HDBinfo vs. HDB info: Diagnostic Tools, Full System Info Dump & Trace Analysis
A definitive operational guide disambiguating the standard process monitoring command "HDB info" from the comprehensive SAP Support diagnostic collection framework "HDBinfo / fullsysteminfo.py", detailing trace collection, crash analysis, and SAP escalation.
Advanced (Senior DBA)HDBinfo & Diagnostics~20 min studyHANA 1.0 & HANA 2.0 (All SPS)
Core Learning Objectives
- 1.Master the critical distinction between "HDB info" (process monitoring) and "HDBinfo / fullsysteminfo.py" (support diagnostics).
- 2.Learn how to locate, execute, and pass arguments to fullsysteminfo.py across HANA versions.
- 3.Analyze output archive contents (traces, OOM dumps, OS configuration, disk stats).
- 4.Execute step-by-step trace analysis on nameserver.trc, indexserver.trc, and crash dumps.
- 5.Prepare technical diagnostic packages for SAP Support ticket escalation.
Prerequisites:hana-architecture-services-overviewhana-linux-ports-services
Definition & Architectural Rationale
Technical Definition:
While "HDB info" (with a space) is a shell script displaying running instance processes, "HDBinfo" (single word, or the Full System Information Dump tool `fullsysteminfo.py`) is an SAP Support diagnostic script that packages instance configuration, runtime metrics, trace files, and OS telemetry into a compressed archive for technical support triage.
Why this component exists:
When a critical P1 production outage occurs (such as an unexplained indexserver crash, memory leak, or storage freeze), SAP Support cannot troubleshoot without a synchronized snapshot of trace files, OS telemetry, parameter files, and memory dumps. Running ad-hoc commands wastes hours; HDBinfo automates the complete diagnostic gathering process.
Internal Working & Mechanics
1. The Crucial Distinction (The "HDB info" vs "HDBinfo" Trap):
- "HDB info" (two words, space separated):
* Location: /usr/sap/<SID>/HDB<NR>/HDB info
* Purpose: Simply parses 'ps' output to list active HANA processes (nameserver, indexserver, daemon), their PIDs, user, CPU%, and memory RSS.
* Execution time: < 1 second.
* Output: Plain text terminal output.
- "HDBinfo" / "Full System Info Dump" (single word / python script):
* Location: /usr/sap/<SID>/SYS/exe/hdb/python_support/fullsysteminfo.py (or standalone support script HDBinfo in older support packages / SAP Note 1732188).
* Purpose: Executes deep diagnostic extraction: collects all .trc files, .ini files, OS kernel parameters, dmesg, /etc/hosts, memory pool statistics, topology, crash dumps, and uncommitted transaction states.
* Execution time: 2 to 10 minutes depending on log volume.
* Output: A compressed .tar.gz archive stored under /usr/sap/<SID>/HDB<NR>/<host>/trace/DB_<tenant>/.
2. Anatomy of the Full System Info Dump:
When fullsysteminfo.py runs, it extracts:
- System Configuration: All .ini files (global.ini, indexserver.ini, etc.) from SYSTEM, DATABASE, and DEFAULT layers.
- Trace Files: alert_<SID>.trc, nameserver_*.trc, indexserver_*.trc, daemon_*.trc.
- OOM Dumps: Any generated oom_*.trc files capturing call stacks and memory allocators.
- Operating System Snapshot: free -m, vmstat, iostat, numactl, uname -a, sysctl settings.
- Database Catalog Views: M_SERVICES, M_SERVICE_MEMORY, M_HEAP_MEMORY, M_SAVEPOINTS.
3. Step-by-Step Execution Workflow:
- Command: python /usr/sap/<SID>/SYS/exe/hdb/python_support/fullsysteminfo.py
- Date range filtering: To prevent creating a 50 GB archive with 6 months of historical logs, specify a timeframe:
python fullsysteminfo.py --from=2026-10-23 --to=2026-10-23
- Target database: For MDC, choose whether to collect for SystemDB or a specific tenant.
Architecture Flow Diagram
Mermaid DiagramHDBinfo vs. HDB info: Diagnostic Tools, Full System Info Dump & Trace Analysis
Architectural Flow: "HDB info" simply checks active process PIDs on the command line. "HDBinfo / fullsysteminfo.py" executes a comprehensive diagnostic harvest across traces, configs, OS stats, and crash dumps, packaging them for support escalation.
Linux Terminal Diagnostics & Operations
Execute SAP Full System Info Dump (HDBinfo Collection)
user: <sid>adm
Harvests full diagnostic snapshot of Tenant PRD for SAP Support escalation.
hdbadm@host $ python3 /usr/sap/$SAPSYSTEMNAME/SYS/exe/hdb/python_support/fullsysteminfo.py -d PRD
Terminal Standard Output (stdout):
Collecting files for database PRD... [INFO] Collecting configuration files... [INFO] Collecting trace files from /usr/sap/HDB/HDB00/hana01/trace/DB_PRD... [INFO] Collecting operating system performance metrics... [INFO] Packaging files into /usr/sap/HDB/HDB00/hana01/trace/DB_PRD/fullsysteminfo_PRD_20261023_083012.tar.gz... [SUCCESS] Collection finished. Archive size: 345 MB.
Output Analysis: Diagnostic collection completed successfully, generating a 345 MB archive containing all traces and configurations.
Search Trace Directory for Fatal Crash Stacks and OOM Dumps
user: <sid>adm
Instantly surfaces the exact line, file, and timestamp where an engine process crashed or reached OOM.
hdbadm@host $ grep -rn -E "(CRASH_STACK|OutOfMemory|Emergency shutdown)" /usr/sap/$SAPSYSTEMNAME/HDB*/$HOSTNAME/trace/*.trc | head -n 30
Terminal Standard Output (stdout):
indexserver_hana01.30003.001.trc:14205: [4890]{-1}[-1/-1] 2026-10-23 07:14:22.102 e oom oom.cc(00420) : OutOfMemory exception occurred.
indexserver_hana01.30003.001.trc:14206: [4890]{-1}[-1/-1] 2026-10-23 07:14:22.105 e oom oom.cc(00425) : Failed to allocate 142058100 bytes.Output Analysis: indexserver crashed at 07:14:22 due to an OutOfMemory exception when a query attempted to allocate an additional 142 MB beyond the Global Allocation Limit.
Troubleshooting & Root Cause Analysis
Production Best Practices & Operational Risks
Mandatory Best Practices
- Always collect the Full System Info Dump BEFORE restarting a crashed database if possible, or immediately after restart before trace files roll over.
- Never run fullsysteminfo.py without a time window filter on high-volume production systems, as zipping gigabytes of old trace files can cause CPU and disk I/O contention.