PostgreSQL 8.3.7: “FATAL: could not reattach to shared memory” and “WARNING: worker took too long to start; cancelled”

问题

We record our office IP phone activity using Xima software's Chronicall, which uses a PostgreSQL backend. The server on which both of these are installed is an ESXi 5.5 VM running Windows Server Standard 2008 SP1. For some time now, we have been getting the following PostgreSQL errors in Windows event viewer:

• "FATAL: could not reattach to shared memory (key=248, addr=02510000): 487"
• "WARNING: worker took too long to start; cancelled"

These errors occur every hour or two, and always occur back-to-back in the order listed above.

Xima support has looked at the issue multiple times and has not been able to resolve it. Upon their recommendation, I have upgraded Java, disabled antivirus, and run the Windows Memory Diagnostic Tool (came back clean), but the errors persist. Xima has specifically stated that PostgreSQL should not be updated, as versions above 8.3.7 are known to cause other issues with Chronicall.

Any other suggestions to resolve this issue?

回答1:

I would say the company (Xima) is at fault here, since PostgreSQL 8.3.7 is hopelessly outdated.

Quoting the official Versioning policy of Postgres:

• Postgres 8.3 has reached EOL in Februar 2013

• Moreover, Postgres strongly recommends:

  We always recommend that all users run the latest available minor release for whatever major version is in use.

  The latest point release of 8.3 is 8.3.23.

Version 8.3.7 is just not right.
_{Running Postgres on a Windows Server 2008 VM wouldn't be my first idea either...}

来源：https://stackoverflow.com/questions/24614314/postgresql-8-3-7-fatal-could-not-reattach-to-shared-memory-and-warning-wor