LightCloud manager can manage your nodes, take backups, purge logs etc.

Getting the LightCloud manager

Simply check out the LightCloud manager in a directory:

$ svn co http://opensource.plurk.com/svn/opensource/tyrant_manager ~/tyrant_manager
$ cd  ~/tyrant_manager

Create a config file

Create a config file config.py. A sample config file is included ~/lightcloud_manager/config.sample.py:

DATA_DIR = '~/lightcloud_manager/data'
TOKYO_SERVER_PARMS = '#bnum=1000000#fpow=13#opts=ld'

USE_MASTER = True

NODES = {
    #Lookup nodes
    'lookup1_A': { 'id': 1, 'host': '127.0.0.1:41201', 'master': '127.0.0.1:51201' },
    'lookup1_B': { 'id': 2, 'host': '127.0.0.1:51201', 'master': '127.0.0.1:41201' },

    #Storage nodes
    'storage1_A': { 'id': 5, 'host': '127.0.0.1:44201', 'master': '127.0.0.1:54201' },
    'storage1_B': { 'id': 6, 'host': '127.0.0.1:54201', 'master': '127.0.0.1:44201' },
}

The options of the manager

$ python -m manager -c config.py
Script that is used to handle Tokyo Tyrant nodes.

Starting nodes:
   python manager.py -c config.py all start
   python manager.py -c config.py lookup1 start

Stopping nodes:
   python manager.py -c config.py all stop
   python manager.py -c config.py lookup1 stop

Status:
   python manager.py -c config.py all status 
   python manager.py -c config.py lookup1 status

Misc:
   python manager.py -c config.py purge_logs
   python manager.py -c config.py delete_logs
   python manager.py -c config.py delete_data

   python manager.py -c config.py hot_copy
   python manager.py -c config.py hot_restore hot_copy_23232.zip

View a sample config file in config.sample.py.

Starting the nodes:

$ python -m manager all start

Checking out status:

$ python manager.py all status
lookup1_A (127.0.0.1:41201):
    node running - node items: 10306
    master 127.0.0.1:51201 running - master items: 10306
lookup1_B (127.0.0.1:51201):
    node running - node items: 10306
    master 127.0.0.1:41201 running - master items: 10306
storage1_A (127.0.0.1:44201):
    node running - node items: 10067
    master 127.0.0.1:54201 running - master items: 10067
storage1_B (127.0.0.1:54201):
    node running - node items: 10067
    master 127.0.0.1:44201 running - master items: 10067

How to debug

If you run into problems, try to start a ttserver manually. python manager.py all start prints out the commands.

How to take a hot backup and restore it

LightCloud Tyrant manager supports taking hot backups of your database, without shutting the database down. The interface is super simple as well.

To take a hot-copy, simply do following thing:

$ python manager.py hot_copy
...
Created hot copy in ~/lightcloud_manager/data/hot_copy_1245055938.zip

This creates a zip file of database files and their log positions.

To restore a hot copy simply do following:

$ python manager.py hot_restore ~/lightcloud_manager/data/hot_copy_1245055938.zip
Restored hot copy of master in ~/lightcloud_manager/data/restore_dir

You should inspect ~/lightcloud_manager/data/restore_dir and ensure that everything looks reasonable (i.e. you have all the database files [tch files] and their log positions [rts files]).

When you have ensured that everything looks reasonable, you can do following to swap the current data directory:

$ python manager.py all stop
$ mv ~/lightcloud_manager/data/data ~/lightcloud_manager/data/data_old
$ mv ~/lightcloud_manager/data/restore_dir ~/lightcloud_manager/data/data
$ python manager.py all start

You can then run status to check consistency:

$ python manager.py all status
lookup1_A (127.0.0.1:41201):
    node running - node items: 10306
    master 127.0.0.1:51201 running - master items: 10306
lookup1_B (127.0.0.1:51201):
    node running - node items: 10306
    master 127.0.0.1:41201 running - master items: 10306
storage1_A (127.0.0.1:44201):
    node running - node items: 10067
    master 127.0.0.1:54201 running - master items: 10067
storage1_B (127.0.0.1:54201):
    node running - node items: 10067
    master 127.0.0.1:44201 running - master items: 10067