High Availability: Upgrade to Chef Backend 2
Warning
Chef Backend is deprecated and no longer under active development. Contact your Chef account representative for information about migrating to Chef Automate HA.
This document is no longer maintained.
This documentation describes the process for upgrading a high availability Chef Infra Server cluster.
The Chef Infra Server 14 upgrade process requires downtime for stopping the server, installing the new package, and then upgrading the server, which will include an automatic Elasticsearch reindexing operation for existing Solr users. We estimate the reindexing operation will take 2 minutes for each 1000 nodes, but the it could take more time, depending on your server hardware and the complexity of your Chef data.
Overview
These instructions cover the process of upgrading a Chef Backend cluster. Please use to the version of Chef Backend already running on your system and your target upgrade version.:
Prerequisite
Download the Chef Backend package for your target upgrade version and copy it onto each of the nodes in your system.
Identify the leader node in the back-end storage cluster:
Find leader node in the Role column in the output of the
cluster-status
subcommand. In this example,backend-1
is the leader node, as indicated by its role in the Role column.chef-backend-ctl cluster-status
Name IP GUID Role PG ES backend-1 192.168.33.215 dc0c6ea77a751f94037cd950e8451fa3 leader leader not_master backend-2 192.168.33.216 008782c59d3628b6bb7f43556ac0c66c follower follower not_master backend-3 192.168.33.217 1af654172b1830927a571d9a5ba7965b follower follower master
Minor Version Upgrade
The minor version update is appropriate for all upgrades of a Chef Backend cluster within a version. For example, updating from 1.3 to 1.4 or 2.3 to 2.4.
Step 1: Block Failover
Prevent the cluster from failing-over to a follower node while during its upgrade. Start by disabling failover, by running the following command on any one backend node:
chef-backend-ctl set-cluster-failover off
Step 2: Update the followers
Update the followers sequentially. Upgrading them simultaneously is not supported and may result in data loss. Verify the successful rejoin after each upgrade.
Install the new Chef Backend package:
RHEL and CentOS:
yum install PATH_TO_FILE.rpm
Debian and Ubuntu:
dpkg -i PATH_TO_FILE.deb
You may also want to look at the chef-ingredient cookbook to automate downloading and installing the latest package.
Run the upgrade command:
chef-backend-ctl upgrade
The update command will make any changes necessary to start the new service and verify that the updated node has rejoined the cluster.
Repeat the previous steps in this section for each remaining follower.
Step 3: Update the leader
Unblock failover, trigger upgrade with failover, block it again.
% chef-backend-ctl set-cluster-failover on
% chef-backend-ctl upgrade --failover
% chef-backend-ctl set-cluster-failover off
Step 4: Re-enable failover
Allow failover again:
chef-backend-ctl set-cluster-failover on
Step 5: Verify the cluster is stable
Check the status of the cluster:
chef-backend-ctl status
Chef Backend 1.x to 2.x Upgrade
Warning
Install the new Chef Backend package on all nodes in the cluster:
RHEL and CentOS:
yum install PATH_TO_FILE.rpm
Debian and Ubuntu:
dpkg -i PATH_TO_FILE.deb
On the leader, run the following command to take the node down for the upgrade:
chef-backend-ctl down-for-upgrade
Then issue the same command on the follower nodes:
chef-backend-ctl down-for-upgrade
Upgrade on the follower nodes first:
chef-backend-ctl upgrade
Then upgrade on the leader node:
chef-backend-ctl upgrade
On each Chef Infra Server front-end node, apply the Chef Backend cluster upgrade configuration from the previous steps:
chef-server-ctl reconfigure
To continue the upgrades on Chef Infra Server front-end nodes using this back-end cluster, see Upgrade Front-ends Associated with a Chef Backend Cluster.
Chef Backend Upgrade 2.x to 3.x
The Chef Backend 3.0 upgrade requires Chef Backend 2.1 or later. Upgrade earlier versions to Chef Backend 2.1 or later. Chef Backend 2.3.16 is the latest of the version 2 stable releases.
Chef Backend Version | Upgrade Target |
---|---|
2.1.0–2.3.16 | 3.0.0 or later |
2.0.0–2.0.45 | 2.3.16 |
0.1.0–1.4.40 | 2.3.16 |
Planning
The versions of Chef Backend, Elasticsearch, and Chef Infra Server running your system determines the strategy, index treatment, and steps required for your upgrade.
Chef Backend Version | Elasticsearch Version | Release Date | |
---|---|---|---|
3.0.0 | 6.8.23 | January 2022 | |
2.3.16 | 5.6.16 | January 2022 | |
2.2.0 | 5.6.16 | September 2020 | |
2.1.0 | 5.6.16 | February 2020 | |
2.0.30 | 5.4.1 | June 2018 | |
2.0.1 | 5.4.1 | October 2017 | |
1.4.6 | 2.3.1 | June 2017 |
See the Chef Backend Release Notes for more information.
Identify the Index Elasticsearch Version
Find the version of Elasticsearch that created the indexes. Run the following command on a back-end node and find value of he created_string:
in the output:
curl -XGET localhost:9200/chef/_settings/index.version.created*?pretty\&human
Upgrade Strategy
The Chef Backend upgrade uses a rolling strategy, but the Elasticsearch version used to create the indexes determines if your upgrade is automatic or requires downtime.
- Indexes created with Elasticsearch 5.4.1 (
"created_string" : "5.4.1"
) or earlier require downtime. - Indexes created with Elasticsearch 5.6.16 (
"created_string" : "5.6.16"
) or later are automatic and do not require downtime.
Running Backend Version | Upgrade Target | Elasticsearch Version | Upgrade Strategy |
---|---|---|---|
2.3.16 | 3.0 | "created_string" : "5.6.16" or higher | Automatic |
2.3.16 | 3.0 | lower than "created_string" : "5.6.16" | Downtime |
2.2.0 | 3.0 | "created_string" : "5.6.16" or higher | Automatic |
2.2.0 | 3.0 | lower than "created_string" : "5.6.16" | Downtime |
2.1.0 | 3.0 | "created_string" : "5.6.16" or higher | Automatic |
2.1.0 | 3.0 | lower than "created_string" : "5.6.16" | Automatic |
2.0.30 | 2.3.16 | "created_string" : "5.4.1" | Automatic |
2.0.30 | 2.3.16 | "created_string" : "2.3.1" | Downtime |
Index Treatment
The the version of Elasticsearch used to create the Chef Infra Server indexes determines the how the you will handle the indexes during the upgrade.
- Indexes created by Elasticsearch 5.4.1 (
created_string": 5.4.1
) or earlier require that you delete and then reindex after the upgrade. - Indexes created by Elasticsearch 5.4.1 (
created_string": 5.4.1
) or earlier and then upgraded to version 6.8.23 require that you delete and then reindex after the upgrade. - Indexes created by Elasticsearch 5.6.16 (
created_string : 5.6.16
) or later will automatically reindexed.
Elasticsearch Version | Index Treatment |
---|---|
"created_string" : "6.8.23" | Automatic reindex |
"created_string" : "5.6.16" | Automatic reindex |
"created_string" : "5.4.1" | Delete and reindex |
"created_string" : "2.3.1" | Delete and reindex |
Upgrade Chef Infra Servers with Elasticsearch 2.3.1 to 5.4.1
The Chef Backend upgrade for installations with indexes created by Elasticsearch 2.3.1–5.4.1 requires downtime to delete and reindex the Chef Infra Servers.
Stop ALL the Front-end Chef Infra Servers
Stop all the front-end Chef Infra Servers. This prevents communication between the front-end and the back-end while upgrade deletes and reindexes the server indexes.
- On Chef Infra Server 14.6.32 or later, put the server nodes into maintenance mode, to temporarily disable the API:
sudo chef-server-ctl maintenance on
- On Chef Infra Server 14.6.31 or earlier, stop all server nodes with the command:
sudo chef-server-ctl stop
Upgrade the Back-end Cluster
Prevent the cluster from failing-over to a follower node while during its upgrade. Start by disabling failover, by running the following command on any one backend node:
chef-backend-ctl set-cluster-failover off
Install the new Chef Backend package on each of the nodes in the back-end storage cluster:
RHEL and Centos:
yum install PATH_TO_FILE.rpm
Debian and Ubuntu:
dpkg -i PATH_TO_FILE.deb
Next, run the upgrade subcommand on each of the follower nodes:
chef-backend-ctl upgrade
After finishing the upgrades on the follower nodes, then upgrade the leader node, using the
--failover
option:chef-backend-ctl set-cluster-failover on chef-backend-ctl upgrade --failover chef-backend-ctl set-cluster-failover off
Stop the Elasticsearch services on ALL backends. ALL backends must be down simultaneously to prevent elastic search trying to recover indices.
sudo chef-backend-ctl stop elasticsearch
- Delete the current Elasticsearch indices from ALL backends. This step ensures that Elasticsearch creates new indexes with the upgraded Elasticsearch version. Use the command:
sudo rm -fr /var/opt/chef-backend/elasticsearch/data/*
- Restart the Elasticsearch services on ALL backends:
sudo chef-backend-ctl start elasticsearch
Upgrade the Index Definitions
Identify a front-end Chef Infra Server that does not serve requests. You can remove it from the load balancer or put in maintenance mode. If the front-end node is still in the load balancer:
If a front-end Chef Infra Server is still in the load balancer and meets both these conditions:
- Running version 14.6.32 or later
chef-server-ctl status
runs all services
Then you can disable the API temporarily by remote access with:
sudo chef-server-ctl maintenance on
Run the
reconfigure
subcommand to apply the upgraded index definitions to the Chef Infra Server:sudo chef-server-ctl reconfigure
Restore failover:
chef-backend-ctl set-cluster-failover on
Reindex the Chef Infra Servers
Note
Test the process by reindexing a single organization:
sudo chef-server-ctl reindex ORGNAME
And then run a standard search with knife to check if the updated indices are working correctly:
sudo /opt/opscode/bin/knife search node *:* -c /etc/opscode/pivotal.rb --server-url https://127.0.0.1:443/organizations/ORGNAME
If the test is successful, reindex all organizations:
sudo chef-server-ctl reindex -a
Restore the Chef Infra Servers
Restore front-end Chef Infra Servers that you placed in maintenance mode to regular status:
sudo chef-server-ctl maintenance off
Restart the services on the front-end servers:
chef-server-ctl start
Upgrade Chef Infra Servers with Elasticsearch 5.6.16
For indexes created by Elasticsearch version 5.6.16 ("created_string" : "5.6.16"
) or later, follow these steps:
Identify the node with the leader role using the
chef-backend-ctl cluster-status
command:Name IP GUID Role PG ES backend-1 192.168.33.215 dc0c6ea77a751f94037cd950e8451fa3 leader leader not_master backend-2 192.168.33.216 008782c59d3628b6bb7f43556ac0c66c follower follower not_master backend-3 192.168.33.217 1af654172b1830927a571d9a5ba7965b follower follower master
In this example,
backend-1
is the leader node, as indicated by its role in the Role column.Install the new Chef Backend package on all nodes in the cluster:
RHEL and CentOS:
yum install PATH_TO_FILE.rpm
Debian and Ubuntu:
dpkg -i PATH_TO_FILE.deb
Upgrade on follower nodes first:
chef-backend-ctl upgrade
Upgrade the leader node with the
--failover
option:chef-backend-ctl upgrade --failover
To continue the upgrades on Chef Infra Server front-end nodes using this back-end cluster, see Upgrade Front-ends Associated with a Chef Backend Cluster.
DRBD/Keepalived HA to Chef Backend 2.x
DRBD configurations are End of Life Products and no longer supported.
For a guide to migrating to Chef Backend from DRBD see the Best Practices for Migrating Your Chef Server and the associated webinar from the Chef Blog.
Was this page helpful?