VERIFIED SOLUTION i
X

e2 Vault - indexerd coredump

Issue

e2 Vault - indexerd coredump

Cause

This issue was caused by a bug in indexerd in Vault on Solaris 6.1m1p0188 (and lower)

Resolution

UPDATED: October 18, 2017


build/6.1M1p0192

This build corrects a locking issue in read only indexerd instances. A commit at the read/write indexerd instance can trigger a root record reload at the read only instance while it is currently executing an operation in the root node. In some cases, the thread executing in root can see a torn or corrupt memory image of the record during such an update. This can in turn lead to a crash.

(The read/write indexerd process loads the root record at start up and maintains it in memory. Since it never reloads the root record, this issue does not occur. The read only indexerd periodically re-reads the root record to track updates made by the read/write instance.)

Environment Details

Product Feature: Server

Operating System: Solaris

Database: Not stated

Configuration: Not stated
 

Downloads

  • No Downloads