Review 1.4: AX batch rules: one finding per job, stuck jobs only once, no crash without a start
- Axapta 3.0/4.0 keep a BATCH row per recurrence, so a failed run and the next waiting run share a job key and gave two SQLI-AX-001 findings with the same subject (the repository kept the last, the hub the first). The batch rules now keep one finding per subject, the most severe. - SQLI-AX-004 treated a single earlier 10-hour run as "normally that long" and stayed silent; it now needs BatchHistoryMinimumRuns runs, like SQLI-AX-002. A stuck job is SQLI-AX-004's finding only, not also -002's. - Running(now)!.Value threw for an executing job with neither a start time nor a first-seen time (and a negative duration meant clock skew); such jobs are skipped. - SQLI-AX-010 no longer calls every session in an AX database "from Dynamics AX", and shows the raw CONTEXT_INFO as sensitive evidence. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This commit is contained in:
parent
8365ba5357
commit
13f22c7541
2 changed files with 82 additions and 7 deletions
|
|
@ -57,11 +57,33 @@ public abstract class DynamicsAxBatchRule : DynamicsAxRule
|
|||
ArgumentNullException.ThrowIfNull(context);
|
||||
var snapshot = context.Data.Get<AxBatchSnapshot>();
|
||||
var (measured, outcome) = Measured(snapshot.Databases, d => d.Database, d => d.NotMeasurable);
|
||||
return outcome ?? Result(measured.SelectMany(d => Check(d, snapshot.Thresholds, context.Now).Take(MaxFindingsPerDatabase)));
|
||||
|
||||
// One finding per subject: Axapta 3.0/4.0 keep a BATCH row per recurrence, so several rows share a job key (e.g. the
|
||||
// failed run and the next waiting one). The most severe wins, then the first (the newest), so the finding the hub
|
||||
// and the repository keep is the same.
|
||||
return outcome ?? Result(measured.SelectMany(d => Check(d, snapshot.Thresholds, context.Now)
|
||||
.GroupBy(f => f.Subject, StringComparer.Ordinal)
|
||||
.Select(g => g.OrderByDescending(f => f.Severity).First())
|
||||
.Take(MaxFindingsPerDatabase)));
|
||||
}
|
||||
|
||||
protected abstract IEnumerable<Finding> Check(AxBatchDatabase database, AxThresholds thresholds, DateTimeOffset now);
|
||||
|
||||
/// <summary>How long an executing job has run, or null when unknown (or negative: clocks or time zones disagree).</summary>
|
||||
protected static TimeSpan? RunningFor(AxBatchJob job, DateTimeOffset now) =>
|
||||
job.Status == AxBatchStatus.Executing && job.Running(now) is { } running && running >= TimeSpan.Zero ? running : null;
|
||||
|
||||
/// <summary>
|
||||
/// SQLI-AX-004's condition: executing at least BatchStuckHours and not normally that long by the job's own history
|
||||
/// (which needs BatchHistoryMinimumRuns runs, as for SQLI-AX-002).
|
||||
/// </summary>
|
||||
protected static bool IsStuck(AxBatchJob job, TimeSpan running, AxThresholds thresholds)
|
||||
{
|
||||
var normallyLong = job.HistoryRuns >= thresholds.BatchHistoryMinimumRuns && job.P90Seconds is { } p90
|
||||
&& running.TotalSeconds <= p90 * thresholds.BatchLongRunningFactor;
|
||||
return running >= TimeSpan.FromHours(thresholds.BatchStuckHours) && !normallyLong;
|
||||
}
|
||||
|
||||
protected Finding JobFinding(AxBatchDatabase database, AxBatchJob job, Severity severity, string title, string explanation, string recommendation,
|
||||
IEnumerable<Evidence> evidence) =>
|
||||
CreateFinding(severity, title, explanation, recommendation,
|
||||
|
|
@ -125,7 +147,12 @@ public sealed class AxBatchLongRunningRule : DynamicsAxBatchRule
|
|||
{
|
||||
foreach (var job in database.Jobs.Where(j => j.Status == AxBatchStatus.Executing && j.HistoryRuns >= thresholds.BatchHistoryMinimumRuns && j.P90Seconds is not null))
|
||||
{
|
||||
var running = job.Running(now)!.Value;
|
||||
// A job that hangs is SQLI-AX-004's finding, not also this one.
|
||||
if (RunningFor(job, now) is not { } running || IsStuck(job, running, thresholds))
|
||||
{
|
||||
continue;
|
||||
}
|
||||
|
||||
var limit = TimeSpan.FromSeconds(Math.Max(job.P90Seconds!.Value * thresholds.BatchLongRunningFactor, thresholds.BatchLongRunningMinimumMinutes * 60.0));
|
||||
if (running < limit)
|
||||
{
|
||||
|
|
@ -195,9 +222,7 @@ public sealed class AxBatchStuckRule : DynamicsAxBatchRule
|
|||
{
|
||||
foreach (var job in database.Jobs.Where(j => j.Status == AxBatchStatus.Executing))
|
||||
{
|
||||
var running = job.Running(now)!.Value;
|
||||
var normallyLong = job.P90Seconds is { } p90 && running.TotalSeconds <= p90 * thresholds.BatchLongRunningFactor;
|
||||
if (running < TimeSpan.FromHours(thresholds.BatchStuckHours) || normallyLong)
|
||||
if (RunningFor(job, now) is not { } running || !IsStuck(job, running, thresholds))
|
||||
{
|
||||
continue;
|
||||
}
|
||||
|
|
@ -464,8 +489,9 @@ public sealed class AxBlockingUserRule : DynamicsAxRule
|
|||
findings.Add(CreateFinding(longest >= 300_000 || victims.Count >= 10 ? Severity.High : Severity.Medium,
|
||||
who.Known
|
||||
? $"AX-brugeren bag session {head.SessionId} blokerer {victims.Count} session(er) i {database.Database}"
|
||||
: $"En AX-session ({head.SessionId}) blokerer {victims.Count} session(er) i {database.Database}",
|
||||
"En session fra Dynamics AX holder låse, som andre AX-brugere eller batchjob venter på. I AX går alle brugere gennem AOS'ens login, "
|
||||
: $"Session {head.SessionId} i AX-databasen blokerer {victims.Count} session(er) i {database.Database}",
|
||||
"En session i Dynamics AX-databasen holder låse, som andre AX-brugere eller batchjob venter på (er det ikke AX selv, ses det af "
|
||||
+ "programmet under målingerne). I AX går alle brugere gennem AOS'ens login, "
|
||||
+ "så SQL Server-login'et siger ikke, hvem det er. " + who.Explanation,
|
||||
who.Known
|
||||
? "Kontakt brugeren (se AX-bruger under målinger) og find ud af, hvad de er i gang med – typisk en stor bogføring, en rapport eller en "
|
||||
|
|
@ -478,6 +504,7 @@ public sealed class AxBlockingUserRule : DynamicsAxRule
|
|||
new Evidence("Kilde", who.Source),
|
||||
new Evidence("SQL-session", $"{head.SessionId} ({head.Status}), {head.OpenTransactionCount} åbne transaktioner"),
|
||||
new Evidence("Vært / program", $"{head.HostName} / {head.ProgramName}", Sensitive: true),
|
||||
new Evidence("CONTEXT_INFO (rå)", who.RawContext ?? "–", Sensitive: true),
|
||||
new Evidence("Blokerede sessioner", victims.Count.ToString(CultureInfo.InvariantCulture)),
|
||||
new Evidence("Længste ventetid", $"{Format.Number(longest / 1000.0)} s"),
|
||||
new Evidence("Sætning", head.SqlText is { Length: > 500 } s ? s[..500] + " …" : head.SqlText ?? "–", Sensitive: true),
|
||||
|
|
|
|||
|
|
@ -296,4 +296,52 @@ public class DynamicsAxRuleTests
|
|||
Assert.Contains("sætter ikke CONTEXT_INFO", finding.Explanation, StringComparison.Ordinal);
|
||||
Assert.Equal(RuleOutcome.Passed, Run(new AxBlockingUserRule(), With(AxDatabases(AxVersion.Axapta30), Blocking(null, waitMs: 5_000))).Outcome);
|
||||
}
|
||||
|
||||
[Fact]
|
||||
public void Ax010_ax30_keeps_host_and_program_out_of_the_explanation_which_may_reach_the_hub()
|
||||
{
|
||||
var finding = Assert.Single(Run(new AxBlockingUserRule(), With(AxDatabases(AxVersion.Axapta30), Blocking(null))).Findings);
|
||||
|
||||
Assert.DoesNotContain("AOS01", finding.Explanation + finding.Title + finding.Subject, StringComparison.Ordinal);
|
||||
Assert.DoesNotContain("Microsoft Dynamics AX", finding.Explanation, StringComparison.Ordinal);
|
||||
Assert.True(finding.Evidence.Single(e => e.Name == "Vært / program").Sensitive);
|
||||
Assert.True(finding.Evidence.Single(e => e.Name == "CONTEXT_INFO (rå)").Sensitive);
|
||||
}
|
||||
|
||||
[Fact]
|
||||
public void Ax001_ax30_rows_of_one_recurring_job_give_one_finding_with_the_highest_severity()
|
||||
{
|
||||
// AX 3.0 keeps a BATCH row per recurrence: the failed run and the next waiting one share the job key.
|
||||
var history = new AxBatchHistory();
|
||||
history.Add("AX30", new AxBatchRun("4711|dat|NAT", "R1", "Lagerafslutning", "dat", Utc.AddHours(-3), Utc.AddHours(-2), AxBatchStatus.Error));
|
||||
var failed = Job30(Batch30(50, 2, Utc.AddHours(-3)), history) with { EndedUtc = Utc.AddHours(-2) };
|
||||
var waiting = Job30(Batch30(51, 0, Utc.AddHours(1)), history);
|
||||
|
||||
var finding = Assert.Single(Run(new AxBatchFailedRule(), With(Batch(failed, waiting))).Findings);
|
||||
|
||||
Assert.Equal(Severity.High, finding.Severity);
|
||||
Assert.Equal("database: AX30 · batch: 4711|dat|NAT", finding.Subject);
|
||||
}
|
||||
|
||||
[Fact]
|
||||
public void Ax004_one_long_earlier_run_does_not_hide_a_stuck_job_and_ax002_leaves_a_stuck_job_to_ax004()
|
||||
{
|
||||
var row = Batch30(60, 1, Utc.AddHours(-9));
|
||||
var oneLongRun = Job30(row, History("4711|dat|NAT", 1, TimeSpan.FromHours(10)), Utc.AddHours(-7));
|
||||
Assert.Single(Run(new AxBatchStuckRule(), With(Batch(oneLongRun))).Findings);
|
||||
|
||||
var stuckWithHistory = Job30(row, History("4711|dat|NAT", 10, TimeSpan.FromMinutes(20)), Utc.AddHours(-7));
|
||||
Assert.Single(Run(new AxBatchStuckRule(), With(Batch(stuckWithHistory))).Findings);
|
||||
Assert.Equal(RuleOutcome.Passed, Run(new AxBatchLongRunningRule(), With(Batch(stuckWithHistory))).Outcome);
|
||||
}
|
||||
|
||||
[Fact]
|
||||
public void An_executing_job_without_any_start_time_is_skipped_not_an_error()
|
||||
{
|
||||
var unknown = Job30(Batch30(70, 1, Utc.AddHours(-9)), History("4711|dat|NAT", 10, TimeSpan.FromMinutes(20)));
|
||||
Assert.Null(unknown.Running(Now));
|
||||
|
||||
Assert.Equal(RuleOutcome.Passed, Run(new AxBatchStuckRule(), With(Batch(unknown))).Outcome);
|
||||
Assert.Equal(RuleOutcome.Passed, Run(new AxBatchLongRunningRule(), With(Batch(unknown))).Outcome);
|
||||
}
|
||||
}
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue