3.9.280 server is crashing on start

[2b2bff]

Hi,

after update to 3.9.280 the server keeps crashing. No dump is created and I cannot see anything in the log besides a warning that the runtime OpenSSL version does not match the compile time version.

Windows Server 2019
Microsoft SQL Server 2017

What can I do?

Cheers

[2b2bff]

Nothing? 

[Filipp Sudanov]

Does it crash right after start, or it runs for some time and then crashes?

You can try setting DebugLevel = 7 in configuration file to see, if last log entries before crash are same for several crashes.

I just tried installing .280 on Windows and it's running so far.

[2b2bff]

Thanks for getting back. This could actually help...

The last line in the log after a lot of startup tasks always is:

DCObject::updateTimeIntervalsInternal(.1.3.6.1.4.1.8072.1.3.2.4.1.2.8.103.101.116.116.101.109.112.115.{instance} [16574]): retentionTime=0, pollingInterval=0

Maybe it is that both are 0?

[2b2bff]

Sadly I'm not any further. Currently I'm trying to restore the database to the time that I upgraded to 3.9.280 from 3.9.235. But so far no luck...

I wonder what the cause is. Is it 3.9 altogether because 3.8 was rock solid for me. Or is it some weird configuration I might have?

I happen to like NetXMS for various reasons. But currently it does just nothing. And I have to way to recover from that. I don't want to switch away...

[Filipp Sudanov]

DCObject looks to be a data collection item. You can try finding it in the database in items table and setting status to 1 (disabled).
More information on the database structure is here: https://www.netxms.org/documentation/datadictionary-3.0/

Please inform, if this helps.

[2b2bff]

Hi again,

finally got time to look into this a bit more.

Disabling that DCI didn't help. I guess it's just the last line in the log, but not necessarily the reason.

I turned on crash dumps and I get a Divide by Zero at 00007FFE4513ABE7 in nxcore.dll (using 3.9.334). Does this help? I could provide the dump as well...

Cheers

[2b2bff]

Ok, after loading the dumps in Visual Studio and trying to understand them, I started to delete all of the affected DCIs, one by one. Seems like they all did belong to the same node, that was not important to monitor. So I decided to delete the node in the database alltogether.

Now NetXMS finally started again and is currently gathering all the data of the last weeks.

[Filipp Sudanov]

Could be related to this:
https://track.radensolutions.com/issue/NX-2168

[victor]

This bug should be fixed in 3.9.361.

Best regards,
Victor