Clustrix Documentation
Space shortcuts

Page tree

Versions Compared

Old Version 3

changes.mady.by.user Christine Lieu

Saved on

compared with

New Version 4

changes.mady.by.user Christine Lieu

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.
Comment: Published by Scroll Versions from space ML1 and version 9.2

As part of the ClustrixDB Installation process you made some configuration selections . Once you have completed installation please use the following directions to modify any of these optionswhich are written to clxnode.conf. Post-installation, you can modify the configuration by editing the file on each node while ensuring that the options specified are the same (other than node specific items like BACKEND_ADDR). 

The clxnode.conf is stored on each node in /etc/clustrix/ and options specified during installation are written to this file. The following table summarizes these options and provides information on how they can be modified post-installation.

...

Note

ClustrixDB requires that all nodes have identical configurations (other than node specific items like BACKEND_ADDR). Modifications to any of the options described below must be made on every node in the cluster. Values commented out with a # will have no effect on the database. Changes to the configuration requires a restart of ClustrixDB services (see below) to take effect.

Default Options for clxnode.conf

...

OptionDescriptionDefault ValueTo Modify
NODE_MEMORYMemory to use for ClustrixDB, in MiBauto detected
  1. Edit

    On every node in your cluster edit the clxnode.conf file

    for every node in your cluster

    and uncomment (if commented out) the line #NODE_MEMORY and change the value to the desired

    value

    amount.

  2. Run the following to propagate this change to all nodes: 
    shell> /opt/clustrix/bin/clx -P cmd 'echo X > /proc/sys/vm/nr_hugepages'
    Where X = NODE_MEMORY/2

  3. Stop/Start all ClustrixDB services.

  4. Stop ClustrixDB, restart hugetlb, and then Start ClustrixDB on all nodes simultaneously by running the following command from a single node:
    clx -P cmd 'systemctl stop clustrix; systemctl restart hugetlb; systemctl start clustrix'

MAX_REDOMaximum ClustrixDB Redo Space, in MiB1024Only modify this variable with guidance from Clustrix Support. 
CPU_CORESNumber of CPU cores ClustrixDB will useAll
  1. Edit the clxnode.conf file for every node in your cluster to the desired value.
  2. Stop/Start all ClustrixDB services.

For additional info, please see section below on setting CPU_CORES.

DATA_PATHClustrixDB Database Storage Path/data/clustrix
  1. Stop all ClustrixDB services.
  2. On each node:
    1. Move the directory to the desired location.
    2. Modify the value of DATA_PATH in clxnode.conf to reflect the new location.
  3. Start all ClustrixDB services.
DATA_ALLOCATEStorage space to pre-allocate to ClustrixDBauto detected 80% of free space on $DATA_PATH volume
 This value here is used for initialization only. To allocate more space on a node that already has ClustrixDB installed, see ALTER CLUSTER RESIZE DEVICES
LOG_PATHClustrixDB Logs Path/data/clustrix/log
  1. Stop all ClustrixDB services.
  2. On each node:
    1. Move the directory to the desired location.
    2. Create a simlink in /data/clustrix/ pointing to the new log folder.
      Example: ln -s /log /data/clustrix/log/
  3. Start all ClustrixDB services.
UI_LOGDIRClustrixDB WebUI Logs Path/data/clustrix/log/clustrix_ui
  1. Stop all ClustrixDB services.
  2. On each node:
    1. Move the directory to the desired location.
    2. Create a simlink in /data/clustrix/log pointing to the new log folder.
      Example: ln -s /log/clustrix_ui /data/clustrix/log/clustrix_ui
  3. Start all ClustrixDB services.
UI_CACHEDIRClustrixDB WebUI Cache Path/var/cache/clustrix/djangoOnly modify this variable with guidance from Clustrix Support.
UNIX_SOCKET_PATHClustrixDB MySQL Protocol Unix Socket/var/lib/mysql/mysql.sockOnly modify this variable with guidance from Clustrix Support.
LISTEN_ADDRClustrixDB Database Listen Address (Front-End IP) Interface0.0.0.0 
MYSQL_PORTClustrixDB Database MySQL Port3306 
BACKEND_ADDRClustrixDB Private Back-End IP Interfaceauto detectedSee section below on changing BACKEND_ADDR.
BACKEND_PORTClustrixDB Back End Network Port24378It is not recommended to modify the default value for the ClustrixDB Back End Network port.
HEALTHMON_PORTThe port used for the ClustrixDB health monitor.3581
  1. Edit value in clxnode.conf
  2. Stop/Start all ClustrixDB services.
HTTP_PORTClustrixDB WebUI HTTP Port80
  1. Edit value in clxnode.conf
  2. Stop/Start all ClustrixDB services.
NANNY_PORTClustrixDB Nanny Port2424It is not recommended to modify the default value for the ClustrixDB nanny port.
CONTROL_PORTClustrixDB Control Port2048It is not recommended to modify the default value for the ClustrixDB control port.
CLX_USERClustrixDB OS UserrootDo not modify this value after installation.
MANAGE_USERClustrixDB Manager UserrootDo not modify this value after installation.
WRITE_HOSTSAllow ClustrixDB to modify sshd_config and /etc/hosts.

This is required used by root installations only, for internode communication for administrative tasks, including upgrades.

#Commented out
  1. On each node: 
    1. Remove comment and set variable to any string.
    2. Stop/Start all ClustrixDB services.
HUGE_TLB_ENABLE

ClustrixDB enables HugeTLB memory allocation for faster startup. NOTE: This causes instability on some systems (particularly certain virtual machines).

Required for non-root installations that have more than 20GiB of $NODE_MEMORY

Contact Clustrix Support before changing from default.

True
  1. On each node, comment out to disable this option. (i.e. #HUGE_TLB_ENABLE)
  2. Stop/Start all ClustrixDB services.

...

The following options are NOT included by default in clxnode.conf, but can be added to the end of that file to modify behavior from the default:. ClustrixDB must be restarted for changes to take effect. 

OptionDescriptionDefault ValueTo Modify
DISABLE_MULTIPORTWhen this option is present, the multiport feature is disabled.This option is not included in clxnode.conf, resulting in multiport being enabled by default.

To disable multiport, add

this entry and define this variable as any value on all nodes. (Its mere presence disables multiport.)

an entry for DISABLE_MULTIPORT = Y.


To enable multiport, remove or comment out this entry from clxnode.conf on all nodes.

MULTIPAGE_ALLOCBytes reserved for multi-page allocators.The option is not included clxnode.conf, resulting in a value of 1GiBAdd this entry and define the number of bytes that Clustrix Support recommends for your workload.

Details of Modifying Startup Configuration Options

...

Stop/Start all ClustrixDB Services

To restart all apply the changes that have been made to clxnode.conf, the ClustrixDB services on all nodes (via nanny), use the following commands. This will restart nanny, and all jobs controlled by nanny, and cause the database to come online. See Overview of Nanny.the nodes will need to be restarted:

...

shell> /opt/clustrix/bin/clx
cmd 'systemctl stop clustrix' shell> /opt/clustrix/bin/clx cmd 'systemctl start clustrix'

...

dbrestart

CPU_CORES

The default setting of CPU_CORES=All will make use of all available cores in the system. If you would like to limit the number of cores utilized by ClustrixDB you must modify the value for CPU_CORES to specify the number of cores that should be used. ClustrixDB licensing will base its enforcement on the number of cores being utilized in the system as specified by CPU_CORES. 

Note

If you are upgrading from v6.X, there may not be an entry for CPU_CORES in clxnode.conf which will implicitly specify CPU_CORES=All. If you would like to specify an alternate value simply add an entry for CPU_CORES.

BACKEND_ADDR

  1. Modify the BACKEND_ADDR entry in the /etc/clustrix/clxnode.conf file with the new IP.
  2. Modify /data/clustrix/broadcast_nodes file with the new IP addresses on all nodes.
    1. Note: Specifying port 24378 is not necessary unless you're using a non-default configuration.

  3. Run the clx command to ensure it has cached the IP address of each node.

    shell> /opt/clustrix/bin/clx cmd 'dateclx cache'
  4. Stop all ClustrixDB services (per above).
  5. Change the backend interface IP address on each node.
  6. Modify iptables as needed.
  7. Start all ClustrixDB services (per above).

...