How to configure PerfMon plugin to visualize CSV data?

[KingRabbid]

Hi,

I cannot visualize properly the data saved into CSV by PerfMon plugin.

The timestamps are completely out of scale, KPIs are not sync'ed; below a sample for a 30 seconds capture.

Sample CSV file content for the picture:

timeStamp,elapsed,label,responseCode,responseMessage,threadName,success,bytes,sentBytes,grpThreads,allThreads,Latency,Hostname,Connect

1620856972279,84593,10.8.96.90 Memory Memory,,,,true,0,0,0,0,0,CCH,0

1620856890199,2512,10.8.96.90 CPU CPU,,,,true,0,0,0,0,0,CCH,0

1620856928688,41000,10.8.96.90 TCP TCP-established conn,,,,true,0,0,0,0,0,CCH,0

1620856972437,84749,10.8.96.91 Memory Memory,,,,true,0,0,0,0,0,CCH,0

1620856930689,43000,10.8.96.91 TCP TCP-established conn,,,,true,0,0,0,0,0,CCH,0

1620856888820,1131,10.8.96.91 CPU CPU,,,,true,0,0,0,0,0,CCH,0

1620856973312,84620,10.8.96.90 Memory Memory,,,,true,0,0,0,0,0,CCH,0

1620856897349,8657,10.8.96.90 CPU CPU,,,,true,0,0,0,0,0,CCH,0

1620856929692,41000,10.8.96.90 TCP TCP-established conn,,,,true,0,0,0,0,0,CCH,0

1620856973446,84754,10.8.96.91 Memory Memory,,,,true,0,0,0,0,0,CCH,0

1620856928692,40000,10.8.96.91 TCP TCP-established conn,,,,true,0,0,0,0,0,CCH,0

1620856889379,687,10.8.96.91 CPU CPU,,,,true,0,0,0,0,0,CCH,0

1620856974313,84614,10.8.96.90 Memory Memory,,,,true,0,0,0,0,0,CCH,0

1620856909310,19611,10.8.96.90 CPU CPU,,,,true,0,0,0,0,0,CCH,0

1620856930699,41000,10.8.96.90 TCP TCP-established conn,,,,true,0,0,0,0,0,CCH,0

1620856974454,84754,10.8.96.91 Memory Memory,,,,true,0,0,0,0,0,CCH,0

1620856929700,40000,10.8.96.91 TCP TCP-established conn,,,,true,0,0,0,0,0,CCH,0

1620856890890,1190,10.8.96.91 CPU CPU,,,,true,0,0,0,0,0,CCH,0

1620856975339,84623,10.8.96.90 Memory Memory,,,,true,0,0,0,0,0,CCH,0

1620856911356,20640,10.8.96.90 CPU CPU,,,,true,0,0,0,0,0,CCH,0

1620856931716,41000,10.8.96.90 TCP TCP-established conn,,,,true,0,0,0,0,0,CCH,0

1620856975471,84754,10.8.96.91 Memory Memory,,,,true,0,0,0,0,0,CCH,0

1620856930717,40000,10.8.96.91 TCP TCP-established conn,,,,true,0,0,0,0,0,CCH,0

1620856891396,679,10.8.96.91 CPU CPU,,,,true,0,0,0,0,0,CCH,0

1620856976347,84621,10.8.96.90 Memory Memory,,,,true,0,0,0,0,0,CCH,0

1620856910231,18505,10.8.96.90 CPU CPU,,,,true,0,0,0,0,0,CCH,0

1620856932726,41000,10.8.96.90 TCP TCP-established conn,,,,true,0,0,0,0,0,CCH,0

1620856976471,84745,10.8.96.91 Memory Memory,,,,true,0,0,0,0,0,CCH,0

1620856931727,40000,10.8.96.91 TCP TCP-established conn,,,,true,0,0,0,0,0,CCH,0

1620856894286,2559,10.8.96.91 CPU CPU,,,,true,0,0,0,0,0,CCH,0

1620856977367,84624,10.8.96.90 Memory Memory,,,,true,0,0,0,0,0,CCH,0

1620856910650,17907,10.8.96.90 CPU CPU,,,,true,0,0,0,0,0,CCH,0

1620856932743,40000,10.8.96.90 TCP TCP-established conn,,,,true,0,0,0,0,0,CCH,0

I'm using JMeter 5.4.1 with latest plugin version.

What do I need to configure to have things right?

Thank you 

[DT]

Your CSV looks weird as the results are in not chronological order, make sure to use a single PerfMon Metrics Collector listener, it can be configured to display/store multiple metrics, just choose what you want to store by adding a new row.

If you have several files from multiple hosts (although I believe one single listener on one single host should be sufficient for collecting server-side metrics) - it might be a good idea to use Merge Results Tool (can be installed using  JMeter Plugins Manager)

Alternatively you can use LibreOffice Calc or MS Excel or equivalent to sort your CSV by the timestamp (first column) ascending, this way you should get at least chronologically compliant results in the chart

[KingRabbid]

I have a single PerfMon listener in the JMX which collects 3 KPIs from 2 servers. It works fine in GUI mode but when I load the CSV file it shows what I've previously attached to the comment.

I don't see how sorting the data can improve things since it shows timestamps that are not related to the test moment (you can compare the pictures and timestamps displayed on the graphs).

Note: I haven't configured anything in the properties files or passed on the command line related to the plugin. The agent running on the servers are configured to collect KPIs every second.

Thank you

[KingRabbid]

Hi,

any suggestion? Do I need to configure something different? Any system property needed?

Thank you

[KingRabbid]

Hi,

repeated trials and same strange behavior; for a 5 min run the data shows KPIs being collected at different moments in time:

sort -g CPU_Memory_usage_report_2021-05-25-17-39-04.jtl | head -n 100

timeStamp,elapsed,label,success,grpThreads,allThreads

1621953546048,1343,10.8.96.90 CPU CPU,true,0,0

1621953546225,1443,10.8.96.91 CPU CPU,true,0,0

1621953546998,1211,10.8.96.91 CPU CPU,true,0,0

1621953547465,675,10.8.96.90 CPU CPU,true,0,0

1621953548195,2409,10.8.96.90 CPU CPU,true,0,0

1621953548305,1512,10.8.96.91 CPU CPU,true,0,0

1621953548820,1020,10.8.96.91 CPU CPU,true,0,0

1621953549311,508,10.8.96.90 CPU CPU,true,0,0

1621953549658,1861,10.8.96.90 CPU CPU,true,0,0

1621953549996,1192,10.8.96.91 CPU CPU,true,0,0

1621953550648,838,10.8.96.91 CPU CPU,true,0,0

1621953551866,1011,10.8.96.91 CPU CPU,true,0,0

1621953553398,3589,10.8.96.90 CPU CPU,true,0,0

1621953559223,8404,10.8.96.90 CPU CPU,true,0,0

1621953568180,14310,10.8.96.91 CPU CPU,true,0,0

1621953569223,17361,10.8.96.91 CPU CPU,true,0,0

1621953569328,16464,10.8.96.91 CPU CPU,true,0,0

1621953571754,16873,10.8.96.91 CPU CPU,true,0,0

1621953573575,21716,10.8.96.90 CPU CPU,true,0,0

1621953573689,19821,10.8.96.90 CPU CPU,true,0,0

1621953574033,18149,10.8.96.91 CPU CPU,true,0,0

1621953576246,19360,10.8.96.91 CPU CPU,true,0,0

1621953577482,22603,10.8.96.90 CPU CPU,true,0,0

1621953578139,20250,10.8.96.91 CPU CPU,true,0,0

1621953578326,21441,10.8.96.90 CPU CPU,true,0,0

1621953579113,23230,10.8.96.90 CPU CPU,true,0,0

1621953580400,19499,10.8.96.90 CPU CPU,true,0,0

1621953580410,22522,10.8.96.90 CPU CPU,true,0,0

1621953580661,27797,10.8.96.90 CPU CPU,true,0,0

1621953581026,21129,10.8.96.90 CPU CPU,true,0,0

1621953581618,22727,10.8.96.90 CPU CPU,true,0,0

1621953584535,23633,10.8.96.91 CPU CPU,true,0,0

1621953584706,40000,10.8.96.90 TCP TCP-established conn,true,0,0

1621953584935,22024,10.8.96.91 CPU CPU,true,0,0

1621953585202,21288,10.8.96.91 CPU CPU,true,0,0

1621953585786,40000,10.8.96.90 TCP TCP-established conn,true,0,0

I've run TCPDUMP on both monitored nodes and I see no missing data for any of the KPIs, the agents are reporting correctly the information, KPI values are in the same data range so no possible side-effect from value format/range/etc.

I'm running some performance/stability tests with ~70 threads in 5 Thread groups, could this impact the responsiveness of the listener? There is only one additional SimpleDataWriter that dumps test case samples into a CSV JTL file, nothing else.

Thanks