Skip to content

Latest commit

 

History

History
82 lines (52 loc) · 3.69 KB

upgrade_notes.md

File metadata and controls

82 lines (52 loc) · 3.69 KB
Raw

Upgrade

Overview

From version 3.5.0, both installing and upgrading with the installer has been simplified. Upgrade differs from Install in that you run an installer diff to discover the differences between the old and new versions and then installer deploy for the new version.

Some guidance from Sysdig Support may be warranted in highly customized installations.

Refer to the appropriate workflow, depending on your environment:

Upgrade Notes

Kubernetes 1.22 not yet supported

Version 5.0.0 does is not yet supported on Kubernetes 1.22 due to breaking changes to the Ingress Controller. We expect to add support for 1.22 in the near future.

OpenShift HostAlreadyClaimed Error in Routes when Prometheus ingestion is enabled (workaround provided)

If Prometheus metric ingestion is enabled (sysdig.beacon.platformMetricsEnabled is true), an overlay is required to avoid an error in Routes which will prevent the Collector Route to be active and able to receive data from the agents.

This is what the error would look like:

oc get route
NAME                                                 HOST/PORT                                                       PATH                                             SERVICES                                             PORT    TERMINATION   WILDCARD
[omitted lines]
sysdigcloud-collector                                HostAlreadyClaimed
[omitted lines]

Use this overlay to avoid the error HostAlreadyClaimed

The domain_name must be different from the name used for the Collectors endpoint and it must be used for Prometheus metrics ingestion.

Cassandra 3 is now the default for fresh installs

Fresh installs by default will use Cassandra version 3.x.

Known limitations for OpenShift 3.11 and Kubernetes < 1.16

PostgreSQL HA is not supported on OpenShift 3.11 and on clusters running Kubernetes < 1.16

MySql removal (breaking change)

Any MySql related entry in the values.yaml must be removed or it will cause the installer to fail.

Preflight checks

Elastic Search

Please verify the Elastic Search health before proceeding with upgrades.

The cluster being Green indicates that the cluster is healthy. If the cluster is yellow please check the unassigned shards and increase heap & requests/limits to help the cluster to Green state before upgrade.

curl -s --cacert /usr/share/elasticsearch/config/root-ca.pem         https://readonly:${ELASTICSEARCH_READONLY_PASSWORD}@$(hostname):9200/_cat/health?v
epoch      timestamp cluster     status node.total node.data shards pri relo init unassign pending_tasks max_task_wait_time active_shards_percent
1616713959 23:12:39  sysdigcloud green           3         3     11   4    0    0        0             0                  -                100.0%

Cassandra

Verifying cassandra health before upgrades. Nodetool status with nodes in UN is the desired state.

nodetool status
Datacenter: datacenter1
=======================
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
--  Address     Load       Tokens  Owns (effective)  Host ID                               Rack
UN  100.96.1.2  1.1 MB     256     100.0%            4fdfcb3c-b57a-421b-b615-61179136842d  rack1
UN  100.96.5.2  364.59 KB  256     100.0%            83579e1a-31b7-402e-9349-d936b545a15b  rack1
UN  100.96.8.2  307.72 KB  256     100.0%            f5766eec-e013-4ddd-9a01-149f5bd9975a  rack1

Release Notes

  • See Release Notes for upgrade matrix, supported platforms & link to full feature notes