OSB BACKUP JOB FAILS WITH "COMMUNICATIONS FAILURE WITH CLIENT" after update to a later version of Oracle Linux 8.x (Doc ID 3014961.1)

Last updated on APRIL 08, 2024

Applies to:

Symptoms

The goal of this document is to provide a resolution to the issue in which Oracle Secure Backup exhibits wide scale backup failures evidenced by an error “COMMUNICATIONS FAILURE WITH CLIENT" appearing in the job transcripts. The issue has been correlated with an upgrade to a higher version of Oracle Linux 8 (either from OL7 or a lower version of OL8)

The problem has hit multiple Oracle Secure Backup domains. The issue occurs when Oracle Linux has been upgraded with affected sssd package as part of an Oracle Linux 8 upgrade. The upgrade can either be from a lower version of Oracle Linux or between versions of OL8. Downgrading/upgrading the sssd package to an older or newer version solves the issue, as does the fix outlined in the solution described in this MOS note.

The first bug in which the problem was seen is Bug 36158134 - OMCSCBBMUYLOSX LINUX CLIENT DAILY BACKUP FAILING WITH NDMP ERROR

This is not specifically an OSB issue but instead it is an Operating System Security Service daemon (SSSD) configuration / code bug issue.

The main OSB Bugs that address this problem are:
Bug 36332928 - SSSD erroneously closing TCP socket of Oracle Secure Backup process causing large scale backup failures.
and
Bug 36158134 - OMCSCBBMUYLOSX LINUX CLIENT DAILY BACKUP FAILING WITH NDMP ERROR

This problem has also been observed at OMCS in OEL 8.9. According to engineering, it occurred during an upgrade to latest update in Oracle Linux 8.x patch. The bug is associated with certain sssd packages and it has been seen multiple times in sssd-2.9.1-4.0.1.el8_9.x86_64
This issue can manifest itself as an index problem because the NOUPDATE file and unprocessed AIF’s will appear in the $OSB_HOME/admin/history/hosts directories, but ultimately the problem is due to the buggy sssd package.
The error observed in filesystem backup job transcripts is typically “ Error: NDMP operation failed: data service reported connect error” but it can exhibit different behavior if it is an ACSLS environment.
Other bugs associated with this issue include:
Bug 36399142 - ACSLS REPORTING "SERVER SYSTEM NETWORK INTERFACE FAILURE." AND INTERFACE TIMEOUT" SINCE ZDLRA PATCH
Bug 36344921 - OSB FAILED WITH "NDMP OPERATION FAILED: DATA SERVICE REPORTED CONNECT ERROR"
Bug 36294146 - SSSD BUG, EVIDENCED BY: OBIXD FAILED WITH NOUPDATE (OSB 18.1.0.2.0) BAD POSITION DATA DETECTED IN INDEX FILE
Bug 35708306 - IMPORT FAILS WITH BAD POSITION DATA
The goal of this MOS note is to provide the action plan to resolve this issue.

Cause

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.