Page tree
Skip to end of metadata
Go to start of metadata

Manual archiving

1.
To archive all datasets in the experiment you have to:

  • go to the "Experiment Detail View"
  • in the "Data Set" section click on the "Perform" button and choose "Archiving -> Archive".
  • if you did not select any dataset all datasets will be archived.
    If you have selected some datasets you can choose if you want to archive only them or all the datasets accessible in the table.

2.
To make the archived datasets available again repeat the steps, but choose "Archiving -> Unarchive".

3.
You can always figure out what is the archiving status of the dataset by showing the "Archiving Status" column in the datasets table.

4.
If you want to disallow archiving, choose "Archiving -> Disable". Remember that you can do this only for available datasets.
The "Archiving Status" column will change to "Available (LOCKED)".
To make archiving possible again, choose "Archiving -> Enable".

Automatic archiving

Archiving can be automated by the Auto Archiver. This is a maintenance task which triggers archiving of data sets fullfulling some conditions (e.g. not accessed since a while). Note that the auto archiver doesn't archives itself. It just automates the selection of data sets to be archived. For all configuration parameters see  AutoArchiverTask.

Archiving Policies

An archiving policy selects from the unarchived data sets candidates (which are either data sets not accessed since some days or data sets marked by a tag) the data sets to be archived. If not specified all candidates will be archived.

The policy can be specified by policy.class property. It has to be the fully-qualified name of a Java class implementing ch.systemsx.cisd.etlserver.IAutoArchiverPolicy. All properties starting with policy. specifying the policy further.

ch.systemsx.cisd.etlserver.plugins.GroupingPolicy

Description: Policy which tries to find a group of data sets with a total size from a specified interval. This is important in case of Multi Data Set Archiving. Grouping can be defined by space, project, experiment, sample, data set type or a combination of those. Groups can be merged if they are too small. Several grouping keys can be specified.

Searching for an appropriate group of data sets for auto archiving is logged. If no group could be found an admin is notified via email (email address specified in log.xml). The email contains the searching log.

Configuration:

Property KeyDescription
minimal-archive-sizeThe total size (in bytes) of the selected data sets has to be equal or the larger than this value. Default: 0
maximal-archive-size

The total size (in bytes) of the selected data sets has to be equal or the less than this value. Default: Unlimited

grouping-keys

Comma separated list of grouping keys. A grouping key has the following form:

<basic key 1>#<basic key 2>#...#<basic key n>[:merge]

A basic key is from the following vocabulary: All, SpaceProjectExperimentSampleDataSetTypeDataSet.

  • All basic keys of a group key define a grouping of all data set candidates. In each group all data sets have the all attributes defined by the basic keys in common. Note, that basic key All means no grouping. For example: Experiment#DataSetType means that the candidates are grouped according to experiment and data set type.
  • The optional :merge is used when no group fulfills the total size condition and there are at least two groups with total size below minimal-archive-size. In this case groups which are too small will be merged until the total size condition is fulfilled.
  • If a grouping key doesn't lead to a group of data set fulfilling the total size condition the next grouping key is used until a matching group is found.
  • If for a grouping key more than one matching group is found the oldest one will be chosen.
  • If merging applies for more than two groups the oldest groups will be merged first.
  • The age of a group is defined by the most recent access time stamp.

Examples:

  1. Grouping policy by experiment:
    DataSetType#Experiment, DataSetType#Project, DataSetType#Experiment#Sample
  2. Grouping policy by space:
    DataSetType#Space, DataSetType#Project:merge, DataSetType#Experiment:merge, DataSetType#Experiment#Sample:merge, DataSet:merge

Example:

plugin.properties
class = ch.systemsx.cisd.etlserver.plugins.AutoArchiverTask
interval = 10 days
archive-candidate-discoverer.class = ch.systemsx.cisd.etlserver.plugins.TagArchiveCandidateDiscoverer
archive-candidate-discoverer.tags = /admin-user/archive
policy.class = ch.systemsx.cisd.etlserver.plugins.GroupingPolicy
policy.minimal-archive-size =  30000000000
policy.maximal-archive-size = 150000000000
policy.grouping-keys = Space#DataSetType, Experiment#Sample:merge

In this example the candidates are unarchived data sets which have been tag by the user admin-user with the tag archive. The policy tries to find a group of data set with total size between 30 Gb and 150 Gb. It first looks for groups where all data sets are of the same type and from the same space. If no group is found it tries to find groups where all data sets are from the same experiment and sample (data set with no samples are assigned to no_sample). If no matching groups are found and at least two groups are below the minimum the policy tries to merge groups to a bigger group until the bigger group match the size condition. If no group can be found an email will be sent describing in detail the several steps of finding a matching group.

  • No labels