Faban collects data about performance from the harness/master and each of the utilised agents. The number of agents utilised is arbitrary (from 0 to N), as well as the use of the harness/master to generate the load is optional. That said, at least one among harness/master and agents must be utilised.
For each of the previous harness/master or agent, Faban produces a set of file. In the following I describe the files that are of our interest. Whenever I refer to fields and names that seems not defined, you can find their definition in http://faban.org/1.3/docs/guide/harness/results.html. The files are located in the /app/faban/FABAN_RUN_ID folder for each trial. An overview of the different information that are present can be accessed here: http://faban.org/1.3/docs/guide/harness/results.html.
I attach a zip file containing one of the folder (name: web101.2Q) I mentioned before, with files generated using the master and one agent.
resultinfo is just a single file containing the Status field
For each harness/master and agent you find a summary.xml.* file
Then you have a summary.xml file that aggregates the information of all the summary.xml.* files
Structure of the file
resultinfo is plain text, tab separated file
summary.xml.* are xml files with a clear structure that is self-explicative. As all the xml files in Faban, they refer to an xslt file (useful to get hints on how to parse these files). I attach them in the xslt zip file. The information relevant for the RunInfo are defined in the benchSummary entity, some of them in attributes, some other as fields.
What to get from these files
The RunInfo information as per in http://faban.org/1.3/docs/guide/harness/results.html for each of the harness/master and agent (which name is identified by the host attribute of the benchSummary entity), and the general and aggregated RunInfo where the host should be set to aggregate.
For each harness/master and agent you find a summary.xml.* file
Then you have a summary.xml file that aggregates the information of all the summary.xml.* files
Structure of the file
summary.xml.* as in the previous section. The drivers information are stored in the driverSummary entity.
Moreover for each driver (the number can be arbitrary) you find the following entities: mix, responseTimes, delayTimes, [customStats]. customStats is optional and we can arbitrary have a number from 0 to N of these entities, each one uniquely identified by the name. The structure of these entities is straightforward, just refer to the name of the fields and attributes.
mix: you can have an arbitrary number of operation entities
responseTimes: you can have an arbitrary number of operation entities and fields. The structure of the content, especially for what concern the percentiles is dynamic
To get information about the "Little's Law Verification" you need to refer to the xlst file because it's computed there. It should be the rtXtps field for each driver.
What to get from these files
The Summary Report information as per in http://faban.org/1.3/docs/guide/harness/results.html for each of the harness/master and agent (which name is identified by the host attribute of the benchSummary entity), and the general and aggregated RunInfo where the host should be set to aggregate.
They are text file where each set of data it's identified by a Section, the type of visualisation by a Display field and the data are stored in a table. The number or rows is arbitrary, as well as the number of columns because it depends on the number of operations. The unit of measure of the timing columns its also arbitrary and its identified in the name of the column itself
What to get from these files
The Detail Report information as per in http://faban.org/1.3/docs/guide/harness/results.html for each of the harness/master and agent (which name is identified by the host field in the file), and the general and aggregated Detail Report where the host should be set to aggregate.
Faban collects data about performance from the harness/master and each of the utilised agents. The number of agents utilised is arbitrary (from 0 to N), as well as the use of the harness/master to generate the load is optional. That said, at least one among harness/master and agents must be utilised.
For each of the previous harness/master or agent, Faban produces a set of file. In the following I describe the files that are of our interest. Whenever I refer to fields and names that seems not defined, you can find their definition in http://faban.org/1.3/docs/guide/harness/results.html. The files are located in the
/app/faban/FABAN_RUN_ID
folder for eachtrial
. An overview of the different information that are present can be accessed here: http://faban.org/1.3/docs/guide/harness/results.html.I attach a zip file containing one of the folder (name: web101.2Q) I mentioned before, with files generated using the master and one agent.
The RunInfo (files:
resultinfo
,summary.xml*
)Files
resultinfo
is just a single file containing theStatus
fieldsummary.xml.*
filesummary.xml
file that aggregates the information of all thesummary.xml.*
filesStructure of the file
resultinfo
is plain text, tab separated filesummary.xml.*
are xml files with a clear structure that is self-explicative. As all the xml files in Faban, they refer to anxslt
file (useful to get hints on how to parse these files). I attach them in thexslt
zip file. The information relevant for the RunInfo are defined in thebenchSummary
entity, some of them in attributes, some other as fields.What to get from these files
host
attribute of thebenchSummary
entity), and the general and aggregated RunInfo where thehost
should be set toaggregate
.The Summary Report (files:
summary.xml*
)Files
summary.xml.*
filesummary.xml
file that aggregates the information of all thesummary.xml.*
filesStructure of the file
summary.xml.*
as in the previous section. The drivers information are stored in thedriverSummary
entity.mix
,responseTimes
,delayTimes
, [customStats
].customStats
is optional and we can arbitrary have a number from 0 to N of these entities, each one uniquely identified by thename
. The structure of these entities is straightforward, just refer to the name of the fields and attributes.mix
: you can have an arbitrary number ofoperation
entitiesresponseTimes
: you can have an arbitrary number ofoperation
entities and fields. The structure of the content, especially for what concern thepercentiles
is dynamicrtXtps
field for each driver.What to get from these files
host
attribute of thebenchSummary
entity), and the general and aggregated RunInfo where thehost
should be set toaggregate
.The Detail Report (files:
detail.xan.*
)Files
detail.xan.*
filedetail.xan
file that aggregates the information of all thedetail.xan.*
filesStructure of the file
Section
, the type of visualisation by aDisplay
field and the data are stored in a table. The number or rows is arbitrary, as well as the number of columns because it depends on the number of operations. The unit of measure of the timing columns its also arbitrary and its identified in the name of the column itselfWhat to get from these files
host
field in the file), and the general and aggregated Detail Report where thehost
should be set toaggregate
.Related issue: https://github.com/benchflow/collectors/issues/36
@Cerfoglg feel free to ask questions so that I can improve the description.
web101.2Q.zip xslt.zip