ArchiverAgent

Archiver Agent

The ArchiverAgent is a platform services agent that uses the smap web services api for retrieving data. The agent listens for a request on the message bus that includes the smap path and a time range, and returns the data retrieved from smap to the requesting agent via the message bus.

Configuration

The configuration is done in the launch config file for the agent. The required elements are the archiver_url, and the source_name. The archiver_url is the url for the server, and includes the path to the web services api (typically /backend/api/query). The source_name corresponds to the Metadata/SourceName in smap and is used as a default in case the requester did not include the source name in the headers for the request.

This is a sample launch configuration:

{
    "agent": {
        "exec": "archiveragent-0.1-py2.7.egg --config \"%c\" --sub \"%s\" --pub \"%p\""
    },
    "archiver_url": "http://<your smap url here>/backend/api/query",
    "agentid": "Archiver1",
    "source_name": "<Your Source Here>"
}

Usage

publish('archiver/request/campus1/building1/realcatalyst1/CoolCall1',{},'(now -1h, now)')

publish('archiver/request/campus1/building1/realcatalyst1/CoolCall1',{},'(1374192541000.0, 1374193541000.0)')

The agent listens for messages on archiver/request. The path that follows this prefix is extracted and used in the Archiver query and represents the path to the value desired. There is also a source name that needs to be specified in one form or another. A default source name can be specified as part of the launch config. Alternatively, the requester may specify the source name as a property in the header of the message. The Agent looks for a SourceName property and will use that if it is given. The message portion of the request is a time range in the form of (start time, end time), and is passed as a simple string. The times may be specified in any way that smap recognizes time. The smap archiver page indicates that you can use a variety of ways to specify the time:

You can select the time region queried using a range query, or a query relative to a reference time stamp. In all these cases, the reference times must either be a timestamp in units of UNIX milliseconds, the string literal now, or a quoted time string. Valid time strings match a time format of either%m/%d/%Y, %m/%d/%Y %M:%H, or %Y-%m-%dT%H:%M:%S. For instance “10/16/1985” and “2/29/2012 20:00” are valid. These strings are interpreted relative to the timezone of the server.

The reference may be modified by appending a relative time string, using unix “at”-style specifications. You can for instance say now + 1hour or now -1h -5m for the last 1:05. Available relative time quantities are days, hours, minutes, and seconds.

The data is returned as a list of lists where the first element in the interior list is the timestamp in unix form, and the second element is the data as it is returned from smap:

[[1371851254000.0,1.0],[1371851314000.0,1.0],[1371851374000.0,1.0],[1371851434000.0,1.0],
 [1371851494000.0,1.0],[1371851554000.0,1.0],[1371851614000.0,1.0],[1371851674000.0,1.0],
 [1371851734000.0,1.0],[1371851794000.0,1.0],[1371851854000.0,1.0],[1371851914000.0,1.0],
 [1371851974000.0,1.0],[1371852034000.0,1.0],[1371852094000.0,1.0],[1371852154000.0,1.0],
 [1371852214000.0,1.0],[1371852274000.0,1.0],[1371852334000.0,1.0],[1371852394000.0,1.0],
 [1371852454000.0,1.0],[1371852514000.0,1.0],[1371852574000.0,1.0],[1371852634000.0,1.0],
 [1371852694000.0,1.0],[1371852754000.0,1.0],[1371852814000.0,1.0],[1371852874000.0,1.0],
 [1371852934000.0,1.0],[1371852994000.0,1.0],[1371853054000.0,1.0],[1371853114000.0,1.0],
 [1371853174000.0,1.0],[1371853234000.0,1.0],[1371853294000.0,1.0],[1371853354000.0,1.0],
 [1371853414000.0,1.0],[1371853474000.0,1.0],[1371853534000.0,1.0],[1371853594000.0,1.0],
 [1371853654000.0,1.0],[1371853714000.0,1.0],[1371853774000.0,1.0],[1371853834000.0,1.0],
 [1371853894000.0,1.0],[1371853954000.0,1.0],[1371854005000.0,1.0]]

There is a known issue where smap returns only up to 10000 data points using the web service, and does so silently. For this reason, requests should avoid broad time ranges to avoid having data request truncated. If necessary, split your desired time range into smaller pieces and join the data together after each request.