System Monitoring & Troubleshooting

The SAP system is running but you can't log on to it at all, so every in-system transaction is off the table. Which OS-level tool do you use for the message server versus the gateway versus the work processes?

The database won't come up and neither will the SAP system, so no SM21 or ST22. Where do the developer traces live, and why are they the right thing to read in this exact situation?

You accidentally deleted a local system log file at the OS level. You recreate it or restart, but SM21 still misbehaves. What did deleting the file alone leave behind, and how do you fix it?

You need to start and stop several instances across the landscape from one place and manage operation modes. Which transaction, and what's the OS-specific catch that bites you on UNIX?

A security review asks you to prove that operating-system and database accounts are visible in the global user overview AL08. Why can't you deliver that from AL08?

A colleague opens SM13 to hunt for a user's failed update from last week, sees an empty list, and concludes there are no cancellations. What did the default selection do to them?

During normal operation, checking SM13 daily is on your list, but what three specific questions about the update system should each check actually answer?

An update got deactivated overnight and nobody noticed for a couple of hours. Why is that far more urgent than it sounds, rather than something users can just wait out?

In SM13 you have a canceled update record and the menu offers both Repeat and Update. They sound identical — which one actually retries, and what does the other do?

A problem reproduces only for one specific user and you can't see why from the system log. What can SM04 do that turns "only for that user" into actual evidence?

You just added a new application server and connections to it fail intermittently. You fixed the hosts/services files — do you have to restart the instance to pick them up?

The whole AS ABAP fails to start and you don't know if it's the database or SAP. What do you check first, and why not just relaunch via the MMC or start script?

A canceled update is holding SAP locks and a user just wants those objects freed. Why is deleting the lock entry in SM12 the wrong fix here?

If you never schedule RSSNAPDL, what happens — and how do you keep an important dump without disabling cleanup?

A user reports an error with no dump on their screen. Where do you start the analysis, and why there rather than jumping straight to a tool?

You need to change rows in a configuration table. Why reach for SM30 (or an object-specific transaction) rather than SE16, even though SE16 can show the data?

In an OS process list you see dw but the instance acts isolated. What do ms, gwrd, and co/se map to, and which absence explains the isolation?

You kill a hung work process in SM50 expecting to free a slot. Why does the process count stay the same, and when does 'Kill with Core' matter?

An instance is up but SM50 looks empty/stuck and you suspect the dispatcher or message server itself. Why won't SM50 help, and what do you use instead?

You want to set up a brand-new client. What must exist before you can copy into it or log on, and where does that live?

A work process is hung. Which two process types must you NOT cancel from SM50, and why is the cost asymmetric?

In SM50 you see several processes in 'Stops' and one in 'Finished'. Which signals a real problem, and which is normal?

You need to raise message-server tracing to debug a connection issue, or take an app server out of rotation. What lets you do it without a restart, and what's the trap people fall into?

For each work-process line in SM50, which fields do you read to triage a stuck process?

Why is checking ST22 daily a standing task, rather than something you do only when a user reports a crash?

SM51 shows one instance as 'Passive' and another as 'Stop'. Which is a graceful drain and which is genuinely disconnected?

You want one consolidated system-log view across all instances and need to size it. Which parameter, and what does the local one NOT cover?

You enabled change logging on a table in the Dictionary but no changes are recorded. What second condition is almost certainly missing?

An SM50 process shows 'stops' reason SLEEP; another shows PRIV. Which points at a resource problem, and which is benign?

The system is slow but you don't know which application server is the culprit. Which transaction finds it, and why is SM50 the wrong first move?

You broadcast an SM02 shutdown warning, then immediately stop the instance. Why might users still be cut off mid-transaction?

You go to SM21 for an error from a few days ago on a busy server, but the entry is gone. Why, given how the system log is written?

What does System > Status show on any screen, including the live-session fields?

Before maintenance you must be sure a user has no active sessions anywhere in the system. Why is SM04 insufficient, and what do you use?

You find what looks like an orphaned lock in SM12 blocking an object. Before deleting it, what two things must you verify, and why?

In SM51 one instance shows status Initial and another shows Starting. Both mean "not ready," so what's the actual difference between them?