The COS core daemon now keeps running when it receives a start or stop notification for a node that was already removed from a process group.

Details

Detail name Value
Changelog Number 27851
Type Bug
Status Resolved
Affected Versions Exasol 8.29.0, Exasol 2025.1.0, Exasol 2025.2.0
Fix Versions Exasol 2025.1.17
Resolution Date 2026-11-12

Description

When the last node of a COS process group was removed while a start or stop notification for that node was still being processed, the COS core daemon (cored) tried to access a node that no longer existed and aborted with SIGABRT. Because cored is the parent of all cluster processes on the node, this also stopped the COS services running on that node.

Workaround

None. The issue depends on the timing of internal cluster messages and cannot be avoided through configuration.

Fix

Before processing a start or stop notification, cored now checks that the referenced node still exists in the process group. Notifications for nodes that have already been removed are discarded with an error message in the log, and cored continues to run normally.