[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17849461#comment-17849461 ] FTR commented on KAFKA-16687: - Handle personal issue, sorry late to reply. Just found more, I saw 1 line code in my before project didn't close after creating KafkaConsumer instance. I think it's caused by this. > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > Attachments: diff > > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17848688#comment-17848688 ] Philip Nee commented on KAFKA-16687: Hey [~fortherightous] - I was wondering if what I saw is in line with your observation. It seems like there are a lot of FetchMetricsManager.partitionRecordLag sensors being allocated. But the memory increase wasn't as fast as you described. Sorry for the bother - trying to identify if this needs to be hotfixed for 3.8 release. > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > Attachments: diff > > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17846313#comment-17846313 ] FTR commented on KAFKA-16687: - [~pnee], give me some time to reproduce this issue and will post here. > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > Attachments: diff > > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17846107#comment-17846107 ] Philip Nee commented on KAFKA-16687: Also, by looking at the heap. It seems like we've been getting leak earlier than 3.6 - I see the sensor class just keeps on getting more allocations. i.e. allocated objects == live object. > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > Attachments: diff > > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17846042#comment-17846042 ] Philip Nee commented on KAFKA-16687: Hey [~fortherightous] - I did a memory profile and found there's a constant increase some sensors in FetchMetricsManager. Is this inline with what you are seeing? I actually don't see much different in unsafe allocation. Could you post a snapshot of your jcmd diff here? > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17845464#comment-17845464 ] Philip Nee commented on KAFKA-16687: [~fortherightous] - Not sure if this is what you are seeing: {code:java} Native Memory Tracking (reserved=3835KB +481KB, committed=3835KB +481KB) 55 (malloc=458KB +234KB #6808 +3631) 56 (tracking overhead=3376KB +247KB) {code} and {code:java} 290 [0x0001040af538] Unsafe_AllocateMemory0(JNIEnv_*, _jobject*, long)+0xcc 291 [0x000147f01034] 292 (malloc=43KB type=Other +43KB #4 +4) 293 294 [0x0001040af538] Unsafe_AllocateMemory0(JNIEnv_*, _jobject*, long)+0xcc 295 [0x00014753d8ac] 296 (malloc=996KB type=Other -1249KB #94 -102) {code} > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17845421#comment-17845421 ] Philip Nee commented on KAFKA-16687: "Looks like it's caused by this" - mind elaborate? > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17845402#comment-17845402 ] FTR commented on KAFKA-16687: - Looks like it's caused by this. I found some Sensor classes in heap memory which I dumped. Turns good now. I moved the kafkaConsumer instance out from a class which I used it to encapsulate, and then this issue was resolved. No native memory leak now! > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17845045#comment-17845045 ] Philip Nee commented on KAFKA-16687: Since you mentioned `disabled JMX reporter` would resolve the issue. I was wondering if this is caused by leaking network connections. I have been wondering if it was caused by the sticky node used by the telemetry sender: [https://github.com/apache/kafka/commit/2b99d0e45027c88ae2347fa8f7d1ff4b2b919089] > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17845042#comment-17845042 ] Philip Nee commented on KAFKA-16687: Thank you. > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17844819#comment-17844819 ] FTR commented on KAFKA-16687: - In order to not waste your time, let me get rid of the business logic and stress test again. > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17844817#comment-17844817 ] FTR commented on KAFKA-16687: - Add -XX:NativeMemoryTracking=detail on startup option. After Java startup, take a NMT baseline: jcmd VM.native_memory baseline. ver a period of time, observe the difference: jcmd VM.native_memory detail.diff > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17844742#comment-17844742 ] Philip Nee commented on KAFKA-16687: Hey @FTR - I wrote a simple consume-produce application and probed it using jcmd. I didn't find much useful info. However, I observed the application's memory footprint increased quite fast (in GB). How did you determine " found that it's caused by [NMT Internal] memory, which created by Unsafe_allocatememory xxx." ? Thanks. > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17844579#comment-17844579 ] FTR commented on KAFKA-16687: - Update: Testing longer for 3.6.0, the resident memory of Java App proccess wasn't more than the max Heap size. But as 3.7.0, total reserved and committed Native Memory was more than OS memory, also it's caused by [NMT Internal] memory. Again, in the end, after free and buffer/cache memory was exhausted, kswapd0 process was working to swap. > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17844408#comment-17844408 ] FTR commented on KAFKA-16687: - Actually, this Kafka client is used for a web application. Just simply use jmeter to test it . > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17844400#comment-17844400 ] Philip Nee commented on KAFKA-16687: thanks for the information this is massively helpful. how do you stress test your application? Could you add a bit of information here? > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17844397#comment-17844397 ] FTR commented on KAFKA-16687: - For Kafka client, I have used KafkaConsumer and AdminClient. But for this stress test case for testing native memory leak, I just request the KafkaConsumer part. > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17844395#comment-17844395 ] Philip Nee commented on KAFKA-16687: Your application uses KafkaConsumer only right? > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17844394#comment-17844394 ] FTR commented on KAFKA-16687: - I am using stress testing to my Java application. Observing with NMT, [Internal] memory is still growing greatly ( scale GB) - native memory leak. > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17844390#comment-17844390 ] FTR commented on KAFKA-16687: - Surely, let me experiment with 3.6 and then feedback here. > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17844387#comment-17844387 ] Philip Nee commented on KAFKA-16687: But this is native memory leak so I suspect it's a different issue than heap leak. would you mind experimenting with 3.6 ? We didn't see the leak using 3.6. > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17844382#comment-17844382 ] FTR commented on KAFKA-16687: - BTW, before that I had also encountered heap leak issue in Kafka client 2.8.2. After upgraded to 3.7.0(>3.4.0) and disabled JMX reporter, and heap leak issue was resolved. > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17844381#comment-17844381 ] FTR commented on KAFKA-16687: - That's it, there are some topics with 30 partitions. I have tested Kafka client 2.8.2 before, with same issue. > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Assignee: Philip Nee >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)
[jira] [Commented] (KAFKA-16687) Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0
[ https://issues.apache.org/jira/browse/KAFKA-16687?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17844379#comment-17844379 ] Philip Nee commented on KAFKA-16687: Hey [~fortherightous] - we also observed some (possible) memory leak. The symptom is particularly obvious when working with large number of partitions. Have you tried kafka client version 3.6 or earlier? > Native memory leak by Unsafe_allocatememory in Kafka Clients 3.7.0 > > > Key: KAFKA-16687 > URL: https://issues.apache.org/jira/browse/KAFKA-16687 > Project: Kafka > Issue Type: Bug > Components: clients, consumer >Affects Versions: 3.7.0 >Reporter: FTR >Priority: Major > > I am building a Java Project which using Maven dependency Kafka-clients with > 3.7.0 version. > My Java application logic is to use Kafka Consumer to poll Kakfa broker topic > continuously. > I have configured my Java application with JVM options with -Xms8G -Xmx8G > -XX:MaxMetaspaceSize=4G, and then run it. > Also, there are 16G physical memory on my virtual machine. > After my Java application running a long time, I have found that resident > memory of the Java Process was being grown to more than 14G. > In the end, the Java process ate Swap space. > I checked it with jmap -heap pid, and found heap memory usage is Ok. > Also with Native Memory Tracking [jcmd pid Native.memory detail.diff], I > found that it's caused by [NMT Internal] memory, which created by > Unsafe_allocatememory xxx. > In my Java application, I don't use any NIO DirectByteBuffer to allocate > memory. > And I check it the Kafka-clients source code, it have codes with use > "sun.misc.unsafe" to allocate memory. And MaxMetaspaceSize not work for it . > > Could you help to check it? How could I to stop this growing native memory to > avoid my System hang? -- This message was sent by Atlassian Jira (v8.20.10#820010)