rust/neon
1
0
Fork 0
mirror of https://github.com/neondatabase/neon.git synced 2026-01-08 14:02:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

[3/4] make initial size estimation task sensitive to task_mgr shutdown requests

Browse Source 
This exacerbates the problem pointed out in the previous commit.
Why? Because with this patch, deleting a timeline also exposes the issue.

Extend the test to expose the problem.
This commit is contained in:
Christian Schwarz
2022-12-15 15:16:25 +01:00
committed by Christian Schwarz
parent 40a3d50883
commit 38ebd6e7a0
2 changed files with 21 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
pageserver/src/tenant/timeline.rs
View File

@@ -1290,6 +1290,9 @@ impl Timeline {
}
Ok(())
},
_ = task_mgr::shutdown_watcher() => {
anyhow::bail!("aborted because task_mgr shutdown requested");
}
new_event = async {
loop {
match timeline_state_updates.changed().await {

27
test_runner/regress/test_timeline_size.py
View File

@@ -9,6 +9,7 @@ from pathlib import Path
import psycopg2.errors
import psycopg2.extras
import pytest
from fixtures.log_helper import log
from fixtures.neon_fixtures import (
NeonEnv,
@@ -216,7 +217,10 @@ def test_timeline_size_quota(neon_env_builder: NeonEnvBuilder):
), "after the WAL is streamed, current_logical_size is expected to be calculated and to be equal its non-incremental value"
def test_timeline_initial_logical_size_calculation_cancellation(neon_env_builder: NeonEnvBuilder):
@pytest.mark.parametrize("deletion_method", ["tenant_detach", "timeline_delete"])
def test_timeline_initial_logical_size_calculation_cancellation(
neon_env_builder: NeonEnvBuilder, deletion_method: str
):
env = neon_env_builder.init_start()
client = env.pageserver.http_client()
@@ -259,16 +263,20 @@ def test_timeline_initial_logical_size_calculation_cancellation(neon_env_builder
assert_size_calculation_not_done()
log.info(
"try to delete the timeline, this should cancel size computation tasks and wait for them to finish"
)
env.pageserver.allowed_errors.append(
f".*initial size calculation.*{tenant_id}.*{timeline_id}.*aborted because task_mgr shutdown requested"
f"try to delete the timeline using {deletion_method}, this should cancel size computation tasks and wait for them to finish"
)
if deletion_method == "timeline_delete":
env.pageserver.allowed_errors.append(
f".*initial size calculation.*{tenant_id}.*{timeline_id}.*aborted because task_mgr shutdown requested"
)
delete_timeline_success: queue.Queue[bool] = queue.Queue(maxsize=1)
def delete_timeline_thread_fn():
try:
client.tenant_detach(tenant_id)
if deletion_method == "tenant_detach":
client.tenant_detach(tenant_id)
elif deletion_method == "timeline_delete":
client.timeline_delete(tenant_id, timeline_id)
delete_timeline_success.put(True)
except PageserverApiException:
delete_timeline_success.put(False)
@@ -278,9 +286,10 @@ def test_timeline_initial_logical_size_calculation_cancellation(neon_env_builder
delete_timeline_thread.start()
# give it some time to settle in the state where it waits for size computation task
time.sleep(5)
assert (
not delete_timeline_success.empty()
), "delete timeline should be stuck waiting for size computation task"
if not delete_timeline_success.empty():
assert (
False
), f"test is broken, the {deletion_method} should be stuck waiting for size computation task, got result {delete_timeline_success.get()}"
log.info(
"resume the size calculation. The failpoint checks that the timeline directory still exists."