From d7bb3d14df48e61f0583f672504c38bb293d512c Mon Sep 17 00:00:00 2001
From: Bojan Serafimov <bojan@pop-os.localdomain>
Date: Mon, 30 May 2022 15:39:08 -0400
Subject: [PATCH] WIP

---
 .circleci/config.yml                |  3 ++-
 pageserver/src/bin/pageserver.rs    | 18 +++++++++++++++++-
 test_runner/batch_others/test_ci.py | 11 +++++++++++
 3 files changed, 30 insertions(+), 2 deletions(-)
 create mode 100644 test_runner/batch_others/test_ci.py

diff --git a/.circleci/config.yml b/.circleci/config.yml
index 5346e35c01..233157d052 100644
--- a/.circleci/config.yml
+++ b/.circleci/config.yml
@@ -293,7 +293,7 @@ jobs:
           # `Too long with no output` error, if a test is running for a long time.
           # In that case, tests should have internal timeouts that are less than
           # no_output_timeout, specified here.
-          no_output_timeout: 10m
+          no_output_timeout: 1m
           environment:
             - ZENITH_BIN: /tmp/zenith/bin
             - POSTGRES_DISTRIB_DIR: /tmp/zenith/pg_install
@@ -354,6 +354,7 @@ jobs:
               fi
             fi
       - run:
+          # TODO wait for processes to die in case of timeout?
           # CircleCI artifacts are preserved one file at a time, so skipping
           # this step isn't a good idea. If you want to extract the
           # pageserver state, perhaps a tarball would be a better idea.
diff --git a/pageserver/src/bin/pageserver.rs b/pageserver/src/bin/pageserver.rs
index ac90500b97..8b88565a63 100644
--- a/pageserver/src/bin/pageserver.rs
+++ b/pageserver/src/bin/pageserver.rs
@@ -1,6 +1,6 @@
 //! Main entry point for the Page Server executable.
 
-use std::{env, path::Path, str::FromStr};
+use std::{env, fs::File, path::Path, process, str::FromStr, thread::sleep, time::Duration};
 use tracing::*;
 
 use anyhow::{bail, Context, Result};
@@ -276,6 +276,22 @@ fn start_pageserver(conf: &'static PageServerConf, daemonize: bool) -> Result<()
 
     let remote_index = tenant_mgr::init_tenant_mgr(conf)?;
 
+    // Create file and frequently check if it's still here
+    thread_mgr::spawn(
+        ThreadKind::HttpEndpointListener,
+        None,
+        None,
+        "http_endpoint_thread",
+        true,
+        move || {
+            File::create("delete-me.txt").expect("FFFF failed creating file");
+            loop {
+                File::open("delete-me.txt").expect("FFFF cannot find file");
+                sleep(Duration::from_millis(10));
+            }
+        },
+    )?;
+
     // Spawn a new thread for the http endpoint
     // bind before launching separate thread so the error reported before startup exits
     let auth_cloned = auth.clone();
diff --git a/test_runner/batch_others/test_ci.py b/test_runner/batch_others/test_ci.py
new file mode 100644
index 0000000000..e936ad076b
--- /dev/null
+++ b/test_runner/batch_others/test_ci.py
@@ -0,0 +1,11 @@
+import pytest
+import os
+import time
+
+from fixtures.zenith_fixtures import ZenithEnvBuilder
+
+def test_timeout(zenith_env_builder: ZenithEnvBuilder, test_output_dir: str):
+    env = zenith_env_builder.init_start()
+
+    # Sleep long enough with no output so CI step times out
+    time.sleep(1000)