Ora-2049 vs ora-12170

Question

We have a 3 node RAC on AIX and have several dbms_jobs running that read and write to a single remote database. Occasionally, but not often, we encounter ORA-2049 timeout: distributed transaction waiting for lock with only one of the jobs. It's the same job/package each time the error occurs.
To troubleshoot, we migrated this environment to a 2 node RAC (less capable hardware) and now we are getting ORA-12170 TNS:Connect timeout occured on several of the jobs. The error is frequent at times and infrequent to non-existent at other times. ORA-2049 is extremely infrequent as compared to the ORA-12170 error.
All DBs are running 10gR2.
The DBA says this is a result of the use of db links and synonyms but my first thought is that these seem to be two different errors. We use a mixture of both public and private db links as well as public and private synonyms to access the remote database. The jobs also use dynamic SQL to execute DML and in some cases triggers to access the remote db. Is our use of links/synonyms/triggers the source of the problem or is there another configuration issue we should be looking at?
Does anyone know of a good reference for the use of synonyms, db links, and triggers with respect to do's and don'ts?
Any insight is greatly appreciated.
Thanks!
Edited by: user10394376 on Feb 23, 2010 12:12 PM

Answer 1

Yup, simplified as in changed. But still exactly the same tables and columns, just realised I was doing an unnecessary join. When I changed the predicates it worked.