I have been trying various methods to collect dump files recently. If I have collected enough files and conducted some basic analysis, I will organize a brief message and send it out as soon as possible. Please help me analyze and provide some suggestions. This problem has been bothering me all along, seriously affecting a batch of projects I am currently implementing

Thank you very much


Simon.
2024.6.11 17:25

Hi , support:
The following information was collected by me from one of the clients, divided into three different log documents
1 opc Ua connection device disconnection log
2 Monitoring of the current number of actual running threads in the ThreadPool of the client program during runtime
3 Tracking logs

I have paid special attention to this exception when multiple machines interrupt and reconnect. Even if only three or more machines are disconnected, the number of client threads will rapidly increase to 2000

Please help to browse the attached document at 20:00:00

Thanks very much ~

Simon
2024-05-24 17:40

Hi support:
This problem has been bothering me all along, and I have been trying to figure out how to solve it recently. Currently, I am not sure where the problem lies, and I am also trying to use exclusion methods to locate the scope

Firstly, I will use one client01 for business collection and write back, while using another console program for subscription only without adding any additional code

At the same time, we also paid attention to the serverconditionchange event and found that when a machine experiences an abnormal disconnection, a large number of logs of disconnection are reported by one machine in almost the same second.

Open tracking with serverconditionchange prompt sourced from the getendpoints() function

Is there any other way to facilitate the investigation and resolution of this difficult problem

Thank you very much

The DMP file size is large, and attachments cannot be uploaded

【About program running Info】
1. DataCollect.Client02-DMP01.zip ( OPC UA SDK 5.72.522 ) , 2024/1/23

2. DataCollect.Client02-DMP01.zip ( Update to OPC UA SDK 5.72.1044) , 2024/1/30

3. windows Event : not found any useful info

4. There are a total of 16 clients, of which 4~6 clients frequently experience abnormal stops, with an average of approximately 50-100 opcSrv connections per client

5. Several fixed clients (subscribing to and writing to the specified target OpcServer) often experience abnormal work stoppages (all thread&Task appear to have all stopped)

6. client stopped subscribe & uaClient.WriteMultipleValues(), and other task(thread), including [client.exe] 's WebAPI response、Logger task、Timer of HeartBeat
But the CPU is still floating up and down, OpcUaClient ServerConditionChange() has a small number of logs that persist for 1-2 hours, and there are no more logs afterwards

【Pay close attention to the stop target time point (approximately January 29, 2024 17:51:24) 】
1. (client06.exe) Write the 26 tag values to one of the OpcServer using function: [WriteMultipleValue]

2. No success/failure logs received and record

3. After 3 consecutive attempts (with an interval of 3500ms), a total of 6 operations were written, and there were no more logs afterwards

4 Client.exe setting: ThreadPool.SetMinThreads(150, 100); ThreadPool.SetMaxThreads(2000, 1000); Unlocking the client's DMP file reveals over 2000 tasks

【Adjustment and monitoring】
1） In the code I wrote myself, task information is marked and recorded before and after [task. Run]/[TaskFactory. StartNew()]. After completing the task, cleaning and logging were carried out to facilitate recording the lifecycle of all tasks.

2）Each log record also includes: [ThreadPool. ThreadCount] [ThreadPool. PendingWorkItem Count], which will provide more reference information for monitoring

3）But this only applies to my own business code section and cannot track internal threads and tasks (such as OpcUaSDK \ third-party components)

4）After modifying the code, it has been reintroduced to the project site and is currently under observation; (As the Spring Festival approaches, in-depth testing cannot be conducted, and there is currently no more useful feedback.)

5) The DMP document is quite large, and I have uploaded two copies to the cloud storage. Here is a link for easy download and reference. I am eager to receive your help
DMP : pan.baidu.com/s/1pXT60kvKNwLN29rpTPqkiA Code: hyk6

【DMP document opening tool】: JetBrains dotMemory (2019.3)

I have conducted troubleshooting specifically for opening DMP files. Attached is the information I found about the Task using the dotMemory tool. A screenshot is attached for your quick browsing

Thanks

Hi,support :

I have encountered the following problem, please provide some suggestions for handling it

[A] Problem description
1 The OPC UA client appears to have stopped working, the application CPU is still floating up and down (0-1%), and the memory is approximately 700MB
2 All subscription data collection stopped on January 24, 2024 at 04:14, and all other worker threads stopped


Operational troubleshooting issues
1 Before closing the application, a memory mirror document (DMP) was stored separately through a process
2. Open the DMP document in the attachment, which appears to have over 2000 tasks, most of which reside in memory for a long time. Open tasks related to opcWriteValue transactions

[C] Randomly viewed a set of tasks
About 20% IPArallelismExtension+<> Task
About 80% UaSCUaBinaryChannel+Write... Task
In addition, there are about 110 tasks for Reconnector

[D] Here is our information:
1 Windows Server20216 enterprise
2 .netcore 7.05 + OPC UA SDK 5.72.522

[E]
1 Closed a batch of problem Tags
2 Fixed subscription methods for many tags, for those that do not support conditional subscriptions