Networking NVIDIA UFM High-Availability User Guide

Bug Fixes in This Release

Ref #

Description

5250145

Description: Resolved an issue where the set_ha_nodes.sh configure command updated HA_Nodes in gv.cfg with fabric GUIDs instead of the actual master and standby nodes.

Keywords: set_ha_nodes.sh, HA_Nodes, gv.cfg

Discovered in Release: v6.3.1

5237214

Description: Resolved an issue where Pacemaker stop operations timed out while UFM was still shutting down on busy fabrics.

Keywords: Pacemaker, timeout, busy fabric

Discovered in Release: v6.3.1

4411296

Description: Resolved an issue where Pacemaker file logging stopped after HA configuration on UFM Enterprise Appliance.

Keywords: Pacemaker, logging, UFM Enterprise Appliance

Discovered in Release: v5.9.0

4667059

Description: Resolved an issue where HA cleanup remounted /opt/ufm/files to the former DRBD disk when it had not been mounted before configuration.

Keywords: Cleanup, DRBD, mount

Discovered in Release: v6.1.1

4763916

Description: Added a clear error message when failover is requested while the cluster is in maintenance mode.

Keywords: Failover, maintenance mode

Discovered in Release: v6.1.1

4836021

Description: Resolved an issue where ufm-ha-cluster_startup_config.yaml had incorrect root ownership on fresh rootless OCI installations.

Keywords: Rootless, OCI, permissions

Discovered in Release: v6.3.0

5010945

Description: Removed redundant backup and restore operations during HA cleanup in NFS mode.

Keywords: NFS, cleanup, backup

Discovered in Release: v6.2.1

5138760

Description: Resolved an issue where HA configuration reported success when DRBD metadata creation failed and the standby remained Diskless.

Keywords: DRBD, Diskless, configure

Discovered in Release: v6.3.0

5161133

Description: Resolved an issue where retrying an interrupted HA configuration left /opt/ufm/files empty while reporting success.

Keywords: DRBD, interrupted configure, data restore

Discovered in Release: v6.1.1

5161921

Description: Resolved an issue where the k0s HA backup failed and produced no backup archive when the host had no Redis service.

Keywords: k0s, backup, Redis

Discovered in Release: v6.3.0

5161972

Description: Resolved an issue where k0s SSH preflight checks failed on hosts whose default UserKnownHostsFile is /dev/null.

Keywords: k0s, SSH, known_hosts

Discovered in Release: v6.3.0

5162066

Description: Resolved an issue where a missing CLI option value caused raw Bash errors or consumed the following option.

Keywords: k0s, CLI, validation

Discovered in Release: v6.3.0

5162110

Description: Resolved an issue where an empty k0s HA configuration was reported as healthy.

Keywords: k0s, configuration, status

Discovered in Release: v6.3.0

5162184

Description: Resolved an issue where kubectl exec, logs, and port-forward commands failed because k0s Konnectivity was not configured with the HA VIP.

Keywords: k0s, Konnectivity, VIP

Discovered in Release: v6.3.0

5182216

Description: Resolved an issue where CIB schema validation errors occurred while configuring UFM4 HA.

Keywords: UFM4, Pacemaker, CIB schema

Discovered in Release: v6.3.0

5211433

Description: Resolved an issue where IPv6 detach and attach failed when Pacemaker identified nodes by hostname.

Keywords: IPv6, detach, attach, Pacemaker

Discovered in Release: v6.3.0

5226594

Description: Resolved an issue where Corosync and Pacemaker did not recover after a direct B2B peer reboot dropped carrier on the active node.

Keywords: UFM Enterprise Appliance, Corosync, direct B2B

Discovered in Release: v6.3.1

5339203

Description: Resolved an issue where k0s is-ha returned false when the Kubernetes API was unavailable, even though HA remained configured.

Keywords: k0s, is-ha, API availability

Discovered in Release: v6.4.1

5355721

Description: Resolved an issue where k0s nodes registered with a hostname-resolved IPv6 address instead of the configured management IP, which prevented cluster networking and core pods from starting.

Keywords: k0s, IPv6, management IP, kube-router

Discovered in Release: v6.4.1

5355724

Description: Resolved an issue where recurring k0s ERROR log messages about Redis replication and removal of an absent VIP appeared in the default two-node profile.

Keywords: k0s, logging, Redis, VIP

Discovered in Release: v6.4.1

5355725

Description: Added a check before start for bare-metal UFM host services, such as apache2, that conflict with the k0s UFM pod on ports 80 and 443.

Keywords: k0s, preflight, ports 80/443, apache2

Discovered in Release: v6.4.1

5354213

Description: Resolved an issue where a retried standby HA configuration reported success without initializing DRBD metadata when an earlier configuration attempt had left a DRBD resource file behind.

Keywords: DRBD, standby, configuration retry

Discovered in Release: v6.4.1

5162087

Description: Resolved an issue where ufm_ha_cluster attach did not import the bundled images on a rejoined k0s node in offline deployments, which left its pods in ImagePullBackOff.

Keywords: k0s, attach, containerd, ImagePullBackOff, offline

Discovered in Release: v6.3.0

5355726

Description: Resolved an issue where k0s HA state under /var/lib/ufm-ha remained after cleanup and package uninstall.

Keywords: k0s, uninstall, cleanup

Discovered in Release: v6.4.1

5364526

Description: Resolved an issue where ufm_replace_standby rejected the documented --enable-single-link option.

Keywords: Standby replacement, single link

Discovered in Release: v6.4.1

5379456

Description: Resolved an issue where the lower-priority k0s node sometimes became master on the first start of a new cluster.

Keywords: k0s, election, priority

Discovered in Release: v6.4.1

5379459

Description: Resolved an issue where ufm_ha_cluster status reported the k0s HA operator as running when no operator pod was running.

Keywords: k0s, status

Discovered in Release: v6.4.1

5387859

Description: Resolved an issue where ufm_ha_cluster is-master reported standby on both nodes, and failover was rejected, when Pacemaker nodes were named by IPv6 address.

Keywords: IPv6, is-master, failover, Pacemaker

Discovered in Release: v6.3.1

5393248

Description: Updated Go to 1.26.10 and the bundled Helm to v3.22.1 to address known security vulnerabilities in the k0s components.

Keywords: k0s, security, Go, Helm

Discovered in Release: N/A

5398982

Description: Resolved an issue where ufm_ha_cluster cleanup restored old pre-HA data instead of the current data when the data directory path ended with a slash.

Keywords: Cleanup, DRBD, data restore

Discovered in Release: v5.9.0



Last updated: