST_ACCESS)"
-	CULLING_THRESHOLD="$(get_attribute CULLING_THRESHOLD)"
-	CULLING_INTERVAL="$(get_attribute CULLING_INTERVAL)"
-	CULLING_QUIET_PERIOD="$(get_attribute CULLING_QUIET_PERIOD)"
-
 	ENABLE_STACKDRIVER_METRICS=$(get_attribute ENABLE_STACKDRIVER_METRICS)
 	METRICS_AGENT_CONFIG=$(get_attribute METRICS_AGENT_CONFIG)
 	ENABLE_KNFSD_AGENT=$(get_attribute ENABLE_KNFSD_AGENT)
@@ -445,34 +439,6 @@ function configure-nfs() {
 
 }
 
-function configure-culling() (
-	function fmt() {
-		if [[ -n "$2" ]]; then
-			printf '%s %s\n' "$1" "$2"
-		fi
-	}
-
-	sed -i '/^nocull/d' /etc/cachefilesd.conf
-
-	if [[ "$CULLING" == "none" ]] || [[ "$CULLING" == "custom" ]]; then
-		echo "nocull" >>/etc/cachefilesd.conf
-	fi
-
-	if [[ "$CULLING" == "custom" ]]; then
-		: >/etc/knfsd-cull.conf
-		fmt last-access "$CULLING_LAST_ACCESS" >>/etc/knfsd-cull.conf
-		fmt threshold "$CULLING_THRESHOLD" >>/etc/knfsd-cull.conf
-		fmt interval "$CULLING_INTERVAL" >>/etc/knfsd-cull.conf
-		fmt quiet-period "$CULLING_QUIET_PERIOD" >>/etc/knfsd-cull.conf
-
-		echo "Starting Custom Culling Agent..."
-		start-services knfsd-cull
-		echo "Finished starting Custom Culling Agent."
-	else
-		echo "Custom Culling Agent disabled. Skipping..."
-	fi
-)
-
 function configure-metrics() {
 
 	# If needed, override the Monitoring API to use an IP address from private.googleapis.com
@@ -552,7 +518,6 @@ function main() {
 
 	configure-read-ahead
 	configure-nfs
-	configure-culling
 	configure-metrics
 
 	start-nfs

deployment/terraform-module-knfsd/resources/tests/configure-culling.bats

@@ -331,59 +331,6 @@ variable "NETAPP_ALLOW_COMMON_NAME" {
   default = false
 }
 
-variable "CULLING" {
-  type    = string
-  default = "default"
-
-  validation {
-    condition     = contains(["default", "custom", "none"], var.CULLING)
-    error_message = "CULLING must be one of [default, custom, none]."
-  }
-}
-
-variable "CULLING_LAST_ACCESS" {
-  type    = string
-  default = "" # calculate default based on LOCAL_SSDS
-
-  validation {
-    # This will also match the empty string, but in this case the empty string is allowed.
-    condition     = can(regex("^(\\d+h)?(\\d+m)?(\\d+s)?$", var.CULLING_LAST_ACCESS))
-    error_message = "CULLING_LAST_ACCESS must be a positive duration, e.g. '1h'. Valid time units are 'h' (hours), 'm' (minutes), 's' (seconds)."
-  }
-}
-
-variable "CULLING_THRESHOLD" {
-  type    = number
-  default = 20
-
-  validation {
-    condition     = var.CULLING_THRESHOLD == null || (var.CULLING_THRESHOLD >= 0 && var.CULLING_THRESHOLD <= 100)
-    error_message = "CULLING_THRESHOLD must between 0% and 100%."
-  }
-}
-
-variable "CULLING_INTERVAL" {
-  type    = string
-  default = "1m"
-
-  validation {
-    # This will also match the empty string, but in this case the empty string is allowed.
-    condition     = can(regex("^(\\d+h)?(\\d+m)?(\\d+s)?$", var.CULLING_INTERVAL))
-    error_message = "CULLING_INTERVAL must be a positive duration, e.g. '1h'. Valid time units are 'h' (hours), 'm' (minutes), 's' (seconds)."
-  }
-}
-
-variable "CULLING_QUIET_PERIOD" {
-  type    = string
-  default = ""
-
-  validation {
-    # This will also match the empty string, but in this case the empty string is allowed.
-    condition     = can(regex("^(\\d+h)?(\\d+m)?(\\d+s)?$", var.CULLING_QUIET_PERIOD))
-    error_message = "CULLING_QUIET_PERIOD must be a positive duration, e.g. '1h'. Valid time units are 'h' (hours), 'm' (minutes), 's' (seconds)."
-  }
-}
-
 variable "CACHEFILESD_DISK_TYPE" {
   type    = string
   default = "local-ssd"