Network DB service stops without any errors

[former_member329524]

Hello, all

This has just started happening on client's site:

The DB service (Sybase 16 latest build) terminates unexpectedly (quote from the event viewer) without leaving any trace of errors.

There is nothing in the console log with debugging info, nothing in the transaction log, nothing in the application event log either.

Can we be sure that something outside the service kills the dbsrv16?

Can there be a crash dump without any record in the event viewer of the failure?

[Former Member]

Hi Arcady

I presume that you are using "-o" option in the DB Service to capture the log.

If you can share the minor version number of the SQLA16 build, it will be helpful.

Generally, I noticed that this kind of problem only happen when there is assertion errors due to corrupted page file etc.

Thanks

Ramendra

[former_member329524]

Hello, Ramendra

They are using build 16.0.0.2043

Regarding what you said about the assertion failure: it is really troublesome that the DB server can crash without telling the OS that it crashed.Even if the reason is unknown, a record in the Application event log "dbsrv16.exe crashed with unknown reason" would eliminate the need for complete security sweep of the network.

Right now that have 10 DB's installed on various machines, with identical schema. 6 of them, located on different storage units crash at the same time, and 4 others do not crash at all.

How do you recommend to proceed. We have to, at the very least, to know if this is an internal problem or an outside attack.

[Former Member]

Hi Arcady

As you mentioned, 6 DBs crash at same time,

As a troubleshooting, can you check if there is any DB Events calling some process which is failing due to some particular Data condition.

Can you also share what all parameter you are using while starting the DB Service?

Thanks

Ramendra

[former_member329524]

Hi Ramendra

You will need to be much more specific than that when asking for failing events. Events in the DB do not fail. There is no log of errors anywhere in the system. It is very difficult to work, when you cannot even tell if the db server crashed because of the internal problem, or someone did "end task" in task manager.

The start up parameters are:

//----------------------------------------

-n server_name

file_name

-z -o log_file_name

//--------------------------------------

[former_member188493]

I agree it is frustrating when SQL Anywhere vanishes without a trace.

16.0.0.2043 is not ancient, but it's not a recent build either.

If six separate instances of dbsrv16.exe all crash at the same time, it is extremely unlikely to be an internal problem... I can't even imagine how one would accomplish such a feat even if it was the desired outcome.

It is unlikely that a conversation on this forum will result in a solution; you need interactive assistance from SAP tech support, or (perhaps even better) hands-on assistance from someone experienced in low-level debugging.

...or, some out-of-the-box thinking, "How would I do this thing if I wanted it to happen?" ( maybe there's an "Ask A Hacker" forum somewhere : )

[former_member188493]

> I noticed that this kind of problem only happen when there is assertion errors

Lucky you : )

[Former Member]

Hi Arcady

As you mentioned all  6 DBs services are closing at the "same time", which means there is some event occurring on all these machines at the "same time" (internal or could be external) which is closing the DB engines.

I do understand that Event do not fail but if this event is calling some function/stored proc. and they are calling any external DLL or data, which could be one potential reason.

As I don't have the DB, I can only guess and think of possible problems.

Please ignore my comments if I am not making sense, the idea was to help if possible.

Thanks

Ramendra

[lucjan_chmura]

Hi Arcady,

download and start dbsrv16 using "procdump" and see if you get the dumps created this way. Once you have a dump you can work with SAP support.

e.g.

D:\sysinternals>procdump -ma -x D:\dumps "%sqlany16%\bin64\dbsrv16.exe" -z -o cons.log D:\v16_crash.db

will start dbsrv16 and create a dump in "D:\dumps" directory when dbsrv16 crashes.

Another option (could impact performance ) , would be to run server with request logging (e.g. -zr ALL -zo req.log ), then check the log after it crashes for the last statement that was executed. This may give you some clues on what is causing this.

You could also look at upgrading the server to the latest and see if problem goes away. The latest 16.0.0 available is 2305. There is a good chance this has been resolved.

Thanks,
Lucjan