11.14.2 How to manage available space in the data storage (using disk caches)
This subsection is applicable when you will be using disk caches and two-way caches.
- Reference note
-
The range of applicability of the total data restriction function on cache files is only the source EADS servers from which data is copied. If a shortage of space occurs on a target EADS server, the target EADS server is isolated. If you will be performing compaction processing on cache files, make sure that the available space is the same on all EADS servers.
Note that the total data restriction function is disabled on an EADS server while it is under restoration processing.
This subsection explains how to monitor available space in the data storage and how to increase available space when using disk caches and two-way caches.
- Organization of this subsection
(1) Monitor available space by using statistics
Estimate the available space in the data storage from the information that is output as statistics by range (eads_store_stats.csv). Monitor these values.
Statistics by range (eads_store_stats.csv) are output only when the total data restriction function is used.
For details about the statistics by range (eads_store_stats.csv), see 11.13.5 Statistics by range (eads_store_stats.csv).
If the available space in the data storage is low, take the actions described in the following subsections.
(a) Deleting unneeded data
Increase the available space in the data storage by deleting unneeded data.
(b) Performing compaction
Perform compaction to reduce the amount of data stored in cache data files.
For details about how to perform compaction on cache data files, see 10.9.1 Performing compaction on cache data files.
(c) Adding EADS servers to the cluster
If enough space cannot be obtained from (a) and (b), add EADS servers to the cluster. Adding EADS servers to the cluster enables you to reduce the number of data items stored in each EADS server.
For details about how to add EADS servers to a cluster, see 11.1.3 How to add EADS servers to a cluster after stopping the cluster (using only disk caches).
(2) If processing results in an error due to a space shortage
This subsection explains how to handle processing errors that result from a shortage of space.
(a) Check the storage that has reached the maximum amount of data that can be stored on EADS servers
Execute the eztool storeusage --cache command to determine from the number of keys, amount of memory usage, and amount of cache file usage the storage that has reached the maximum amount of data that can be stored on EADS servers.
(b) Terminate all EADS servers in the cluster
Terminate all EADS servers in the cluster.
For details about this procedure, see 10.6 Terminating the EADS servers (terminating caches on disk).
(c) Change the properties
Re-estimate the Java heap size, the explicit heap size, and the capacity of cache files, and then change the properties based on the estimation results. If necessary, add server machines, memory, and disks.
For details about how to estimate these values, see the following:
-
Estimating the Java heap size
-
Estimating the explicit heap size
-
Estimating the capacity of cache files
For details about how to change the properties, see 11.4 Changing the properties.
(d) Start all EADS servers in the cluster (relocate data)
Relocate data after all EADS servers in the cluster have started.
For details about this procedure, see 11.1.3(5) Start all EADS servers in the cluster through 11.1.3(9) Verify that the EADS servers have been opened.