Memory leak in CKMSQuantiles.java #550
Comments
|
Can you provide more details? |
|
My application is running on Tomcat JVM. With increasing in requests/second to my application there is a memory leak in the JVM the memory is not getting released. Per my analysis the CollectorRegistry class usage causing the CKMSQuantiles$Quantile objects are not released. The Heap dump in regular intervals show the leak suspect of 95% on the class CKMSQuantiles$Quantile. |
|
Can you share graphs and code? |
|
Unfortunately I cannot share them as per the policies of my organization. But I observed it is pretty similar to the graphs of the issue #422 |
|
That wasn't an issue with client_java. I'm afraid without data, there's not much we can do. |
|
Lacking information there's not much we can do here, and it's likely this code will be replaced at some point anyway. |
DieBauer
added a commit
to DieBauer/client_java
that referenced
this issue
Jan 22, 2022
CKMSQuantiles is copied from an implementation of 2012, where it states that a ‘HACK’ was done, admitting a space leak. This leak has been noticed several times (prometheus#422, prometheus#550, prometheus#654). By correctly applying the algorithm from the paper we fix the leak. I have added unit-tests to show that the behaviour is correct. I have also added a Benchmark in the benchmark module showing the difference with the old and current implementation. According to my benchmarks, is in the new implementation a `get` of a quantile that has ‘seen’ 1 million elements 440 times faster. Inserting 1 million elements is 3.5 times faster. While going through the CKMS paper and the Java implementation I have added remarks and snippets from the paper, to clarify why certain choices are made.
DieBauer
added a commit
to DieBauer/client_java
that referenced
this issue
Jan 22, 2022
CKMSQuantiles is copied from an implementation of 2012, where it states that a ‘HACK’ was done, admitting a space leak. This leak has been noticed several times (prometheus#422, prometheus#550, prometheus#654). By correctly applying the algorithm from the paper we fix the leak. I have added unit-tests to show that the behaviour is correct. I have also added a Benchmark in the benchmark module showing the difference with the old and current implementation. According to my benchmarks, is in the new implementation a `get` of a quantile that has ‘seen’ 1 million elements 440 times faster. Inserting 1 million elements is 3.5 times faster. While going through the CKMS paper and the Java implementation I have added remarks and snippets from the paper, to clarify why certain choices are made. Signed-off-by: Jens <jenskat@gmail.com>
fstab
pushed a commit
that referenced
this issue
Jan 30, 2022
CKMSQuantiles is copied from an implementation of 2012, where it states that a ‘HACK’ was done, admitting a space leak. This leak has been noticed several times (#422, #550, #654). By correctly applying the algorithm from the paper we fix the leak. I have added unit-tests to show that the behaviour is correct. I have also added a Benchmark in the benchmark module showing the difference with the old and current implementation. According to my benchmarks, is in the new implementation a `get` of a quantile that has ‘seen’ 1 million elements 440 times faster. Inserting 1 million elements is 3.5 times faster. While going through the CKMS paper and the Java implementation I have added remarks and snippets from the paper, to clarify why certain choices are made. Signed-off-by: Jens <jenskat@gmail.com>
fstab
pushed a commit
that referenced
this issue
Jan 30, 2022
CKMSQuantiles is copied from an implementation of 2012, where it states that a ‘HACK’ was done, admitting a space leak. This leak has been noticed several times (#422, #550, #654). By correctly applying the algorithm from the paper we fix the leak. I have added unit-tests to show that the behaviour is correct. I have also added a Benchmark in the benchmark module showing the difference with the old and current implementation. According to my benchmarks, is in the new implementation a `get` of a quantile that has ‘seen’ 1 million elements 440 times faster. Inserting 1 million elements is 3.5 times faster. While going through the CKMS paper and the Java implementation I have added remarks and snippets from the paper, to clarify why certain choices are made. Signed-off-by: Jens <jenskat@gmail.com> Fix assertion in HistogramTest Median of 1..11 = 6 Signed-off-by: Jens <jenskat@gmail.com> Address PR remarks Signed-off-by: Jens <jenskat@gmail.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
This issue is more related to Memory leak in CKMSQuantiles.java? #422
I am getting this issue on a CKMSQuantiles$Quantile
The text was updated successfully, but these errors were encountered: