Talend memory leak in tHashOutput / tHashInput

I have a slow memory leak in Tal Tallet. I am using a tHashOutput / tHashInput pair in the middle of the span because I need to figure out how many lines are in the stream. So I push them to tHashOutput and then to the tHashOutput_1_NB_LINE link from globalMap.

stream of roles

I have what I think are the correct options:

  • allRows - "append" - FALSE
  • tHashinput_1 - "Clear after reading" is TRUE

However, when I run this for some time and analyze with the Eclipse Memory Analyzer, I see the objects grow over time. This is what I get 12 hours later:

memory usage

This usage (64MB / 12 hours) is growing steadily and is not related to what the job is doing (i.e. actively pumping data or just idling - and this code is also called to idle). If I look into the memory references in MAT, I can see lines that point me to this place in the code, like

tHashFile_DAAgentProductAccountCDC_delete_BPpuaT_jsonToDataPump_1_tHashOutput_2

(jsonToDataPump is the name of the new job). Am I doing something wrong in using these hash components?

+3


source to share


1 answer


I believe you should set your garbage counter to a minimum duration so that it takes care of an unused object in your application



0


source







All Articles