From deb26e63e9c95fa966d9282b10cfa22d47428a5f Mon Sep 17 00:00:00 2001
From: Huon Wilson <huon@exoflare.io>
Date: Fri, 5 Apr 2024 07:53:44 +1100
Subject: [PATCH] Add workflow to clear persistent cache dirs on self-hosted
 runners

---
 .../clear_self_hosted_persistent_caches.yaml  | 67 +++++++++++++++++++
 .../generate_github_workflows.py              | 44 ++++++++++++
 2 files changed, 111 insertions(+)
 create mode 100644 .github/workflows/clear_self_hosted_persistent_caches.yaml

diff --git a/.github/workflows/clear_self_hosted_persistent_caches.yaml b/.github/workflows/clear_self_hosted_persistent_caches.yaml
new file mode 100644
index 00000000000..32f3040ede1
--- /dev/null
+++ b/.github/workflows/clear_self_hosted_persistent_caches.yaml
@@ -0,0 +1,67 @@
+# GENERATED, DO NOT EDIT!
+# To change, edit `src/python/pants_release/generate_github_workflows.py` and run:
+#   ./pants run src/python/pants_release/generate_github_workflows.py
+
+
+jobs:
+- name: clear_macos10_15_x86_64
+  runs-on:
+  - self-hosted
+  - macOS-10.15-X64
+  steps:
+  - name: df before
+    run: df -h
+  - name: Deleting ~/Library/Caches
+    run: du -sh ~/Library/Caches || true; rm -rf ~/Library/Caches
+  - name: Deleting ~/.cache
+    run: du -sh ~/.cache || true; rm -rf ~/.cache
+  - name: Deleting ~/.nce
+    run: du -sh ~/.nce || true; rm -rf ~/.nce
+  - name: Deleting ~/.rustup
+    run: du -sh ~/.rustup || true; rm -rf ~/.rustup
+  - name: Deleting ~/.pex
+    run: du -sh ~/.pex || true; rm -rf ~/.pex
+  - name: df after
+    run: df -h
+- name: clear_linux_arm64
+  runs-on:
+  - self-hosted
+  - Linux
+  - ARM64
+  steps:
+  - name: df before
+    run: df -h
+  - name: Deleting ~/Library/Caches
+    run: du -sh ~/Library/Caches || true; rm -rf ~/Library/Caches
+  - name: Deleting ~/.cache
+    run: du -sh ~/.cache || true; rm -rf ~/.cache
+  - name: Deleting ~/.nce
+    run: du -sh ~/.nce || true; rm -rf ~/.nce
+  - name: Deleting ~/.rustup
+    run: du -sh ~/.rustup || true; rm -rf ~/.rustup
+  - name: Deleting ~/.pex
+    run: du -sh ~/.pex || true; rm -rf ~/.pex
+  - name: df after
+    run: df -h
+- name: clear_macos11_arm64
+  runs-on:
+  - self-hosted
+  - macOS-11-ARM64
+  steps:
+  - name: df before
+    run: df -h
+  - name: Deleting ~/Library/Caches
+    run: du -sh ~/Library/Caches || true; rm -rf ~/Library/Caches
+  - name: Deleting ~/.cache
+    run: du -sh ~/.cache || true; rm -rf ~/.cache
+  - name: Deleting ~/.nce
+    run: du -sh ~/.nce || true; rm -rf ~/.nce
+  - name: Deleting ~/.rustup
+    run: du -sh ~/.rustup || true; rm -rf ~/.rustup
+  - name: Deleting ~/.pex
+    run: du -sh ~/.pex || true; rm -rf ~/.pex
+  - name: df after
+    run: df -h
+name: Clear persistent caches on long-lived self-hosted runners
+'on':
+  workflow_dispatch: {}
diff --git a/src/python/pants_release/generate_github_workflows.py b/src/python/pants_release/generate_github_workflows.py
index 83c32f0aa66..2c0511536d3 100644
--- a/src/python/pants_release/generate_github_workflows.py
+++ b/src/python/pants_release/generate_github_workflows.py
@@ -1570,6 +1570,38 @@ def gen_goals(use_default_version: bool) -> Sequence[object]:
     return PublicReposOutput(jobs=jobs, inputs=inputs, run_name=run_name)
 
 
+def clear_self_hosted_persistent_caches_jobs() -> list[dict[str, Any]]:
+    def make_job(platform: Platform) -> dict[str, Any]:
+        helper = Helper(platform)
+
+        clear_steps = [
+            {
+                "name": f"Deleting {directory}",
+                "run": f"du -sh {directory} || true; rm -rf {directory}",
+            }
+            for directory in [
+                # not all of these will necessarily exist (e.g. ~/Library/Caches is macOS-specific),
+                # but the script is resilient to this
+                "~/Library/Caches",
+                "~/.cache",
+                "~/.nce",
+                "~/.rustup",
+                "~/.pex",
+            ]
+        ]
+        return {
+            "name": helper.job_name("clear"),
+            "runs-on": helper.runs_on(),
+            "steps": [
+                {"name": "df before", "run": "df -h"},
+                *clear_steps,
+                {"name": "df after", "run": "df -h"},
+            ],
+        }
+
+    return [make_job(platform) for platform in SELF_HOSTED]
+
+
 # ----------------------------------------------------------------------
 # Main file
 # ----------------------------------------------------------------------
@@ -1740,12 +1772,24 @@ def generate() -> dict[Path, str]:
         Dumper=NoAliasDumper,
     )
 
+    clear_self_hosted_persistent_caches = clear_self_hosted_persistent_caches_jobs()
+    clear_self_hosted_persistent_caches_yaml = yaml.dump(
+        {
+            "name": "Clear persistent caches on long-lived self-hosted runners",
+            "on": {"workflow_dispatch": {}},
+            "jobs": clear_self_hosted_persistent_caches,
+        }
+    )
+
     return {
         Path(".github/workflows/audit.yaml"): f"{HEADER}\n\n{audit_yaml}",
         Path(".github/workflows/cache_comparison.yaml"): f"{HEADER}\n\n{cache_comparison_yaml}",
         Path(".github/workflows/test.yaml"): f"{HEADER}\n\n{test_yaml}",
         Path(".github/workflows/release.yaml"): f"{HEADER}\n\n{release_yaml}",
         Path(".github/workflows/public_repos.yaml"): f"{HEADER}\n\n{public_repos_yaml}",
+        Path(
+            ".github/workflows/clear_self_hosted_persistent_caches.yaml"
+        ): f"{HEADER}\n\n{clear_self_hosted_persistent_caches_yaml}",
     }