On this page


Teamwork Cloud uses Apache Cassandra, an open-source NoSQL distributed database. Before installing Teamwork Cloud, please follow the steps below to install Apache Cassandra.

Prerequisites

  • chkconfig package (for service startup on boot)

Installing with script

The script downloads and installs the necessary packages, Cassandra, and the Cassandra tools from the Apache Software Foundation repository, and creates the necessary firewall rules to allow proper operation both for a single node or a cluster installation. The script will also install Java 11 and set it as the default system Java. 

The script can also be used for offline installation, download the prerequisite RPM packages and place them in the same location as the installation script. Manually install Java 11 and set it as the system default Java. Installation with Apache Cassandra RPM still uses the older System V init to automate service startup on boot. In order to enable the Cassandra service with systemctl, installation of the chkconfig package is needed.

To install Apache Cassandra

  1. Install Apache Cassandra by executing the install_cassandra4x_ol_rhel.sh installation script.

    Example
    sudo ./install_cassandra4x_ol_rhel.sh

  2. Start Apache Cassandra by executing the following command:

    sudo systemctl start cassandra

  3. Check if Apache Cassandra is running by executing the following command:

    nodetool status


    If Apache Cassandra is running, you should receive the output displayed below. If the service is fully operational, the first 2 characters of the last line are "UN", indicating that the node status is Up, and its state is Normal.


    Datacenter: datacenter1
=======================
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
--  Address    Load       Tokens   Owns (effective)  Host ID                               Rack
UN  127.0.0.1  128.4 KB   256      100.0%            ea3f99eb-c4ad-4d13-95a1-80aec71b750f  rack1

    Wait for a few minutes until Cassandra starts for the first time before checking if it is running. If Cassandra has not started yet, you will get the error: "No nodes present in the cluster. Has this node finished starting up?" This means that you need to give Cassandra more time to start.

  4. If Apache Cassandra is not running or if you used installation options other than the one described in this chapter, optionally configure Apache Cassandra.

Developing a backup strategy

Before deploying Teamwork Cloud and Apache Cassandra in a production environment, it is imperative to have a fully implemented backup strategy. The Cassandra database stores all project and user data associated with Teamwork Cloud. Review the backup and restore data procedure document. Ensure that you test the entire backup and restore process before your deployment goes live to users.

During the backup process, user access to Cassandra should be suspended. Refer to the Cassandra backup documentation for more information. 

Improper backup procedure can lead to total data loss! For example, taking an image snapshot of the storage system while Cassandra is actively accepting read and write requests will result in unrecoverable data.


Configuring Apache Cassandra for Teamwork Cloud

If you used other installation options and not the provided script or if Apache Cassandra does not start, configure it as described below.

Before starting, note that you do not need to configure Apache Cassandra if you installed it using the installation script we provided (install_cassandra<version_number>_<os_version>.sh). It should start without any additional configuration.

To configure Apache Cassandra

  1. Edit the cassandra.yaml file by executing the following command:

    sudo nano /etc/cassandra/default.conf/cassandra.yaml

  2. Find the following parameters related to the Cassandra node IP address and communication settings, and change them as shown below:

    Example
    seeds: "192.168.130.10"
listen_address: 192.168.130.10
broadcast_rpc_address: 192.168.130.10
rpc_address: 0.0.0.0
    • seeds - a comma-delimited list containing all of the seeds in the Cassandra cluster. Since our cluster consists of a single node, it contains only one entry - our IP address.
    • listen address - the IP address that Cassandra uses to listen for connections.
    • broadcast_rcp_address - the IP address used to broadcast to other Cassandra nodes in the cluster. This parameter may be commented. In such case, remove "#" and make sure there are no leading spaces.
    • rcp_address - when set to to 0.0.0.0, Cassandra listens to rpc requests on all interfaces.

  3. Find the following parameters that control thresholds to ensure that the data being sent is processed properly, and change them as shown below:

    Example
    commitlog_segment_size: 192MiB
read_request_timeout: 1800000ms
range_request_timeout: 1800000ms
write_request_timeout: 1800000ms
cas_contention_timeout: 1000ms
truncate_request_timeout: 1800000ms
request_timeout: 1800000ms
batch_size_warn_threshold: 3000KiB
batch_size_fail_threshold: 5000KiB

  4. To ensure that the default commit log size is 8GB (recommended), uncomment the commitlog_total_space_in_mb parameter as show as below.

    Example
    commitlog_total_space: 8192MiB

    Ensure that the partition where the commit log is installed has enough space to accommodate a commit log of 8GB.

  5. To point the data to the appropriate locations, find the following parameters and change them as shown below:

    Example
    data_file_directories:
- /data/data
commitlog_directory: /logs/commitlog
hints_directory: /data/hints
saved_caches_directory: /data/saved_caches

  6. Start Apache Cassandra by executing the following command:

    sudo systemctl start cassandra

  7. Check if Apache Cassandra is running by executing the following command:

    nodetool status

    If Apache Cassandra is running, you should receive the output displayed below. If the service is fully operational, the first 2 characters of the last line are "UN", indicating that the node status is Up, and its state is Normal.

    Example
    Datacenter: datacenter1
=======================
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
--  Address    Load       Tokens   Owns (effective)  Host ID                               Rack
UN  127.0.0.1  128.4 KB   256      100.0%            ea3f99eb-c4ad-4d13-95a1-80aec71b750f  rack1

Configuring Cassandra memory usage

If you did not use the installation script or want to increase the RAM usage by Cassandra, make the following changes. Otherwise, these configuration changes are set automatically by the Cassandra installation script.

Configuration files are located in /etc/cassandra/conf/.

  • By default, the maximum RAM usage for Cassandra is 8GB. To change the amount of RAM used by Cassandra, uncomment -Xms4G (min) and -Xmx4G (max) in the jvm-server.options file and specify their values.
  • In the jvm11-server.options and jvm8-server.options files, comment all lines from"### CMS Settings" to "### G1 Settings".

  • In the jvm11-server.options and jvm8-server.options files, uncomment the following lines:

    #-XX:+UseG1GC
#-XX:MaxGCPauseMillis=500

  • In the jvm11-server.options and jvm8-server.options files, uncomment the following lines and set the values to the physical CPU core count (the values of both parameters should be the same):

    #-XX:ParallelGCThreads=16
#-XX:ConcGCThreads=16
  • In the jvm8-server.options file, comment all lines from "### GC logging options" to the end of the file.

  • Synchronize CPU clocks on all Cassandra cluster nodes. Otherwise, you may encounter issues when creating an empty Cassandra cluster.

  • When using cqlsh, use Python 3.6.0 or a later version. Python 2.7 series is no longer supported.
  • In the logback.xml file, comment the "<appender-ref ref="ASYNCDEBUGLOG" />" line. This will increase Cassandra's performance by disabling the debug log.

Configuring Linux environment for Cassandra performance

If you install Teamwork Cloud using the install_twc_mcs_centos_rhel.sh script, Cassandra performance is tuned automatically. However, if you plan to use other installation options or if you need to set other parameters after running the script, you can do it manually as described in this section.

To improve Apache Cassandra performance

  1. Open the sysctl.conf file by executing the following command:

    sudo nano /etc/sysctl.conf

  2. To configure the TCP settings, add the following tuning parameters to the file:

    Example
    net.core.rmem_max=16777216
net.core.wmem_max=16777216
net.core.optmem_max=40960
net.core.default_qdisc=fq
net.core.somaxconn=4096
net.ipv4.conf.all.arp_notify = 1
net.ipv4.tcp_keepalive_time=60
net.ipv4.tcp_keepalive_probes=3
net.ipv4.tcp_keepalive_intvl=10
net.ipv4.tcp_mtu_probing=1
net.ipv4.tcp_rmem=4096 12582912 16777216 
net.ipv4.tcp_wmem=4096 12582912 16777216 
net.ipv4.tcp_max_syn_backlog=8096
net.ipv4.tcp_slow_start_after_idle = 0
net.ipv4.tcp_tw_reuse = 1 
vm.max_map_count = 1048575
vm.swappiness = 0
vm.dirty_background_ratio=5
vm.dirty_ratio=80
vm.dirty_expire_centisecs = 12000

  3. To apply the setting without rebooting, execute the following command:

    sudo sysctl -p

For more information about tuning Linux, see DSE 6.8 Administrator Guide.

Using jemalloc memory allocator

The jemalloc memory allocator package can potentially improve Cassandra performance. Our installation script does not install jemalloc. The easiest way to install this optional package is to first install the epel-release package. You can then pull the latest jemalloc release from the EPEL repository. An older version jemalloc is available for direct download and install.

  • EPEL package (optional, for pulling jemalloc package)
  • jemalloc package (optional for performance)