|
This last weekend we upgraded our Mobilink Sync Server to the latest v 17.0.0.1359 from the latest V 12 (been running v12 for years without a problem). We had tested it in a dev environment and all seemed fine. But now that it's getting production volume it crashes within minutes of starting it. For now we started back up the V12 server until we can figure out what's going on. Below are the details of this environment: Mobilink Server: V17.0.0.1359 Consolidated Server: MS SQL Server 2008 R2 Remote: ultralite iOS v12 Protocol: TCP - E2EE RSA Additional Details: This mobilink server syncs some binary columns along with typical db columns (varchar, int, guid...) (we have another environment which syncs non-binary tables on V17 in production and so far no issue) The mobilink server crashes so hard that we don't even know where to start to get a log to help you guys out. Any ideas? |
|
The MobiLInk Log is almost the best place to start in debugging any issue. In the case of a crash, we may at least be given an idea what the ML Server was doing at the time. If the process is crashing, by default, mini-dump files will be placed in "%ALLUSERSPROFILE%\SQL Anywhere 17\diagnostics" and will start with "MLS". The file name also indicates the time that the crash occured. These files can be submitted automatically using the dbsupport tool, and you can post the submission ID here, or althernatively you could simply ZIP up the crash logs and dumps files and associated Mobilink Server Logs from the time of the crash to the forum. Reg Here is my dbsupport setup: [Error Reporting] Host=sqlasupport.sap.com Port=80 Service=/projects/submission/sa_submission StatusService=/projects/submission/sa_status [Autosubmit] PromptLevel=autosubmit LastSAutoSubmit=1460852462 ResponseRetryCount=10 RetryCount=10 RetryDelay=30 So it looks like autosubmit is turned on. I don't see ane MLS logs in the folder but I assume once they're submitted they get cleared. Below is what I see for my submissions: [2418886] Date=Sun Apr 17 00:19:23 2016 [2419095] Date=Sun Apr 17 12:22:51 2016 [2419096] Date=Sun Apr 17 12:22:51 2016 [2419997] Date=Mon Apr 18 12:36:37 2016 [2419998] Date=Mon Apr 18 12:36:37 2016 So it looks like you guys got the errors so i'll just hang tight and continue to use ML12 until a fix can be issued. Thanks!
(18 Apr '16, 11:44)
Jason MS
None of those seem to be a valid or complete ML dump. Anything else of interest in your "%ALLUSERSPROFILE%\SQL Anywhere 17\diagnostics" directory? What did the MobiLink server console log show around the time of the crash?
(18 Apr '16, 16:24)
Nick Elson S...
2
While MobiLink version 12 was tested with older versions of MS SQL Server, version 17 was only ever tested with the Microsoft Native Clients from Microsoft SQL Server 2012/2014 If you are using untested ODBC drivers, then the crash might be related to those older drivers and you might to see if switch to the newer drivers makes any difference.
(18 Apr '16, 17:15)
Nick Elson S...
Replies hidden
...although that should have been noticed in the "dev environment" Jason speaks of in the OP (assuming it has had an identical setup)...
(18 Apr '16, 20:00)
Volker Barth
Good call Nick. I found the component docs that states ML17 only supports 2012-2014 as a consolidated. We'll upgrade to v16 as it's the latest version SQL Server 2008 is supported as a consolidated. Thanks!
(18 Apr '16, 20:41)
Jason MS
Hopefully that is all there is to this. If not, hopefully you will be able capture more context about the nature of the crash.
(19 Apr '16, 08:57)
Nick Elson S...
Replies hidden
It's too bad there were no crash dump files, as we would have been able to pull a stack trace from the dump file and determine whether the crash was in the MSS ODBC driver. If the crash was in the ODBC drvier, it's also possible that the crash in the ODBC driver only occurs under production load, which could explain why it wasn't seen in the dev environment.
(19 Apr '16, 09:27)
Reg Domaratzki
So 16 crashed too, just not as hard as 17. When I hopped onto server errors were flying in the console but in a panic I just shut it down and said ok to the send diagnostics dialog. Started v12 back up to take care of our users. Diagnostics show the following: [2421691] Date=Tue Apr 19 13:55:21 2016 [2421692] Date=Tue Apr 19 13:55:22 2016 I just started back up the v12 mobilink server. I'm going to have to spend some time in our dev environment to see if I can recreate the circumstances that's been bringing down 16 & 17...for now we'll just hold on to the v12 workhorse that's been running for years without issue :)
(19 Apr '16, 15:46)
Jason MS
|
