Monitoring, Alerts & Troubleshooting

In which directory are HANA trace files for the system database located, and where do the tenant databases' trace files live?

What is the name form of a SAP HANA alert trace file, and which parameter limits its size?

You query STATISTICS_SCHEDULE and every row's key is just a bare number, with no hint of what each collector does. How do you find out which collector is which before editing one?

A HANA instance crashed and you grab the crashdump expecting a complete image of memory at the moment of failure. What do you actually get, and why?

A junior admin asks where the 'system views browser' is in HANA cockpit or Studio so they can click through the monitoring views. What do you tell them?

A user reports a performance anomaly from 7 weeks ago and asks you to pull the statistics history for that day. Why might the data already be gone?

Mini checks abort with error 463. Which parameter must you set first, and to what value depending on the HANA version?

Which two parameters control standard trace-file size and retention count, where are they set, and with what defaults?

You want to read a trace file's contents via SQL rather than logging onto the OS. M_TRACEFILES or M_TRACEFILE_CONTENTS — which gives you the text, and what does the other give you?

To send HANA alert emails you must configure both the sender and recipients. What are the sender's parts, and what is the default SMTP port?

A monitoring-view query shows a service only as HOST + PORT, but you need the human-readable service name. What's the join, and what's the easy mistake?

An ABAP admin with only DBACOCKPIT access needs to browse HANA system views. Where do they go, and what schema choice do they face?

You try to change HANA alert collection intervals and email setup from DBA Cockpit and can't find the screen. What's the constraint, and how do you actually send alert emails from the ABAP side?

What is the naming form of a standard (non-alert) HANA trace file, and what is the last component?

You create a new reset point on a monitoring view to start fresh, assuming the old reset point is archived somewhere. What actually happens to it?

While cleaning up a full trace directory you're about to delete kill.sap and INSTSTAT. Why must you not, and where do these live?

A monitoring view shows a counter of '0 errors' — does that mean the system has had zero errors since it was installed?

SAP's SQL statement collection is delivered as an attachment to which SAP Note, and roughly how many scripts are in it?

You can query M_CS_TABLES without writing a schema prefix, but _SYS_STATISTICS views fail without one. What underlying difference explains this?

You reset M_VOLUME_IO_TOTAL_STATISTICS but the main view's numbers don't change. Where did your reset baseline actually land?

Logged in as <hsid>adm, what does the cdtrace alias do?

A developer wants to find why a query is slow and reaches for the HANA SQL trace. Why is that the wrong tool, and what is the SQL trace actually for?

In a script you see M_CS_TABLES and HOST_RESOURCE_UTILIZATION_STATISTICS. From the name alone, which is a live (volatile) monitoring view, and how can you tell?

M_DATABASES returns only one row when you expected to see every tenant. What's wrong, and how do you fix it?

An admin scans the cockpit alert list, sees nothing, and reports 'no alerts at all'. Why might that conclusion be premature?

You try to set a reset point on a monitoring view to measure 'since 9am today', but the view has no reset companion. Why might a view not support reset points?

Which SAP Note explains how to handle the 'Check for New Crash Dump File' alert?

Monitoring-view data vanishes on restart, yet you can still review last week's metrics. What mechanism preserves the data, and what's the catch on granularity?

You need the alert firing right now, but you query STATISTICS_ALERTS and get a wall of old entries. Which view should you have used, and how do the two differ?

You want one query in the system database that reports a metric per tenant without connecting to each. Which view set helps, and what makes it special?

On a HANA system both alerting AND history collection stopped at the same moment. Which single thread/service, if impaired, explains both at once?

You want to script-enable a specific alert by its ID. Where are alert definitions stored, and what precondition stops you from acting on the wrong object?

A crash spanned several services and you have a dozen diagnostic files with interleaved timestamps. What Studio feature saves you from reading them one by one?

You're investigating a tenant issue but can't see one tenant's trace files from where you're connected. What's likely going on, and where should you connect?

You're starting to troubleshoot a service and face a giant standard trace plus a smaller alert trace. Which do you open first, and why?

A node ran out of memory and crashed. Among the dump files on disk, which filename term points you at the OOM event versus a generic runtime or emergency dump?

To stop an alert check via SQL you set its STATUS in STATISTICS_SCHEDULE. A colleague says 'just set it to OFF (or ACTIVE to re-enable)'. What are the real status values?

A symptom appears only when one application connects via a specific driver. Which three client APIs each have their own HANA client-side trace, and which one covers hdbsql / S/4HANA / NetWeaver?

In STATISTICS_CURRENT_ALERTS, what do ALERT_RATING values 1 through 5 mean?

You follow the HANA Studio online help to change alert check start times, but the configuration tab simply isn't there. Why, and what does that tell you about the help?

During a crash investigation you have the main crashdump plus several sibling dumps. Which term marks the IPMM dump, and how do you confirm it belongs to the same crash?

You're editing a collector in STATISTICS_SCHEDULE: one column changes how long data is kept, another how often it's collected. Which is which, and what's the units trap?

After a go-live a colleague says 'the system runs, so we don't need mini checks'. What do mini checks catch that a running system won't reveal on its own?

You're scripting an alert-threshold change in STATISTICS_ALERT_THRESHOLDS. Which column must you write, which must you leave alone, and why does it matter?

A support engineer hands you scripts from SQLStatements_Internal.zip to run yourself on prod. Why should you push back, and what access do they even require?

When do you actually have to install SAP's SQL statement collection on the database server itself, versus not bothering?

To configure HANA alerts entirely from the command line, which four _SYS_STATISTICS views cover status/interval, thresholds, recipients, and sender settings?

You've uploaded dozens of SQL collection scripts via DBACOCKPIT and now want them all gone. Where are they stored, and why isn't cleanup as easy as you'd hope?

A SQL collection script has 'MDC' in its name. A colleague skips it on a single-tenant box 'because MDC means multitenant'. Is that the right call?

Someone fills in the username and password fields in the SQL trace .py file to make replay easier. Why is that dangerous by default?

You're enabling the SQL trace from the command line. Which ini section and file — and what changes for the system database versus a tenant?

On a 2.0 SPS 04 system a cursor that had been idle for hours suddenly got terminated and nobody issued a KILL. What HANA feature likely did it, and what's the surprise?

Which hdbsql option runs a script from SAP's SQL collection, and why also add -m?

A colleague wants full visibility and sets the SQL trace to ALL_WITH_RESULTS on a busy production index server. What's the asymmetric risk?

Mini-check output is long and you have limited time. Which category deserves attention first, and which SAP Note helps you interpret the results?

An alert is firing and you need to know what to actually DO about it, from the alert data itself. Which column tells you, and what's its limitation?

In mini-check output, what does an 'X' in the C (criticality) column mean, and what should you do about those rows?

A single business transaction crossed several systems and you need to follow its trace records across all of them. What links the records, and what do you typically need to drive the tracing?

One specific SQL statement misbehaves but the user runs many statements fine. You want extra tracing for just that statement without flooding the trace. What does user-specific tracing let you target, and what's the precise scope?

What are the five trace levels available for the HANA SQL trace?

You've tuned an alert's activation, interval, and thresholds on one tenant and now have a dozen more tenants to match. What's the faster path in the cockpit than redoing each by hand?

You add a recipient to one specific alert so a specialist gets paged for it. Afterwards the general distribution list stops getting that alert's emails. Bug or expected?

In a HANA trace message, what does the severity-level field indicate and what are its possible single-letter values?

The SQL trace produced a .py file. A teammate ignores it expecting a plain text log. What is it really, and why does the format matter?

On a multitenant HANA system you run the mini checks once against the system database and call it done. Why is the coverage incomplete?

You need to raise the trace verbosity for a HANA service. The trace is split into many components — what's the practical way to set a level broadly, and where does the database trace write?

Which SQL statement turns the HANA SQL trace ON from the command line (and how do you turn it off)?

You enabled a 'profile trace' expecting a readable performance report, but got two cryptic files and a note to send them to SAP. What does the kernel profile trace produce, and who's the audience?

Which SQL statement creates (resets) a reset point for the M_VOLUME_IO_TOTAL_STATISTICS_RESET view?

In which ini section is the HANA kernel sentinel controlled, and which SAP Note lists all its parameters?

Which SQL activates HANA alert ID 77 and sets its check interval to 30 minutes (current schema _SYS_STATISTICS)?

You're about to restart a HANA instance to clear a glitch, but a colleague is watching a live counter in a monitoring view to diagnose it. What's lost on restart, and what's the safety net?

From the system database you query M_DATA_VOLUMES expecting to see every database's data volumes since they share one filesystem. What do you actually get, and why?

You want to add an index to speed up reads from a monitoring view, or trust a value you read an hour ago. Both are mistakes — why?

Your script queries STATISTICS_ALERTS without a schema and gets 'not found', though the same bare-name style works for M_ views. What's different about statistics views?

Two SQL collection scripts share the same functional name but one ends in _2.00.040+. You're tempted to grab the higher-looking number. What's the rule, and the risk if you guess wrong?

What are the five groups of SAP HANA mini check scripts?