giltene / jhiccup Goto Github PK

After chatting with an Azul engineer, it was suggested I create this issue to ask for a clarification in the documentation (if no code change is needed) to describe which clocks are being used when printing otu the log line and which clocks are being used to measure pauses (System.currentTimeMillis() and System.naonTime() respectively).

We have an issue (still unsolved at this point) where there was a 35s delta between two log lines (lines before and after are 1s apart). The maximum seen pause during this iteration was only 0.688 which wouldn't seem to be possible if the log line's clock showed a diff of 35s.

I've tried use Solr with Jhiccup to analyze garbage collectors of Azul zing JVM and G1GC of Java-Hotspot. I launched solr, and want include jHiccup on it. of solr process is 1078. I used following command: ./jHiccup -l /tmp/jHiccup-2.0.9/hiccup.%date.%pid -p 1078. It should be work, but an error is displayed : A JNI error has occured, please check your installation and try again. Help me please!

Hi Gil,
Could you please guide me on the following

• How can I achieve log file rotation after the log has reached a specific size limit. I understand we can add %pid_%host_%date to the log file name. But that I assume will not get rolled over, until the application restarts?
• The intention is to run jHiccup on a application under load (high volume) running on a pod, that might have disk capacity restrictions. Could you please help me on what extra disk capacity is required assuming jHiccup runs inside such a high volume application.
• Also , if you have a rough guideline on the size of the generated log, based on duration of run, it would be great.
  Thanks in Advance,
  Divya

example :
folder name : "new folder"
files inside:

• jHiccup.jar
• orkTerminal2021.jar
• run.bat file that contain the line :
  java -javaagent:jHiccup.jar="-c" -XX:+UseZGC -jar orkTerminal2021.jar

tested on win 10 with jdk 15,16,17

changing folder name to Newfolder solved.
thanks, Or.

This would lower the barrier of entry in parsing the data with scripts (i.e. plotting the series in R would be trivial). I'll look at implementing this myself if I can get some time - if so, you can expect a pull request coming your way.

Not any of the mentioned log processing tools were able to process my logs

jHiccupLogProcessor, failed to parse logs produced with jhiccup 2.0.10:

./jHiccupLogProcessor -i ~/jhiccup.log -o jhiccup_processed.log
Log file parsing error at line number 0: line appears to be malformed.

Using this file (only gzipped to make attachment smaller):

jhiccup.log.gz

I also tried some other tools linked in the README, none of them worked either:

https://github.com/ennerf/HdrHistogramVisualizer

https://hdrhistogram.github.io/HdrHistogramJSDemo/logparser.html

http://hdrhistogram.github.io/HdrHistogram/plotFiles.html

Please provide an up-to-date list of tools that support the log format of 2.0.10 or instructions on how to convert logs to other formats.

Hi There,

This project is still using an old implementation of LGTM's automated code review, which has now been disabled. To continue using automated code review, and receive checks on your Pull Requests, please install the GitHub App on this repository.

Thanks,
The LGTM Team

What is the unit or measure for Interval_Max in hiccup.*.hlog file ?
"StartTimestamp","Interval_Length","Interval_Max","Interval_Compressed_Histogram"

Sure looks like milliseconds, but can't find it documented anywhere.

Thanks,
--Erik

This looks interesting, but graphs need to be created in something other than excel. How about adding a script using R (or some other open/platform-independent language/graphing library) to create the graphs?

Hi,

We used the following parameters on Linux Centos 3.10.0-514.6.1.el7.x86_64:

-javaagent:/opt/jHiccup-2.0.8/jHiccup.jar=-c,-d,60000,-i,1000,-l,../var/log/program-name.%date

This example should record hiccups experienced during the running of MyProgram.jar in log file hiccuplog, while at the same time recording the hiccups experienced by a control process running in a separate JVM in the log file c.hiccuplog.

The control process running in a separate JVM is not generated with JDK 9.0.4 and JDK 10.0.1.

One single file is generated:

#[Logged with jHiccup version 2.0.8]
#[Histogram log format version 1.3]
#[StartTime: 1524526871.595 (seconds since epoch), Mon Apr 23 23:41:11 UTC 2018]
"StartTimestamp","Interval_Length","Interval_Max","Interval_Compressed_Histogram"
60.007,1.000,0.033,HISTFAAAACV42pNpmSzMwMDAzAABTBDKTwEo5GawYwGD/QeIyDM+JgBnDAWw
61.007,1.000,0.049,HISTFAAAACh42pNpmSzMwMDAygABTBDKT4GBgdnNYMcCBvsPEJF7PEwdTABzmgYy

No issue with JDK 1.8_0_144: two files are generated.

Thanks for your help.

Regards,

Annabelle

Hi, I found an interesting case in this project.
It has “Thread.sleep()” in a “while(true)” loop. I found some discussion in Stackoverflow and other website. It can be refactored by using a executor service.
https://ejrh.wordpress.com/2012/07/13/sleeping-in-loops-considered-harmful/
https://stackoverflow.com/questions/3535754/netbeans-java-new-hint-thread-sleep-called-in-loop
or
Maybe use CountDownLatch.await to automatically awake and shutdown.

Detailed websites and lines:

620 | https://github.com/giltene/jHiccup/blob/master/src/main/java/org/jhiccup/HiccupMeter.java

Best regards

Using jHiccupPlotter, when generating a chart with default options and no SLA

Run-time error '-2147024809 (80070057)': The index into the specified collection is out of bounds.

This happens in the following line inside the drawIntervalChart method.

ch.SeriesCollection.Add Source:=Sheets(sourceSheetName).Range("A1")

Here is the hlog and files generated by the log processor
logs.zip

I have a use case where I'd like to track how much time is spent in "hiccup" compared to a total time taken to serve a request. What's the recommended setup?

The wrapper script example of:

jHiccup /usr/bin/java -jar MyProgram.jar -a -b -c

Doesn't work. The only way this works is if options are given to jHiccup.

For example:

$ jHiccup java -version
Error: launched with the following args:

Which was parsed as an error, indicated by the following exception:
java.lang.Exception: Invalid args:
[...]

But this works:

$ jHiccup -d 0 java -version

Looks like a bug in the option parsing where options are not, in fact, optional.

Trying to use jHiccup to monitor a Cassandra process but getting an error. Is there a way to attach? Or what are my alternative ways to monitor using jHiccup?

$ locate tools.jar
/usr/java/jdk1.7.0_76/lib/tools.jar
$ ps jfax | egrep java
1 20400 20337 20337 ? -1 SLl 397 36362:24 /usr/java/jdk1.7.0_76/jre/bin/java -ea -javaagent:/usr/share/cassandra/lib/jamm-0.3.0.jar ...
$ sudo ./jHiccup -p 20400
com.sun.tools.attach.AttachNotSupportedException: Unable to open socket file: target process not responding or HotSpot VM not loaded

Running with "-l hiccup -c" causes control thread log file to be written in / where normal users don't have privileges to write.

We're going to use statsd+graphite for monitoring, and I'd like to see data collected by jhiccup there as well. Would it be possible to send data to statsd/graphite instead of logging it to a file?

Hello,

it's not a big deal but the Readme.md mentions version 2.0.5 evnev in the 2.0.6 zip archive.

So we don't know if there are big changes between 2.0.5 and 2.0.6.

Kind regards,
Thomas

Hello all,

I've been using jHiccup on close to a 5 min interval to listen for pausing on a bare-metal linux machine.

java -jar jHiccup.jar -t 295000

Normally the pausing stays within a pretty small timeframe, but on one of the files we had a pause of 5841155.523. Accounting for it being in milliseconds this is still 5,841.15 seconds which is way over my 5 min interval the profiler runs for.

I was wondering if we might know what occurred to cause this?

Attached is the .hlog file in question along with an example of a normal run for my process.
suspect_file.csv
normal_run.csv