-tq server option witout Shuts down

[former_SQLA_member326303]

We are faced with the following problem (SQL Anywhere Network Server Version 12.0.1.3537).
On the ASA12 database server, there are 3 services (and 3 databases).
For each of the services we set -tq dbsrv12 server option (for example, "-tq 23:40") in service command line.
Sometimes automatic shutdown for one (always for the same) service is not performed.
In this case, we see no error in the ASA12 message & Windows event logs.
For example:

• In ASA12 message log:

  I. 11/18 23:40:04. Database server reached quitting time at Fri Nov 18 2016 23:40
  I. 11/18 23:40:04. Database server shutdown requested via server console
  W. 11/18 23:40:04. TCPIP listener on IP address (::):49157 is exiting
  W. 11/18 23:40:04. TCPIP listener on IP address 0.0.0.0:49157 is exiting
  I. 11/18 23:40:04. Starting checkpoint of "SImg" (SImg.db) at Fri Nov 18 2016 23:40
  I. 11/18 23:40:04. Finished checkpoint of "SImg" (SImg.db) at Fri Nov 18 2016 23:40

  (without "Database server stopped"-message !)

• In Windows event logs no error events and no events like:

  Normal shutdown of service SQLANYs_CSS_V12.0~SImg
  SQL Anywhere - CSS_V12.0~SImg stopped

  Where can I "dig"?

[Former Member · Accepted solution]

Then that does appear to be a 'stuck' dbsrv12 process. I do suspect it could be related to some feature(s) you are using. Maybe external environments, maybe proxy tables, maybe events, ... it would be hard to guess.

My leading theory (for the range of builds you are using) would be that you are experiencing a problem related to Intra-Query Parallelism. You can test for this possibility in either of these 2 ways:

1. execute, once, from dbisql the statement: SET OPTION "public"."max_query_tasks"=1;
2. upgrade the EBF/SP to a level past where all of the parallism changes were made {build# >= 4072 and >= 4081 for Solaris}

The original Intra-query parallelism implementation could cause a hang on shutdown in some cases and the builds mentioned (#2 above) were past all the relevant corrections to that feature. The first option (#) just disables intra-query parallelism and will allow you to factor that feature out from your testing and should be preventative if your issue is related to intra-query parallelism; but can be expected to have some impact upon your performance.

[former_SQLA_member326303 · Accepted solution]

Thanks. Most likely I will suggest to the client upgrade (to SQL Anywhere 12.01.4314).
Returning to the previous council (dbsupport -...). Crash reports were not found in November:

...
SA12_20161009_234508_9008
scj_errors-2
scj_errors-3
scj_errors

[Former Member]

You could just be seeing a system failure to flush the last lines (SQL Anywhere - .... stopped) to the media. If this is what has occurred then I would not expect there to be any negative consequences due to it.

Do you have any other evidence of a shutdown failure? Do you have extra (or have to kill) dbsrv12.exe processes or reboot the system to correct for this?

There might be evidence of a crash or assertion if you run these cmdlines: dbsupport -ls dbsupport -lc {do report any submissions from the first and you will need to be in contact with Product Support for either}

[former_SQLA_member326303]

Symptoms: unable to connect to the DB on mornings. In order to get them up and running we need to end the process via Task Manager as it is unable to be stopped via SCJVIEW.

Windows event log:

Service "SQL Anywhere - CSS_V12.0~SImg" terminated unexpectedly:
SystemTime="2016-11-18T21:47:22.763115400Z" 
"param1": SQL Anywhere - CSS_V12.0~SImg 
"param2": 3

ASA12 message log:

...
I. 11/18 23:40:04. Finished checkpoint of "CMSImg" (CMSImg.db) at Fri Nov 18 2016 23:40
I. 11/19 07:47:49. SQL Anywhere Network Server Version 12.0.1.3537