Diagnostic Tools: Tips on diagnostic tools / utilities / commands to troubleshoot complex problems

Document created by hdevanath Employee on Jul 2, 2017
Version 1Show Document
  • View in full screen mode


Two of the most commonly used diagnostic tools are guts and mapr-support-dump which are both sourced by MapR. However, several other open source and Linux native troubleshooting tools exist that are very useful. The word-cloud below attempts to capture the more popular tools with the font size. The table below it, outlines the description and other details for each tool. MapR Support engineers highly recommend the use of these tools for collecting useful diagnostic information to solve common problems.



 

ToolDescriptionPurposeSourceUsed for
gdb  Used to debug a running process or debug a generated core fileDiagnosticOpen SourceProcess
gstack  Used to generate a stack trace from a running processDiagnosticOpen SourceProcess
jstat  Used to print Java heap utilization and memory statisticsDiagnosticOpen SourceProcess
jstack  Used to generate a stack trace from a running JVMDiagnosticOpen SourceProcess
iostat  Used to print IO utilization stats from the Linux kernelDiagnosticOpen SourceIO
vmstat  Used to print memory utilization stats from the Linux kernelDiagnosticOpen SourceMemory
mpstat  Used to print CPU utilization stats from the Linux kernelDiagnosticOpen SourceCPU
top  Used to print per process resource utilization stats from the Linux kernelDiagnosticOpen SourceMem, CPU
sar  Used to print resource utilization stats from the Linux kernelDiagnosticOpen SourceALL
iperf  Used to benchmark network bandwidth on a hostBenchmarkOpen SourceNetworking
rpctest  Used to test and benchmark MapR's RPC layer between hostsDiagnostic, BenchmarkingMapRNetworking
guts  Used to print MapR internal metricsDiagnostic MapRProcess
RWSpeedTest  Used to benchmark read/write performance to MapR-FS from a clientDiagnostic, BenchmarkingMapRNetworking, IO, MapR
mapr-support-dump Used to collect MapR and system logs from a host and collect system & cluster diagnosticsDiagnostic MapRProcess
TPCH or TPCDS benchmarkfor SQL performanceBenchmarkIndustry StandardALL
Parquet-toolfor detect parquet file schemaDiagnosticIndustry Standard 
Netstat  Used to print network resources in use from the Linux kernelDiagnosticIndustry StandardNetworking
Tcpdump  Used to capture network traffic on a host interfaceDiagnosticIndustry StandardNetworking
Lsof  Used to print open files from the Linux kernelDiagnosticIndustry StandardProcess
iotop Used to display top disk I/O events by process from the Linux kernelDiagnosticOpen SourceIO
top -H -p Used to print thread specific resource utilization for a particular processDiagnosticOpen SourceMem, CPU
kill -3 Used to send SIGQUIT to a process to terminate and generate a coreDiagnosticOpen SourceProcess
kill -11 Used to send SIGSEGV to a process to terminate and generate a coreDiagnosticOpen SourceProcess
gcore Used to generate a core file from a given processDiagnosticOpen SourceProcess
pmap Used to report the memory map for a given processDiagnosticOpen SourceMemory, Process
YCSBUsed as a database benchmark, particularly for MaprDB and HBase Benchmark, DiagnosticOpen SourceALL
DFSIOUsed as an IO benchmark in distributed filesystemsBenchmark, DiagnosticOpen SourceIO
TERASORT/TERAGENUsed as a compute layer benchmark (map-reduce/YARN) in distributed filesystemsBenchmark, DiagnosticOpen SourceALL
Bonnie, IOZoneUsed for benchmarking Posix & NFS filesystemsBenchmark, DiagnosticOpen SourceALL
fioUsed for benchmarking  - disk & filesystem [ Posix & NFS ]  Benchmark, DiagnosticOpen SourceStorage & FS
1 person found this helpful

Attachments

    Outcomes