TP Connect process stops

[MarcinG]

Hi,

My ConnectSynchronizationService stops responding when process memory exceeds 1,6 GB RAM, despite there is still lots of free system memory. I've noticed thet every mobile connection during scanning adds 20-40 MB used memory to tpconnect process. Connect is installed on virtual machine - Windows 10 pro, system is up to date.

 

[Dan Peluso]

Hi,
This is not a known issue so I would be curious on a couple of things.

1. What version of TP Connect and TracerPlus on the mobile device are you using?
2. How many and what are the sync processes syncing to/from? Also, what type of data?

As I mentioned, I don't think this is a known issue but I am curious on a couple of these questions.

 

[MarcinG]

Hi,

1.I'm using 10.1.0.1747 version of TP Connect and the same on Mobile devices (Zebra TC20).
2.I have 5 mobile devices. They sync to MS SQL via ODBC.

I've tested my Tracer Plus project on Windows 7 Pro, with TP Connect in trial mode, and it works well - memory usage don't grows with every scan. I thought it has something to do with ODBC drivers. I've tried several versions of ODBC drivers and there is no difference.

 

[Dan Peluso]

Hi,

OK. So that is the latest public release of TracerPlus so an upgrade suggestion would not make sense.

I don't have any reports of this as an issue. I wonder a few things.

1. Could the VM environment be the culprit? I wouldn't think so and we have a ton of customers using this setup so I expect we would have heard something.

2. Windows 10 vs 7 doesn't seem like it would be a difference either.

3. I wonder if it is related to trial mode vs registered. Meaning trial mode may just limit the amount of data/memory since you can only send/receive five records.

4. I suppose it could be related to the ODBC drivers. That would be hard to determine though I think.

What is the typical size of your data transfers? Average number of records synced?

 

[MarcinG]

I did more tests yesterday. I used my mobile phone (unregistered app) as a terminal and sync worked fine on Windows 7 and 10. Next i decided to clear cache and application data on terminals. I had to manually delete files in tracerplus10\systeminfo\system folder. I uploaded my project once again and now it works fine. TP connect service memory don't grow with every scan and it don't stops. I dont't have tpconnect errors in windows event log. Another benefit is that performance is much better. I hope it's permanent fix.

 

[MarcinG]

... memory issue continues. This time not during scan/sync process, but when starting session on mobile device. After starting session syncing works good - quick and without consuming memory, but starting session on another device consumes 125MB ram memory, and critical amount is still 1,6GB (more or less). When memory usage exceeds 1,6GB TPconnect service stops responding and then stops.

Windows event log errors (sorry for Polish descriptions)
Error No.1

Aplikacja: tpconnect_ss.exe
Wersja architektury: v4.0.30319
Opis: proces został przerwany z powodu nieobsłużonego wyjątku.
Informacje o wyjątku: System.OutOfMemoryException
w System.Threading.ThreadPool.QueueUserWorkItemHelper(System.Threading.WaitCallback, System.Object, System.Threading.StackCrawlMark ByRef, Boolean)
w System.Threading.ThreadPool.UnsafeQueueUserWorkItem(System.Threading.WaitCallback, System.Object)
w System.Threading.TimerQueue.QueueTimerCompletion(System.Threading.TimerQueueTimer)
w System.Threading.TimerQueue.FireNextTimers()
w System.Threading.TimerQueue.AppDomainTimerCallback(Int32)

Error No.2

Nazwa aplikacji powodującej błąd: tpconnect_ss.exe, wersja: 10.1.0.1747, sygnatura czasowa: 0x5c829ae5
Nazwa modułu powodującego błąd: KERNELBASE.dll, wersja: 10.0.18362.267, sygnatura czasowa: 0x91664cab
Kod wyjątku: 0xe0434352
Przesunięcie błędu: 0x00113442
Identyfikator procesu powodującego błąd: 0x1950
Godzina uruchomienia aplikacji powodującej błąd: 0x01d568a9362b0654
Ścieżka aplikacji powodującej błąd: C:\Program Files (x86)\PTS\TracerPlus Connect 10\tpconnect_ss.exe
Ścieżka modułu powodującego błąd: C:\WINDOWS\System32\KERNELBASE.dll
Identyfikator raportu: 39de2cd5-648e-4c53-86ed-2dccce0dcf7a
Pełna nazwa pakietu powodującego błąd:
Identyfikator aplikacji względem pakietu powodującego błąd:

 

[MarcinG]

I tested various scenarios today. Conclusion is that errors occur on windows 7 and 10, on virtual machine and real machine, on fresh install of win10 vm. Errors happen when TP Connect service memory usage exceeds 1,6 GB ram. Then service stops responding, cpu usage rises to 100 and 1-5 minutes later service stops. TP Connect service memory usage rises when mobile terminal starts session, previously also on every sync/scan. Memory usage increment depends on project complexity.

 

[Dan Peluso]

Hi,

If you have an active support agreement, I would recommend creating a support ticket with us. This way, we can take a more specific look into your configuration, etc

 

[MarcinG]

I get critical error on every data insert:
[11-21-19|16:29:23.061 - 10.1.0.1747 - CriticalError]: Data source 'TEST' not found. Adding as new data source. <-info from event log
Despite this, the rows are inserted into the table. What could be the reason for this error?

 

[Dan Peluso]

Hi,

That error may be expected behavior and just a poor choice of error level which causes it to post as a critical error. Basically, that error comes up the first time when loading a sync process if that data source has not been seen before and it indicates it just needs to be loaded because it is not already in memory. If the actual sync process is working well, this should be ok. I believe we have already logged this as an enhancement request to remove this over-zealous method of logging.

Do let me know if your sync is not behaving normally otherwise though.

 

[MarcinG]

Sync works, but i still have problem with memory usage. I've created new database and copied to it 3 tables which are used in my project . In this scenario memory usage looks normal. Maybe database size is the problem? I'm trying to sync directly to Microsoft Dynamics NAV database.

 

[Dan Peluso]

Hi,

I imagine a larger database size could be reflected in a larger TPConnect footprint during the time of sync (depending on your settings) but it should definitely be releasing. Even so, I am not sure if that should even show that much increase in memory. If the backend database is SQL Server via ODBC, I don't think the Microsoft NAV as the application would be a factor.

I am curious on one thing though. Have you tried making a native connection to SQL Server. This would allow a direct connection to SQL server and will not require ODBC. I don't know of any existing issues with ODBC and memory issues but it may be worth a shot going direct via SQL Server native.

 

[MarcinG]

Yes, i've tried both odbc and native sql connections. My database size is near 300GB (more then 4600 tables). I have two solutions:
1. restart tpconnect process once per hour to release memory. Disadvantage - restarting may occur during scannig
2. create separate database for tracer plus and sync it with my production database - Navision. Disadvantage - syncing two databases.

 

[Dan Peluso]

As I am sure you agree, I don't love either of these solutions for you. Since TP Connect does have some startup overhead directly related to the database size, I do feel like the 2nd option might be the better approach.

Using this approach, I wonder if you couldn't use some kind of trigger on the new smaller database to then push the same change over to your final destination db.

Let me know if that helps.

 

[MarcinG]

Second option works. It seems that TP Connect doesn't work well with large databases.
Sorry for late reply