Live Backup Procedure

Overview

This page outlines the steps to be followed to back up all data of an Appian installation at an application level. It is not meant to replace or eliminate the need to backup the server at a drive and operating system level.

A live data backup can be performed on a running system in order to replicate the state of the data to an external storage device for restoration later should some catastrophic failure occur on the primary system.

Configuring replication

A live backup of Appian involves five different types of data that require five different replication mechanisms.

• File system: Appian stores some of its system data, such as uploaded documents, archived processes, and more, on the host server's file system.

• Data service: Appian stores data for record types with sync enabled in the data service, along with other application metadata. Only self-managed customers need to configure data server data replication by using point-in-time file system backups, as data server data is automatically backed up for Appian Cloud customers.

• Internal Messaging Service: The Internal Messaging Service holds data flowing between different components of Appian as well as transaction logs for the Appian engines and the data service.

• RDBMS: Appian typically connects to at least two relational databases: an Appian Data Source and one or more business Data Sources.

• Search server: Appian runs Elasticsearch to provide search and data retrieval capabilities.

File system replication

The locations of application data listed in the following table should be replicated to your backup location using your standard file replication mechanism, such as rsync, disk snapshotting, or similar.

If you have more than the default three shards of Process Execution and Process Analytics, the gw1/ directories for those shards must be backed up as well.

When running in a system with multiple application servers, some of these directories are shared between servers. In those cases, the data only needs to be backed up once from one of the servers. For those directories that are not shared between servers, the data needs to be backed up from each of the servers. See High Availability and Distributed Installations for a list of directories that are shared when configuring a multiple server setup.

Internal Messaging Service replication

The Internal Messaging Service is able to replicate its data to a backup location. Appian must be installed on the backup server, but only the Internal Messaging Service needs to be running there for the backup to take place.

This is a continuously running process that backs up the Internal Messaging Service data on a transaction-by-transaction basis. Only the transaction data is backed up by this process and any related data, such as engine .kdb files, must be backed up by other means like File System Replication

Data service replication

Data service replication can only be achieved by using point-in-time file system backups, such as EBS incremental snapshots, VMware snapshots, or other similar procedures. These backups must include the data service Kafka Topic (ads_transaction_effects-0). rsync and other non-point-in-time backups are not supported. With a point-in-time backup, the data service can be recovered by starting the recovery site using the backed-up data.

If you are unable to use a point-in-time file system backup, the data service will have to be started without data on the backup server and all synced record types will need to be resynced. If you are using MirrorMaker to replicate your Kafka logs, the data service topic will need to be deleted.

To start Appian when you cannot use a point-in-time file system backup:

1. Delete the data service Kafka topic by running the following:

   1
   {APPIAN_HOME}/data-server/bin/delete-kafka-data.{sh/bat} {ABSOLUTE PATH TO KAFKA BIN DIRECTORY} {HOSTNAME} -f
   

2. Delete historical store and snapshot data from the data service by running the following:

   1
   2
   3
   rm -rf <APPIAN_HOME>/ae/data-server/data/hs
   rm -rf <APPIAN_HOME>/ae/data-server/data/ss 
   rm -f <APPIAN_HOME>/ae/data-server/data/.no-op
   

3. Start Appian.
4. Start a full sync on all synced record types.

Topology

In conf/appian-topology.xml on the backup server, add sourceKafkaCluster and mirrorMakerCluster elements as in the examples below. In these examples, the backup system runs on backupserver1.example.com and the system to be replicated from runs on primaryserver1.example.com.

Example: Single-node

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
<topology port="5000">
  <server host="backupserver1.example.com">
    <engine name="forums"/>
    <engine name="notify"/>
    <engine name="notify-email"/>
    <engine name="channels"/>
    <engine name="content"/>
    <engine name="collaboration-statistics"/>
    <engine name="personalization"/>
    <engine name="portal"/>
    <engine name="process-design"/>
    <engine name="process-analytics0"/>
    <engine name="process-analytics1"/>
    <engine name="process-analytics2"/>
    <engine name="process-execution0"/>
    <engine name="process-execution1"/>
    <engine name="process-execution2"/>
  </server>
  <search-cluster port="9300">
    <search-server/>
  </search-cluster>
  <kafkaCluster>
    <broker host="backupserver1.example.com" port="9092"/>
  </kafkaCluster>
  <zookeeperCluster>
    <zookeeper host="backupserver1.example.com" port="2181"/>
  </zookeeperCluster>
  <sourceKafkaCluster>
    <broker host="primaryserver1.example.com" port="9092"/>
  </sourceKafkaCluster>
  <mirrorMakerCluster>
    <instance host="backupserver1.example.com"/>
  </mirrorMakerCluster>
  <data-server-cluster>
    <data-server host="backupserver1.example.com" port="5400" rts-count="2"/>
  </data-server-cluster>
</topology>

Example: High availability

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
<topology port="5000">
  <server host="backupserver1.example.com">
    <engine name="forums"/>
    <engine name="notify"/>
    <engine name="notify-email"/>
    <engine name="channels"/>
    <engine name="content"/>
    <engine name="collaboration-statistics"/>
    <engine name="personalization"/>
    <engine name="portal"/>
    <engine name="process-design"/>
    <engine name="process-analytics0"/>
    <engine name="process-analytics1"/>
    <engine name="process-analytics2"/>
    <engine name="process-execution0"/>
    <engine name="process-execution1"/>
    <engine name="process-execution2"/>
  </server>
  <server host="backupserver2.example.com">
    <engine name="forums"/>
    <engine name="notify"/>
    <engine name="notify-email"/>
    <engine name="channels"/>
    <engine name="content"/>
    <engine name="collaboration-statistics"/>
    <engine name="personalization"/>
    <engine name="portal"/>
    <engine name="process-design"/>
    <engine name="process-analytics0"/>
    <engine name="process-analytics1"/>
    <engine name="process-analytics2"/>
    <engine name="process-execution0"/>
    <engine name="process-execution1"/>
    <engine name="process-execution2"/>
  </server>
  <server host="backupserver3.example.com">
    <engine name="forums"/>
    <engine name="notify"/>
    <engine name="notify-email"/>
    <engine name="channels"/>
    <engine name="content"/>
    <engine name="collaboration-statistics"/>
    <engine name="personalization"/>
    <engine name="portal"/>
    <engine name="process-design"/>
    <engine name="process-analytics0"/>
    <engine name="process-analytics1"/>
    <engine name="process-analytics2"/>
    <engine name="process-execution0"/>
    <engine name="process-execution1"/>
    <engine name="process-execution2"/>
  </server>
  <search-cluster>
    <search-server host="backupserver1.example.com"/>
    <search-server host="backupserver2.example.com"/>
    <search-server host="backupserver3.example.com"/>
  </search-cluster>
  <data-server-cluster>
    <data-server host="backupserver1.example.com" port="5400" rts-count="2"/>
    <data-server host="backupserver2.example.com" port="5400" rts-count="2"/>
    <data-server host="backupserver3.example.com" port="5400" rts-count="2"/>
  </data-server-cluster>
  <kafkaCluster>
    <broker host="backupserver1.example.com" port="9092"/>
    <broker host="backupserver2.example.com" port="9092"/>
    <broker host="backupserver3.example.com" port="9092"/>
  </kafkaCluster>
  <zookeeperCluster>
    <zookeeper host="backupserver1.example.com" port="2181"/>
    <zookeeper host="backupserver2.example.com" port="2181"/>
    <zookeeper host="backupserver3.example.com" port="2181"/>
  </zookeeperCluster>
  <sourceKafkaCluster>
    <broker host="primaryserver1.example.com" port="9092"/>
    <broker host="primaryserver2.example.com" port="9092"/>
    <broker host="primaryserver3.example.com" port="9092"/>
  </sourceKafkaCluster>
  <mirrorMakerCluster>
    <instance host="backupserver1.example.com"/>
  </mirrorMakerCluster>
</topology>

The host of the broker elements inside the sourceKafkaCluster elements should point to the primary site's Kafka cluster (the one you want to replicate data from).

The number of broker elements inside sourceKafkaCluster must match the number of broker elements inside kafkaCluster as well as the number of brokers in the primary site's kafkaCluster. For example, if the primary site is running three Kafka brokers then the backup site must also run three Kafka brokers as well as have three brokers listed in the sourceKafkaCluster element.

The mirrorMakerCluster controls which server will run the replication process. There can only be one instance of this configured and it must be one of the servers listed in kafkaCluster.

Firewall

In order for the Internal Messaging Service live backup procedure to work, the backup site must be able to connect to the primary site via port 9092.

Starting the backup process

1. Preparing the Backup Instance

Delete the following directories from the servers hosting the backup site if they exist:

APPIAN_HOME/services/data/zookeeper

APPIAN_HOME/services/data/kafka-logs

2. Start ZooKeeper

On the servers that host the backup site's Zookeeper cluster, run:

1
APPIAN_HOME/services/bin/start.sh -p <password> -s zookeeper

3. Start Kafka

Verify that all of the Kafka brokers are running on the primary site. Then, run the following command on all the servers that host the backup site's Kafka cluster:

1
APPIAN_HOME/services/bin/start.sh -p <password> -s kafka

4. Start MirrorMaker

Now that the Kafka instances are up and running on the backup site, we need to start a process that will watch the main site's Kafka cluster for messages and replicate them into the backup site's Kafka cluster.

On the server that matches the hostname in the mirrorMakerCluster element in the backup site's appian-topology.xml file run:

1
APPIAN_HOME/services/bin/start.sh -p <password> -s mirror-maker

5. Validate Startup

Run the Status Script to validate MirrorMaker started successfully:

1
APPIAN_HOME/services/bin/status.sh -p <password>

Review the MirrorMaker log file for errors or warnings that may have occurred during the startup process. The MirrorMaker log file can be found at the following location:

logs/service-manager/kafka/mirror-maker.log

Failover procedure

When failing over from your primary site to your backup site you must:

1. Stop the MirrorMaker processes

On the server that matches the hostname in the mirrorMakerCluster element in the backup site's appian-topology.xml file run:

1
APPIAN_HOME/services/bin/stop.sh -p <password> -s mirror-maker

2. Update the Appian Topology file

In conf/appian-topology.xml on the backup server, remove the mirrorMakerCluster and sourceKafkaCluster elements.

3. Start the backup Appian instance as normal

Start all Appian services.

4. Sync your record types

Perform a full sync for all of your record types with data sync enabled.

Monitoring Replication Latency

Monitoring and alerting for data replication latency is available via the status script. By default alerts are triggered for topics where data replication has exceeded 5 minutes. The alerting threshold can be configured to meet specific requirements by setting the serviceManager.mirrorMaker.replicationLagWarningThreshold property in custom.properties.

Example: serviceManager.mirrorMaker.replicationLagWarningThreshold = 1 minute

Specify the time-based configurations in the format of <number> <units>, for example, 5 minutes or 12 hours. If no units are specified the assumed unit is milliseconds.

Alerting thresholds should be configured to meet specific Recovery Point Objective (RPO) requirements, when applicable.

See Status Script for a full listing of available alerts.

RDBMS replication

Use your preferred backup mechanism for the specific RDBMS(s) used by your Appian environment.

Search server replication

If your site is using the Document Extraction Suite, you should regularly backup Elasticsearch data. While the feature will continue to work in the event of data loss, some or all field mappings created through past reconciliation tasks will be lost. You will need to remap fields during future reconciliation tasks, which may temporarily affect the performance and accuracy of document extraction.

You cannot backup search server data by simply copying the data directory <APPIAN_HOME>/search-server/data/. Instead, you must use the Elasticsearch snapshot and restore APIs. You can automate snapshots by configuring an Elasticsearch snapshot lifecycle management policy.

When restoring search server data from a snapshot, only the following indices should be restored:

• document-extraction
• document-extraction-table
• document-extraction-snippet

Feedback