add convenience script to start/stop the bot

README.md

@@ -1409,6 +1409,29 @@ The job manager writes log information to the file `eessi_bot_job_manager.log`.
 
 The job manager can run on a different machine than the event handler, as long as both have access to the same shared filesystem.
 
+## <a name="step7.3"></a>Step 7.3: Managing the bot with a single command
+
+The `bot` script starts, stops, restarts, and checks the status of the 2 bot
+components: the event handler and the job manager.  It launches both component
+processes in the background.
+
+```bash
+./bot start [options]
+./bot stop
+./bot restart [options]
+./bot status
+```
+
+Available `start`/`restart` options are the same as those for
+`event_handler.sh` and `job_manager.sh`.
+
+The following example starts the bot with 10 job manager iterations, managing
+only job ids 1234 and 5678, and listening for events on port 8080:
+
+```bash
+./bot start -i 10 -j 1234,5678 --port 8080
+```
+
 # Example pull request on software-layer
 
 For information on how to make pull requests and let the bot build software, see

bot.py

@@ -0,0 +1,177 @@
+#!/usr/bin/env python3
+#
+# This file is part of the EESSI build-and-deploy bot,
+# see https://github.com/EESSI/eessi-bot-software-layer
+#
+# The bot helps with requests to add software installations to the
+# EESSI software layer, see https://github.com/EESSI/software-layer
+#
+# author: Samuel Moors (@smoors)
+#
+# license: GPLv2
+#
+
+import argparse
+import os
+import shlex
+import signal
+import subprocess
+import time
+from pathlib import Path
+
+RED = "\033[31m"
+BLUE = "\033[34m"
+GREEN = "\033[32m"
+RESET = "\033[0m"
+
+JOB_MANAGER_MODULE = "eessi_bot_job_manager"
+EVENT_HANDLER_MODULE = "eessi_bot_event_handler"
+DEFAULT_BOT_NAME = "eessi-bot"
+
+
+def parse_args():
+    parser = argparse.ArgumentParser(
+        usage="%(prog)s {start|stop|restart|status} [options]"
+    )
+    parser.add_argument("command", choices=("start", "stop", "restart", "status"))
+    parser.add_argument(
+        "--bot-name",
+        default=DEFAULT_BOT_NAME,
+        help=f"bot name (default: {DEFAULT_BOT_NAME})",
+    )
+    parser.add_argument(
+        "--job-manager-opts",
+        default="",
+        help="additional job manager options as a string",
+    )
+    parser.add_argument(
+        "--event-handler-opts",
+        default="",
+        help="additional event handler options as a string",
+    )
+    args = parser.parse_args()
+
+    return args
+
+
+def get_cmdline(pid):
+    try:
+        data = Path(f"/proc/{pid}/cmdline").read_bytes()
+    except (FileNotFoundError, PermissionError, ProcessLookupError):
+        return []
+    return [arg.decode(errors="replace") for arg in data.split(b"\0") if arg]
+
+
+def find_processes(module, bot_name):
+    processes = []
+
+    for proc in Path("/proc").iterdir():
+        if not proc.name.isdigit():
+            continue
+
+        pid = int(proc.name)
+        if pid == os.getpid():
+            continue
+
+        cmdline = get_cmdline(pid)
+        if not cmdline:
+            continue
+
+        try:
+            module_index = cmdline.index("-m")
+        except ValueError:
+            continue
+
+        if module_index + 1 >= len(cmdline) or cmdline[module_index + 1] != module:
+            continue
+
+        for i, arg in enumerate(cmdline[:-1]):
+            if arg == "--bot-name" and cmdline[i + 1] == bot_name:
+                processes.append(pid)
+                break
+
+    return processes
+
+
+def start_process(module, bot_name, opts):
+    cmd = ["python3", "-m", module, "--bot-name", bot_name]
+    if opts:
+        cmd.extend(shlex.split(opts))
+
+    env = os.environ.copy()
+    env["PYTHONUNBUFFERED"] = "1"
+
+    subprocess.Popen(cmd, env=env, start_new_session=True)
+
+
+def start_bot(bot_name, job_manager_opts, event_handler_opts):
+    if find_processes(JOB_MANAGER_MODULE, bot_name):
+        print(f"{BLUE}>>> job manager for bot '{bot_name}' is already running{RESET}")
+    else:
+        print(f"{BLUE}>>> starting job manager for bot '{bot_name}'...{RESET}")
+        start_process(JOB_MANAGER_MODULE, bot_name, job_manager_opts)
+
+    if find_processes(EVENT_HANDLER_MODULE, bot_name):
+        print(f"{GREEN}>>> event handler for bot '{bot_name}' is already running{RESET}")
+    else:
+        print(f"{GREEN}>>> starting event handler for bot '{bot_name}'...{RESET}")
+        start_process(EVENT_HANDLER_MODULE, bot_name, event_handler_opts)
+
+
+def stop_processes(module, bot_name):
+    for pid in find_processes(module, bot_name):
+        try:
+            cmdline = get_cmdline(pid)
+            os.kill(pid, signal.SIGTERM)
+            print(f"killed (pid {pid}): {shlex.join(cmdline)}")
+        except ProcessLookupError:
+            pass
+
+
+def stop_bot(bot_name):
+    if find_processes(EVENT_HANDLER_MODULE, bot_name):
+        print(f"{RED}>>> stopping event handler for bot '{bot_name}'...{RESET}")
+        stop_processes(EVENT_HANDLER_MODULE, bot_name)
+    else:
+        print(f"{RED}>>> event handler for bot '{bot_name}' is not running{RESET}")
+
+    if find_processes(JOB_MANAGER_MODULE, bot_name):
+        print(f"{RED}>>> stopping job manager for bot '{bot_name}'...{RESET}")
+        stop_processes(JOB_MANAGER_MODULE, bot_name)
+    else:
+        print(f"{RED}>>> job manager for bot '{bot_name}' is not running{RESET}")
+
+
+def status_bot(bot_name):
+    if find_processes(JOB_MANAGER_MODULE, bot_name):
+        print(f"{BLUE}>>> job manager for bot '{bot_name}' is running{RESET}")
+    else:
+        print(f"{RED}>>> job manager for bot '{bot_name}' is not running{RESET}")
+
+    if find_processes(EVENT_HANDLER_MODULE, bot_name):
+        print(f"{GREEN}>>> event handler for bot '{bot_name}' is running{RESET}")
+    else:
+        print(f"{RED}>>> event handler for bot '{bot_name}' is not running{RESET}")
+
+
+def main():
+    args = parse_args()
+
+    if args.command == "start":
+        start_bot(args.bot_name, args.job_manager_opts, args.event_handler_opts)
+        time.sleep(2)
+        status_bot(args.bot_name)
+    elif args.command == "stop":
+        stop_bot(args.bot_name)
+    elif args.command == "restart":
+        stop_bot(args.bot_name)
+        time.sleep(1)
+        start_bot(args.bot_name, args.job_manager_opts, args.event_handler_opts)
+        time.sleep(2)
+        status_bot(args.bot_name)
+    elif args.command == "status":
+        status_bot(args.bot_name)
+
+
+if __name__ == "__main__":
+    main()

tests/test_tools_args.py

@@ -44,15 +44,17 @@ def test_parse_common_args(test_args, expected_parsed, expected_unknown):
 # Test event_handler_parse()
 @pytest.mark.parametrize("test_args,expectation", [
     # No args
-    ([], nullcontext(Namespace(debug=False, build=False, test=False, cron=False, file=None, port=3000))),
+    ([], nullcontext(Namespace(debug=False, build=False, test=False, cron=False, file=None, port=3000, bot_name=None))),
 
     # Short-form args
     (["-d", "-b", "-t", "-c", "-f", "file.json", "-p", "8000"],
-     nullcontext(Namespace(debug=True, build=True, test=True, cron=True, file="file.json", port="8000"))),
+     nullcontext(Namespace(debug=True, build=True, test=True, cron=True, file="file.json", port="8000",
+                           bot_name=None))),
 
     # Long-form args
-    (["--debug", "--build", "--test", "--cron", "--file", "file2.json", "--port", "9000"],
-     nullcontext(Namespace(debug=True, build=True, test=True, cron=True, file="file2.json", port="9000"))),
+    (["--debug", "--build", "--test", "--cron", "--file", "file2.json", "--port", "9000", "--bot-name", "test-bot"],
+     nullcontext(Namespace(debug=True, build=True, test=True, cron=True, file="file2.json", port="9000",
+                           bot_name="test-bot"))),
 
     # Unknown args - should fail and exit
     (["-u"], pytest.raises(SystemExit)),
@@ -66,15 +68,15 @@ def test_event_handler_parse_known_args(test_args, expectation):
 # Test job_manager_parse()
 @pytest.mark.parametrize("test_args,expectation", [
     # No args
-    ([], nullcontext(Namespace(debug=False, max_manager_iterations=-1, jobs=None))),
+    ([], nullcontext(Namespace(debug=False, max_manager_iterations=-1, jobs=None, bot_name=None))),
 
     # Short-form args
     (["-d", "-i", "0", "-j", "17"],
-     nullcontext(Namespace(debug=True, max_manager_iterations="0", jobs="17"))),
+     nullcontext(Namespace(debug=True, max_manager_iterations="0", jobs="17", bot_name=None))),
 
     # Long-form args
-    (["--debug", "--max-manager-iterations", "10", "--jobs", "4,18,48"],
-     nullcontext(Namespace(debug=True, max_manager_iterations="10", jobs="4,18,48"))),
+    (["--debug", "--max-manager-iterations", "10", "--jobs", "4,18,48", "--bot-name", "test-bot"],
+     nullcontext(Namespace(debug=True, max_manager_iterations="10", jobs="4,18,48", bot_name="test-bot"))),
 
     # Unknown args - should fail and exit
     (["-u"], pytest.raises(SystemExit)),

tools/args.py

@@ -85,6 +85,11 @@ def event_handler_parse(args=None):
         help="listen on a specific port for events (default 3000)",
     )
 
+    parser.add_argument(
+        "--bot-name",
+        help="bot name (not used by the bot itself), useful for keeping track of multiple instances",
+    )
+
     return parser.parse_args(args=unknown_args, namespace=parsed_args)
 
 
@@ -110,4 +115,9 @@ def job_manager_parse(args=None):
         help="limits the processing to a specific job id or list of comma-separated list of job ids",
     )
 
+    parser.add_argument(
+        "--bot-name",
+        help="bot name (not used by the bot itself), useful for keeping track of multiple instances",
+    )
+
     return parser.parse_args(args=unknown_args, namespace=parsed_args)