[Groonga-commit] droonga/droonga.org at 1013419 [gh-pages] Copy a memo about catalog.json from github wiki.

Back to archive index

Daijiro MORI null+****@clear*****
Thu Nov 14 16:56:14 JST 2013


Daijiro MORI	2013-11-14 16:56:14 +0900 (Thu, 14 Nov 2013)

  New Revision: 1013419acce65a7fde412af68b98aa2adc9030fd
  https://github.com/droonga/droonga.org/commit/1013419acce65a7fde412af68b98aa2adc9030fd

  Message:
    Copy a memo about catalog.json from github wiki.

  Added files:
    reference/catalog/index.md

  Added: reference/catalog/index.md (+91 -0) 100644
===================================================================
--- /dev/null
+++ reference/catalog/index.md    2013-11-14 16:56:14 +0900 (a4e9ea3)
@@ -0,0 +1,91 @@
+A droonga network consists of several resources.
+**Catalog** is a series of data which represent the resouces in the network.
+**Catalog** is shared all the nodes in the network.
+So far, a **catalog** is only a json file which must be written and delivered manually.
+Hopefully it would be generated by some utility program in near future, furthermore it would be maintained automatically and shared via droonga network itself.
+Resources which **catalog** manages are as following.
+
+### effective_date
+
+A date string representing the day the **catalog** becomes effective.
+
+### zones
+
+**Zone** is an array of **farms** (or other **zones**). The elements in a **zone** are expected to be close to each other, like in the same host, in the same switch, in the same network.
+
+### farms
+
+**Farms** correspond with fluent-plugin-droonga instances. A fluentd process may have multiple **farms** if more than one **match** entry with type **droonga** appear in the "fluentd.conf".
+Each **farm** has its own job queue.
+Each **farm** can attach to a data partition which is a part of a **dataset**.
+
+### datasets
+
+A **dataset** is a set of **tables** which comprise a single logical **table** virtually.
+Each **dataset** must have a unique name in the network.
+
+### ring
+
+**Ring** is a series of partitions which comprise a dataset. **replica_count**, **number\_of\_partitons** and **time-slice** factors affects the number of partitions in a **ring**.
+
+### number\_of\_partitions
+
+**number\_of\_partitions** is an integer number which represents the number of partitions devided by the hash function. The hash function which determine where each record resides the partion in a dataset is compatible with memcached.
+
+### date_range
+
+**date_range** determines when to split the dataset. If a string "infinity" is assigned, dataset is never split by time factor.
+
+### number\_of\_replicas
+
+**number\_of\_replicas** represents the number of replicas of dataset maintained in the network.
+
+### examples
+
+    {
+      "effective_date": "2013-06-05T00:05:51Z",
+      "zones": ["localhost:23003/farm0", "localhost:23003/farm1"],
+      "farms": {
+        "localhost:23003/farm0": {
+          "device": "/disk0",
+          "capacity": 1024
+        },
+        "localhost:23003/farm1": {
+          "device": "/disk1",
+          "capacity": 1024
+        }
+      },
+      "datasets": {
+        "Wiki": {
+          "workers": {
+            "search": 4,
+            "update": 1
+          },
+          "number_of_replicas": 2,
+          "number_of_partitions": 2,
+          "partition_key": "_key",
+          "date_range": "infinity",
+          "ring": {
+            "localhost:23004": {
+              "weight": 10,
+              "partitions": {
+                "2013-07-24": [
+                  "localhost:23003/farm0.000",
+                  "localhost:23003/farm1.000"
+                ]
+              }
+            },
+            "localhost:23005": {
+              "weight": 10,
+              "partitions": {
+                "2013-07-24": [
+                  "localhost:23003/farm1.001",
+                  "localhost:23003/farm0.001"
+                ]
+              }
+            }
+          }
+        }
+      }
+    }
+
-------------- next part --------------
HTML����������������������������...
Download 



More information about the Groonga-commit mailing list
Back to archive index