From cb14c932142a2e2dae814c86ec4810da32af9328 Mon Sep 17 00:00:00 2001
From: Grant Gainey <ggainey@redhat.com>
Date: Fri, 7 Jun 2024 12:44:35 -0400
Subject: [PATCH] Dropped repo_concurrency in favor of a setting.

Defaults to 5-workers.
---
 pulp_rpm/app/serializers/prune.py | 11 -----------
 pulp_rpm/app/settings.py          |  1 +
 pulp_rpm/app/tasks/prune.py       | 17 +++++++++++++++--
 pulp_rpm/app/viewsets/prune.py    |  1 -
 4 files changed, 16 insertions(+), 14 deletions(-)

diff --git a/pulp_rpm/app/serializers/prune.py b/pulp_rpm/app/serializers/prune.py
index 4b3e09f30..d59781a09 100644
--- a/pulp_rpm/app/serializers/prune.py
+++ b/pulp_rpm/app/serializers/prune.py
@@ -33,17 +33,6 @@ class PrunePackagesSerializer(serializers.Serializer, ValidateFieldsMixin):
         default=14,
     )
 
-    repo_concurrency = serializers.IntegerField(
-        help_text=(
-            "Number of concurrent workers to use to do the pruning. "
-            "If not set then the default value will be used."
-        ),
-        allow_null=True,
-        required=False,
-        min_value=1,
-        default=10,
-    )
-
     dry_run = serializers.BooleanField(
         help_text=_(
             "Determine what would-be-pruned and log the list of packages. "
diff --git a/pulp_rpm/app/settings.py b/pulp_rpm/app/settings.py
index 02186b359..79add69ca 100644
--- a/pulp_rpm/app/settings.py
+++ b/pulp_rpm/app/settings.py
@@ -16,3 +16,4 @@
 SOLVER_DEBUG_LOGS = True
 RPM_METADATA_USE_REPO_PACKAGE_TIME = False
 NOCACHE_LIST = ["repomd.xml", "repomd.xml.asc", "repomd.xml.key"]
+PRUNE_WORKERS_MAX = 5
diff --git a/pulp_rpm/app/tasks/prune.py b/pulp_rpm/app/tasks/prune.py
index 6d371e1da..606924bc6 100644
--- a/pulp_rpm/app/tasks/prune.py
+++ b/pulp_rpm/app/tasks/prune.py
@@ -1,6 +1,7 @@
 from datetime import datetime, timedelta
 from logging import getLogger, DEBUG
 
+from django.conf import settings
 from django.db.models import F, Subquery
 from django.utils import timezone
 
@@ -105,7 +106,6 @@ def prune_repo_packages(repo_pk, keep_days, dry_run):
 def prune_packages(
     repo_pks,
     keep_days=14,
-    repo_concurrency=10,
     dry_run=False,
 ):
     """
@@ -126,6 +126,19 @@ def prune_packages(
     repos_to_prune = RpmRepository.objects.filter(pk__in=repo_pks)
     task_group = TaskGroup.current()
 
+    # We want to be able to limit the number of available-workers that prune will consume,
+    # so that pulp can continue to work while doing an import. We accomplish this by creating
+    # a reserved-resource string for each repo-prune-task based on that repo's index in
+    # the dispatch loop, mod number-of-workers-to-consume.
+    #
+    # By default, prune will consume up to 5 workers.
+    #
+    # (This comment and code below based on
+    #   https://github.com/pulp/pulpcore/blob/main/pulpcore/app/tasks/importer.py#L503-L512
+    # When we have a generic-approach to throttling mass-task-spawning, both places should
+    # be refactored to take advantage thereof.
+    prune_workers = int(settings.get("PRUNE_WORKERS_MAX", 5))
+
     gpr = GroupProgressReport(
         message="Pruning old Packages",
         code="rpm.package.prune",
@@ -140,7 +153,7 @@ def prune_packages(
     # This will keep an "all repositories" prune from locking up all the workers
     # until all repositories are completed.
     for index, a_repo in enumerate(repos_to_prune):
-        worker_rsrc = f"rpm-prune-worker-{index % repo_concurrency}"
+        worker_rsrc = f"rpm-prune-worker-{index % prune_workers}"
         exclusive_resources = [worker_rsrc, a_repo]
 
         dispatch(
diff --git a/pulp_rpm/app/viewsets/prune.py b/pulp_rpm/app/viewsets/prune.py
index 4080b5738..1037871e1 100644
--- a/pulp_rpm/app/viewsets/prune.py
+++ b/pulp_rpm/app/viewsets/prune.py
@@ -66,7 +66,6 @@ def prune_packages(self, request):
             kwargs={
                 "repo_pks": repos_to_prune_pks,
                 "keep_days": serializer.validated_data["keep_days"],
-                "repo_concurrency": serializer.validated_data["repo_concurrency"],
                 "dry_run": serializer.validated_data["dry_run"],
             },
         )