we switched one of our production server systems from ASA 9 to ASA 12 (64 bit). The database is running on one VM and there are three other VM's running EAServer that are connecting to the database using the ASA 12 32 bit odbc driver using TCP/IP. Liveness timeout is set to 0 and idle timeout is set to 240. We are getting random disconnects of the connections from the EAServer processes. The time between these is much less than the 240 minutes in the idle timeout. The majority of the time it runs fine, but we get these disconnects about every 30 minutes on each server. They do not happen at the same time on all servers. We see the disconnects as the EAServer is processing a query and gets an error back stating the the connection was terminated cm__odbc_error: SQLError: -308, SQLState: 08S01, [Sybase][ODBC Driver][SQL Anywhere]Connection was terminated. Any ideas? |
We are aware of one issue affecting heavily loaded servers and disconnecting clients:
We are also aware of other bugs in the database server that may cause sporadic connection drops:
We have also had sporadic reports of TCP/IP "localhost" connections dropping in some virtualized environments, although we strongly suspect these are related to underlying OS issues with name resolution as changing the client connection to use "127.0.0.1" as the target address for the server have resolved the issue. Other timeouts or disconnections would currently be undiagnosed and would likely require additional investigation via a technical support case. |
I neglected to post the error we are getting in the database server log Disconnected TCPIP client's AppInfo: IP=208.140.4.32;HOST=MDAM-2;OSUSER=SYSTEM;OS='Windows 2008R2 Build 7601 Service Pack 1';EXE='C:Program Files (x86)SybaseEAServerbinjagsrv.exe';PID=0x7b8;THREAD=0x2fc;VERSION=12.0.1.3152;API=ODBC;TIMEZONEADJUSTMENT=-300 I. 10/31 10:28:50. Connection terminated abnormally
We also get this error Disconnected TCPIP client's AppInfo: IP=208.140.4.32;HOST=MDAM-2;OSUSER=SYSTEM;OS='Windows 2008R2 Build 7601 Service Pack 1';EXE='C:Program Files (x86)SybaseEAServerbinjagsrv.exe';PID=0x878;THREAD=0x8e8;VERSION=12.0.1.3152;API=ODBC;TIMEZONEADJUSTMENT=-300 I. 10/31 10:07:41. Connection terminated abnormally; client socket shut down
When you say SQLError I assume you mean SQLCODE. FWIW the docs don't list -308 together with 08S01, and 08S01 is not listed as a SQLSTATE at all; it is listed as ODBC 2 State and ODBC 3 State. Maybe running dbsrv12 with -z will shed some light on what is happening... more than the Help for SQLCODE -308 http://dcx.sybase.com/index.html#1201/en/saerrors/errm308.html
A quick Google leaves one with the impression "it could be just about anything" :)
FYI, running the same configuration with ASA 9 32 bit did not have these dropped connections.
The errors are cut and pasted out of our error logs. NO editing was performed.
This appears to only be happening under heavy loads.
If you have heavy loads, why are you running the database on a VM? ...just asking.
Have you tried to use a newer EBF? - AFAIK, 12.0.1.3152 was the first 12.0.1 release (MR) so some fixes might be available...
Heavy is a relative term. At low transaction volumes, we don't have an issue. At high volumes the database server hits 50% cpu.
Applied the latest EBF (12.0.1.3484) to one of the two servers that we have upgraded. It had one dropped connection so far today. This is a ASA12 32 bit server using Shared memory
Rolled the ODBC driver on the other server (The one with three EAServers connecting to it) back to the ASA 9.0.2.3508 driver. Not one dropped connection since doing this.
From the server that we turned the -z option on for.
I. 11/01 06:42:12. TCP/IP: Received connection request from 66.18.108.32:49696. I. 11/01 06:42:12. Connection from 66.18.108.32:49696 assigned connection ID 59 I. 11/01 06:42:12. Connection ID 59: "xxx" connected to database "emr" from SQL Anywhere 12.0.1 (3152) client over TCPIP I. 11/01 06:42:12. Connection ID 59: AppInfo is IP=208.140.4.32;HOST=MDAM-2;OSUSER=SYSTEM;OS='Windows 2008R2 Build 7601 Service Pack 1';EXE='C:Program Files (x86)SybaseEAServerbinjagsrv.exe';PID=0xbe8;THREAD=0x1ec;VERSION=12.0.1.3152;API=ODBC;TIMEZONEADJUSTMENT=-300 I. 11/01 06:42:12. Connection ID 59: Server character set conversion is enabled, but not required using character set "windows-1252"
I. 11/01 07:26:17. Connection ID 59: connection "XXX" terminated abnormally I. 11/01 07:26:17. Connection ID 59: Disconnected TCPIP client's AppInfo is IP=208.140.4.32;HOST=MDAM-2;OSUSER=SYSTEM;OS='Windows 2008R2 Build 7601 Service Pack 1';EXE='C:Program Files (x86)SybaseEAServerbinjagsrv.exe';PID=0xbe8;THREAD=0x1ec;VERSION=12.0.1.3152;API=ODBC;TIMEZONEADJUSTMENT=-300 I. 11/01 07:26:17. Connection ID 59: "XXX" disconnected from database "emr"
I'm somewhat puzzled: Are these log excerpts from the server with the dropped connection that should be using Shared memory (as you state in the first paragraph? - That would be irritating as the connection is using a TCP/IP link...
FWIW, there is a - possibly connected - FAQ with a short discussion on how Windows Server 2008 R2 might prevent local processes from using Shared memory links...
The log excerpts are from the servers using tcpip. Did not turn on the -z option for the one running shared memory.
None of these issued showed up in our development and test environments
The problem we have now is that things are working and we are reluctant to change what is in production. Will endeavor to reproduce in a test environment so we can then try different options (once we determine what those options might be)
I am having this EXACT issue on the latest EBF (12.0.1.3492). Abnormal disconnects on a VM 2008 Server (64 bit) under heavey loads so they only show up in production. Server messages complain the client disconnected and ODBC log complains server disconnected. Logs look the same as Alex's posts. Issue happens when connecting using tcpip or shared memory. For now we've set the idle timeout and liveness timeout to 0 in ODBC connection parameters and set the -ti 0 -tl 0 on the server parameters but it doesn't seem to be helping.
If anybody has any information we could really use the help.
12.0.1.3492 is not the latest EBF for Windows x86/x64 - 12.0.1.3967 is. See here to download the EBF. I would recommend that you apply this patch to see if the issues I have mentioned in my answer are resolved.
Jeff I misspoke, we have loaded EBF 3942. Since even that's not the latest I will try the EBF you mentioned above. Thanks!
hi,
may i know have this problem resolved? i think i also having the similar problem here using SQLA ver 10
Please ask a new question. The symptom "Connection terminated abnormally" has an almost infinite number of possible causes. There haven't been any fixes for Version 10 for a long time (nor will there be any) so the bug fixes described in this conversation likely won't help you. What MIGHT help is a new question, describing EXACTLY what symptoms you are seeing... just because there won't be any new fixes doesn't mean there isn't any support... and it is entirely possible you problem has a solution that doesn't involve a bug fix... but no one will know until you tell us what's going on.
To recap: Ask a new question, do not post a comment here.