Merge pull request #24631 from nvartolomei/nv/cst-cache-ubsan

cst/cache: ubsan fix, fix highly unlikely full cache trim if free disk info is delayed, more tests

src/v/cloud_storage/cache_service.cc

@@ -356,6 +356,10 @@ ss::future<> cache::trim(
         target_size *= _cache_size_low_watermark;
     }
 
+    if (!_free_space.has_value()) {
+        throw std::runtime_error("Free space information is not available.");
+    }
+
     // In the extreme case where even trimming to the low watermark wouldn't
     // free enough space to enable writing to the cache, go even further.
     if (_free_space < config::shard_local_cfg().storage_min_free_bytes()) {
@@ -1533,7 +1537,7 @@ bool cache::may_exceed_limits(uint64_t bytes, size_t objects) {
 
     auto would_fit_in_cache = _current_cache_size + bytes <= _max_bytes;
 
-    return !_block_puts && _free_space > bytes * 10
+    return !_block_puts && _free_space.value_or(0) > bytes * 10
            && _current_cache_objects + _reserved_cache_objects + objects
                 < _max_objects()
            && !would_fit_in_cache;
@@ -1871,7 +1875,12 @@ ss::future<> cache::do_reserve_space(uint64_t bytes, size_t objects) {
                     // all skipping the cache limit based on the same apparent
                     // free bytes.  This counter will get reset to ground
                     // truth the next time we get a disk status notification.
-                    _free_space -= bytes;
+                    if (unlikely(!_free_space.has_value())) {
+                        throw std::runtime_error(
+                          "Free space information must be available by the "
+                          "time we execute this code path");
+                    }
+                    *_free_space -= bytes;
                     break;
                 } else {
                     // No allowance, and the disk does not have a lot of

src/v/cloud_storage/cache_service.h

@@ -335,7 +335,7 @@ class cache
     /// and have to decide whether to block writes, or exceed our configured
     /// limit.
     /// (shard 0 only)
-    uint64_t _free_space{0};
+    std::optional<uint64_t> _free_space;
 
     ssx::semaphore _cleanup_sm{1, "cloud/cache"};
     std::set<std::filesystem::path> _files_in_progress;

src/v/redpanda/admin/server.cc

@@ -103,6 +103,7 @@
 #include <seastar/core/map_reduce.hh>
 #include <seastar/core/prometheus.hh>
 #include <seastar/core/reactor.hh>
+#include <seastar/core/shard_id.hh>
 #include <seastar/core/sharded.hh>
 #include <seastar/core/shared_ptr.hh>
 #include <seastar/core/smp.hh>
@@ -4347,6 +4348,14 @@ admin_server::delete_cloud_storage_lifecycle(
 ss::future<ss::json::json_return_type>
 admin_server::post_cloud_storage_cache_trim(
   std::unique_ptr<ss::http::request> req) {
+    co_await ss::smp::submit_to(ss::shard_id{0}, [this] {
+        if (!_cloud_storage_cache.local_is_initialized()) {
+            throw ss::httpd::bad_request_exception(
+              "Cloud Storage Cache is not available. Is cloud storage "
+              "enabled?");
+        }
+    });
+
     auto max_objects = get_integer_query_param(*req, "objects");
     auto max_bytes = static_cast<std::optional<size_t>>(
       get_integer_query_param(*req, "bytes"));

tests/rptest/tests/cloud_storage_cache_test.py

@@ -0,0 +1,94 @@
+import pytest
+from ducktape.utils.util import wait_until
+from requests import HTTPError
+
+from rptest.services.admin import Admin
+from rptest.services.cluster import cluster
+from rptest.services.redpanda import SISettings
+from rptest.tests.redpanda_test import RedpandaTest
+
+
+class CloudStorageCacheAdminApisNoCacheTest(RedpandaTest):
+    """
+    Test the Cloud Storage Cache Admin APIs when tiered storage is not configured.
+    """
+    def __init__(self, test_context):
+        super().__init__(test_context, num_brokers=1)
+        self.admin = Admin(self.redpanda)
+
+    @cluster(num_nodes=1)
+    def test_admin_apis(self):
+        for node in self.redpanda.nodes:
+            with pytest.raises(HTTPError) as excinfo:
+                self.admin.cloud_storage_trim(byte_limit=None,
+                                              object_limit=None,
+                                              node=node)
+
+            assert "Cloud Storage Cache is not available. Is cloud storage enabled?" == excinfo.value.response.json(
+            )['message']
+
+
+class CloudStorageCacheAdminApisTest(RedpandaTest):
+    """
+    Test the Cloud Storage Cache Admin APIs when tiered storage is configured.
+    """
+    def __init__(self, test_context):
+        super().__init__(test_context,
+                         num_brokers=1,
+                         si_settings=SISettings(test_context))
+        self.admin = Admin(self.redpanda)
+
+    def setUp(self):
+        pass
+
+    @cluster(num_nodes=1,
+             log_allow_list=["Free space information is not available"])
+    def test_admin_apis(self):
+        num_objects = 100
+        object_size = 4096
+
+        for node in self.redpanda.nodes:
+            node.account.ssh(
+                f"mkdir -p {self.redpanda.cache_dir} ; "
+                f"for n in `seq 1 {num_objects}`; do "
+                f"dd if=/dev/urandom bs={object_size} count=1 of={self.redpanda.cache_dir}/garbage_$n.bin ; done",
+            )
+
+            self.redpanda.start(clean_nodes=False)
+
+            # Assert initial conditions.
+            usage = self.admin.get_local_storage_usage(node)
+            assert usage['cloud_storage_cache_objects'] == num_objects, usage
+
+            # Trim with default settings. Nothing should be trimmed as we are well
+            # below reasonable limits.
+            # Wrapped with wait_until as it will fail until a background fiber
+            # updates information about free disk space.
+            wait_until(lambda: self.admin.cloud_storage_trim(
+                byte_limit=None, object_limit=None, node=node),
+                       timeout_sec=30,
+                       backoff_sec=1,
+                       retry_on_exc=True)
+
+            usage = self.admin.get_local_storage_usage(node)
+            assert usage['cloud_storage_cache_objects'] == num_objects, usage
+
+            # Trim with byte limit. We should trim half of objects.
+            self.admin.cloud_storage_trim(byte_limit=object_size * 50,
+                                          object_limit=None,
+                                          node=node)
+            usage = self.admin.get_local_storage_usage(node)
+
+            # Although we set the limit to size of 50 objects, the value
+            # gets multiplied by 0.8 internally so we end up with 40 objects left.
+            assert usage['cloud_storage_cache_objects'] == 40, usage
+
+            # Trim with object limit. We should trim 20 objects.
+            self.admin.cloud_storage_trim(byte_limit=None,
+                                          object_limit=20,
+                                          node=node)
+            usage = self.admin.get_local_storage_usage(node)
+
+            # Although we set the limit to 20 objects, the value
+            # gets multiplied by 0.8 internally so we end up with 16 objects left.
+            assert usage['cloud_storage_cache_objects'] == 16, usage