SA12 - connection stuck: DB and applications freeze

[former_SQLA_member1694890]

In our product new version we are using SA 12.0.1.3827 Linux (previous product version based on SA 12.0.1.3519 Linux).On two customers sites after 1-2 month running appears follow problem - DB stopped allows any activities :

1. In DB output log stop appears regular lines about Starting checkpoint/Finished checkpoint
2. Only messages about 'Connection terminated abnormally' appears, in the same time applications working with DB are freezed- not crash(exit) but not real activities
3. Checking TCPIP connections status (netstat -apn | grep {db_port}, most of connections are in status CLOSE_WAIT
4. Trying connect from dbisql command-line - Communication timeout error appears
5. Trying stop DB with dbstop utility doesn't work the same error about Communication appears
6. Only 'kill -9' stop the DB. And in the same time all applications working with DB wake up and signal about B connection problem

What could be reason for this problem? Please Advise

[Breck_Carter]

Your best bet would be to contact Tech Support: http://www.sybase.com/contactus/support/

After killing the process, can they restart the server, and does it work OK after that? How long does it take before crashing again?

Historically speaking, SQL Anywhere 9 exhibited the following symptom at one financial services client site I was familiar with: The database would become unresponsive, with no activity at all, no new connections allowed. The workaround was to kill the process and restart the server. This became a regular activity, performed by IT staff without discussion, and the management and user community became used to it. Upgrading to SQL Anywhere 10 eliminated the symptom.

I am NOT saying this is happening again, to you... I am just telling you that "unresponsive server" is one of the hardest symptoms to deal with.

The good news is, Version 12 has a lot more tech support instrumentation built in, so maybe they can help.

[former_SQLA_member1694890]

After DB server kill and restart DB server and applications. The service continue to work properly. Meanwhile it doesn't happens twice on the same customers sites, but as mention it was happen on two different customers sites, after upgrade product version based on SA 12.0.1.3519 to SA 12.0.1.3827