feat(xceadmin): add 'config dump' command and size-string thresholds

Adds `xceadmin config dump CLUSTER_BIN_PATH`, the reverse of `config
build` -- decodes a serialized ClusterConfig back to human-readable text
so a built config can be inspected without a throwaway pycapnp script.
Motivated directly by how bug-1556 (hosts.yaml's max_slot: -1 auto-detect
sentinel silently becoming mxj=0) was actually found: by hand-decoding a
live cluster.bin. The dump renders the same sentinel as
"max_slot=-1 (auto-detect from CPU count)", not the raw uint32.

Also adds parse_size_or_number() to xcecommon.mapping so a queue/host's
sched_stop threshold on a byte-valued resource (maxmem/mem/tmp/maxswap/
maxtmp) can be written as a human-friendly size string ("500MB", "2GB")
instead of a raw byte count; a plain number still passes through
unchanged for non-byte resources like load averages. hosts.schema.json
and queues.schema.json's sched_stop schema gains a matching SizeOrNumber
type. dump_cluster_config() humanizes the same byte-valued thresholds
back on the way out (_format_bytes_human).

New ClusterBinFileError covers every local cluster.bin read/decode
failure mode without letting a raw OSError/capnp.KjException escape the
CLI. RESOURCE_VALTYPE/RESOURCE_ORDTYPE/RESOURCE_FLAG_* made public in
mapping.py so dump_cluster_config can reuse the same int<->name tables
build_cluster_config already uses, keeping both directions in sync by
construction.

256 xceadmin tests pass (incl. new test_config_dump.py and the
parse_size_or_number/SizeOrNumber cases in test_mapping.py/test_cli.py),
verified via `xce-build.sh xceadmin`'s full pinned out-of-tree install
+ pytest run before this commit.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This commit is contained in:
Rongsong Shen 2026-08-16 03:56:52 +00:00
parent d479b70764
commit 3afe3d5549
Signed by: rshen
SSH key fingerprint: SHA256:8Vf2pZ3h6rJa7EhGK+WaAQ9Mta/apZvteW6g9U/WKKk
9 changed files with 560 additions and 19 deletions

View file

@ -51,6 +51,7 @@ def test_cli_config_help_shows_build_and_push() -> None:
assert result.exit_code == 0
assert "build" in result.output
assert "push" in result.output
assert "dump" in result.output
def test_cli_group_invoked_with_no_args_does_not_traceback() -> None:
@ -132,6 +133,39 @@ def test_config_build_missing_configs_dir_does_not_raw_traceback(tmp_path):
assert result.exit_code == 0, "a nonexistent dir behaves like one with zero known files present (1.5)"
# ---------------------------------------------------------------------------
# config dump
# ---------------------------------------------------------------------------
def test_config_dump_prints_decoded_text(tmp_path):
(tmp_path / "xce.yaml").write_text(
yaml.safe_dump(
{"cluster_name": "eda1", "heartbeat_interval": 30, "inactive_heartbeat_count": 3}
)
)
bin_path = tmp_path / "cluster.bin"
runner = CliRunner()
build_result = runner.invoke(cli, ["config", "build", str(tmp_path), "-o", str(bin_path)])
assert build_result.exit_code == 0, build_result.output
result = runner.invoke(cli, ["config", "dump", str(bin_path)])
assert result.exit_code == 0, result.output
assert "=== ClusterConfig ===" in result.output
assert "clusterName: eda1" in result.output
def test_config_dump_missing_file_exits_nonzero_with_message(tmp_path):
missing = tmp_path / "does-not-exist.bin"
runner = CliRunner()
result = runner.invoke(cli, ["config", "dump", str(missing)])
assert result.exit_code != 0
assert not isinstance(result.exception, KeyError | AttributeError | TypeError)
assert result.output.strip() != ""
# ---------------------------------------------------------------------------
# config push / config push --dry-run (mock controller)
# ---------------------------------------------------------------------------

View file

@ -0,0 +1,128 @@
"""Tests for ``xceadmin.build.dump_cluster_config`` -- the reverse of
``build_cluster_config`` (decode a ``cluster.bin`` back to human-readable
text). Uses the same ``tests/fixtures/full`` real-config fixture as
test_config_build.py rather than a second golden text fixture: the
assertions below check for specific substrings (mirrors
test_scheduler.py's ``dump()`` test style), not a byte-exact text match,
so formatting can evolve without a brittle golden file.
"""
from __future__ import annotations
from pathlib import Path
import pytest
import yaml
from xcecommon.errors import ClusterBinFileError
from xceadmin import build
FIXTURES_DIR = Path(__file__).parent / "fixtures"
@pytest.fixture
def full_cluster_bin(tmp_path):
data = build.build_cluster_config(str(FIXTURES_DIR / "full" / "configs"))
path = tmp_path / "cluster.bin"
path.write_bytes(data)
return path
# ---------------------------------------------------------------------------
# Section content
# ---------------------------------------------------------------------------
def test_dump_lists_cluster_scalars(full_cluster_bin):
text = build.dump_cluster_config(str(full_cluster_bin))
assert "=== ClusterConfig ===" in text
assert "clusterName:" in text
assert "heartbeatInterval:" in text
def test_dump_lists_each_resource_with_decoded_type_and_order(full_cluster_bin):
text = build.dump_cluster_config(str(full_cluster_bin))
assert "=== resources (" in text
# "mem" is numeric/decrease/not consumable -- valtype/ordtype/flags int
# codes must render back to their YAML names, not raw ints.
assert "name=mem type=numeric order=decrease" in text
assert "consumable=False" in text
def test_dump_renders_mxj_auto_detect_sentinel_not_raw_uint32(full_cluster_bin):
"""The exact bug that motivated this feature: hosts.yaml's `default`
host has `max_slot: -1`, which wire-encodes as HOSTCONFIG_MXJ_AUTO_DETECT
(0xFFFFFFFF). The dump must show the human meaning, not the raw UInt32.
"""
text = build.dump_cluster_config(str(full_cluster_bin))
assert "=== host: default ===" in text
assert "max_slot=-1 (auto-detect from CPU count)" in text
# worker2 has an explicit max_slot: 256 -- shown as a plain number.
assert "=== host: worker2 ===" in text
assert "max_slot=256" in text
def test_dump_lists_queues_groups_scalers_and_limits(full_cluster_bin):
text = build.dump_cluster_config(str(full_cluster_bin))
assert "=== queues (" in text
assert "=== ugroups (" in text
assert "=== hgroups (" in text
assert "=== hproviders (" in text
assert "=== limits (" in text
assert "=== limit: ug1_slots ===" in text
def test_dump_output_ends_with_trailing_newline(full_cluster_bin):
text = build.dump_cluster_config(str(full_cluster_bin))
assert text.endswith("\n")
assert not text.endswith("\n\n")
def test_dump_humanizes_byte_valued_threshold(tmp_path):
"""A host's sched_stop threshold on a byte-valued resource (maxmem) is
shown as a human-readable size, not a raw float byte count -- mirrors
the mem/tmp/maxswap/maxtmp/maxmem unit-consistency work this session.
"""
(tmp_path / "hosts.yaml").write_text(
yaml.safe_dump(
{
"hosts": [
{
"hostname": "h1",
"sched_stop": [{"resname": "maxmem", "sched": "500MB", "stop": "1GB"}],
}
]
}
)
)
data = build.build_cluster_config(str(tmp_path))
bin_path = tmp_path / "cluster.bin"
bin_path.write_bytes(data)
text = build.dump_cluster_config(str(bin_path))
assert "threshold: name=maxmem sched=500.00 MB stop=1.00 GB" in text
# ---------------------------------------------------------------------------
# Error handling
# ---------------------------------------------------------------------------
def test_dump_missing_file_raises_cluster_bin_file_error(tmp_path):
missing = tmp_path / "does-not-exist.bin"
with pytest.raises(ClusterBinFileError):
build.dump_cluster_config(str(missing))
def test_dump_empty_file_raises_cluster_bin_file_error(tmp_path):
empty = tmp_path / "empty.bin"
empty.write_bytes(b"")
with pytest.raises(ClusterBinFileError):
build.dump_cluster_config(str(empty))
def test_dump_garbage_bytes_raises_cluster_bin_file_error_not_crash(tmp_path):
garbage = tmp_path / "garbage.bin"
garbage.write_bytes(b"not a cluster config, just noise")
with pytest.raises(ClusterBinFileError):
build.dump_cluster_config(str(garbage))

View file

@ -293,6 +293,61 @@ def test_map_list_section_host_config_sets_mxj_auto_detect_sentinel_when_max_slo
assert mapping.HOSTCONFIG_MXJ_AUTO_DETECT != 0
@pytest.mark.parametrize(
("value", "expected"),
[
(3.5, 3.5),
(2000, 2000.0),
(0, 0.0),
("500MB", 500 * 1024**2),
("2GB", 2 * 1024**3),
("1.5GB", 1.5 * 1024**3),
("500 MB", 500 * 1024**2), # space between number and unit
("500mb", 500 * 1024**2), # lowercase unit
("1024B", 1024.0),
("-1KB", -1024.0),
],
)
def test_parse_size_or_number_accepts_plain_numbers_and_size_strings(value, expected):
assert mapping.parse_size_or_number(value) == expected
@pytest.mark.parametrize(
"value",
[
"not a number",
"500XB", # unrecognized unit
"GB", # missing number
"500", # missing unit -- ambiguous, must be a plain number instead
None,
[1, 2],
],
)
def test_parse_size_or_number_rejects_unparseable_values(value):
with pytest.raises((TypeError, ValueError)):
mapping.parse_size_or_number(value)
def test_map_list_section_host_config_sched_stop_accepts_size_strings(cluster_config):
"""The production case for parse_size_or_number: an operator writes a
human-friendly size string for a byte-valued resource's threshold
(e.g. maxmem/mem/tmp/maxswap/maxtmp -- see test.c's
is_bytes_valued_resource) instead of computing the raw byte count by
hand."""
items = [
{
"hostname": "worker4",
"sched_stop": [{"resname": "tmp", "sched": "500MB", "stop": "100MB"}],
}
]
mapping.map_list_section(_spec_for("hostConfigs"), items, cluster_config)
threshold = cluster_config.hostConfigs[0].thresholds[0]
assert threshold.name == "tmp"
assert threshold.sched == 500 * 1024**2
assert threshold.stop == 100 * 1024**2
def test_time_window_day_prefixed_spec_parses_day_hour_minute(cluster_config):
items = [
{