feat(pageserver): support schedule gc-compaction (#9809)

## Problem

part of https://github.com/neondatabase/neon/issues/9114

gc-compaction can take a long time. This patch adds support for
scheduling a gc-compaction job. The compaction loop will first handle
L0->L1 compaction, and then gc compaction. The scheduled jobs are stored
in a non-persistent queue within the tenant structure.

This will be the building block for the partial compaction trigger -- if
the system determines that we need to do a gc compaction, it will
partition the keyspace and schedule several jobs. Each of these jobs
will run for a short amount of time (i.e, 1 min). L0 compaction will be
prioritized over gc compaction.

## Summary of changes
 
* Add compaction scheduler in tenant.
* Run scheduled compaction in integration tests.
* Change the manual compaction API to allow schedule a compaction
instead of immediately doing it.
* Add LSN upper bound as gc-compaction parameter. If we schedule partial
compactions, gc_cutoff might move across different runs. Therefore, we
need to pass a pre-determined gc_cutoff beforehand. (TODO: support LSN
lower bound so that we can compact arbitrary "rectangle" in the layer
map)
* Refactor the gc_compaction internal interface.

---------

Signed-off-by: Alex Chi Z <chi@neon.tech>
Co-authored-by: Christian Schwarz <christian@neon.tech>

test_runner/regress/test_compaction.py

@@ -15,7 +15,7 @@ from fixtures.pageserver.http import PageserverApiException
 from fixtures.utils import skip_in_debug_build, wait_until
 from fixtures.workload import Workload
 
-AGGRESIVE_COMPACTION_TENANT_CONF = {
+AGGRESSIVE_COMPACTION_TENANT_CONF = {
     # Disable gc and compaction. The test runs compaction manually.
     "gc_period": "0s",
     "compaction_period": "0s",
@@ -24,6 +24,7 @@ AGGRESIVE_COMPACTION_TENANT_CONF = {
     # Compact small layers
     "compaction_target_size": 1024**2,
     "image_creation_threshold": 2,
+    # "lsn_lease_length": "0s", -- TODO: would cause branch creation errors, should fix later
 }
 
 
@@ -51,7 +52,7 @@ def test_pageserver_compaction_smoke(
 page_cache_size=10
 """
 
-    env = neon_env_builder.init_start(initial_tenant_conf=AGGRESIVE_COMPACTION_TENANT_CONF)
+    env = neon_env_builder.init_start(initial_tenant_conf=AGGRESSIVE_COMPACTION_TENANT_CONF)
 
     tenant_id = env.initial_tenant
     timeline_id = env.initial_timeline
@@ -120,14 +121,25 @@ page_cache_size=10
     assert vectored_average < 8
 
 
+@skip_in_debug_build("only run with release build")
 def test_pageserver_gc_compaction_smoke(neon_env_builder: NeonEnvBuilder):
-    env = neon_env_builder.init_start(initial_tenant_conf=AGGRESIVE_COMPACTION_TENANT_CONF)
+    SMOKE_CONF = {
+        # Run both gc and gc-compaction.
+        "gc_period": "5s",
+        "compaction_period": "5s",
+        # No PiTR interval and small GC horizon
+        "pitr_interval": "0s",
+        "gc_horizon": f"{1024 ** 2}",
+        "lsn_lease_length": "0s",
+    }
+
+    env = neon_env_builder.init_start(initial_tenant_conf=SMOKE_CONF)
 
     tenant_id = env.initial_tenant
     timeline_id = env.initial_timeline
 
-    row_count = 1000
-    churn_rounds = 10
+    row_count = 10000
+    churn_rounds = 50
 
     ps_http = env.pageserver.http_client()
 
@@ -141,20 +153,27 @@ def test_pageserver_gc_compaction_smoke(neon_env_builder: NeonEnvBuilder):
         if i % 10 == 0:
             log.info(f"Running churn round {i}/{churn_rounds} ...")
 
-        workload.churn_rows(row_count, env.pageserver.id)
-        # Force L0 compaction to ensure the number of layers is within bounds, so that gc-compaction can run.
-        ps_http.timeline_compact(tenant_id, timeline_id, force_l0_compaction=True)
-        assert ps_http.perf_info(tenant_id, timeline_id)[0]["num_of_l0"] <= 1
         ps_http.timeline_compact(
             tenant_id,
             timeline_id,
             enhanced_gc_bottom_most_compaction=True,
             body={
-                "start": "000000000000000000000000000000000000",
-                "end": "030000000000000000000000000000000000",
+                "scheduled": True,
+                "compact_range": {
+                    "start": "000000000000000000000000000000000000",
+                    # skip the SLRU range for now -- it races with get-lsn-by-timestamp, TODO: fix this
+                    "end": "010000000000000000000000000000000000",
+                },
             },
         )
 
+        workload.churn_rows(row_count, env.pageserver.id)
+
+    # ensure gc_compaction is scheduled and it's actually running (instead of skipping due to no layers picked)
+    env.pageserver.assert_log_contains(
+        "scheduled_compact_timeline.*picked .* layers for compaction"
+    )
+
     log.info("Validating at workload end ...")
     workload.validate(env.pageserver.id)