We have a build process using ant that is externally monitored (web page)
and shows the current goal in a group of goals. The active goal is being
uploaded to elasticsearch every 60 seconds and is allowing us to monitor a
builds progress. So for three goals (A,B & C) we would see something like
"AAABBBBCCCCCC" with A taking 3 minutes, B taking 4 and C
taking 6.
Using this, we are able to manually/visually determine how long each goal
is taking. What we would like to be able to do is query and ascertain
trends/expected times from this data to, for example, allow us to flag when
builds/goals take 20% longer than the average. Is this possible with the
current data set or do we need to change what we upload?
Appreciate any/all help.
(Note: currently using elasticsearch 1.1.2-1 on CentOS 6)
We have a build process using ant that is externally monitored (web page)
and shows the current goal in a group of goals. The active goal is being
uploaded to elasticsearch every 60 seconds and is allowing us to monitor a
builds progress. So for three goals (A,B & C) we would see something like
"AAABBBBCCCCCC" with A taking 3 minutes, B taking 4 and C
taking 6.
Using this, we are able to manually/visually determine how long each goal
is taking. What we would like to be able to do is query and ascertain
trends/expected times from this data to, for example, allow us to flag when
builds/goals take 20% longer than the average. Is this possible with the
current data set or do we need to change what we upload?
Appreciate any/all help.
(Note: currently using elasticsearch 1.1.2-1 on CentOS 6)
When you say "split the fields out", do you mean take
"AAABBBBCCCCCC" and have it post multiple entries, one for each
60 second snapshot. If so, then that is what is already being done, perhaps
I should have explained that better. Each character was meant to represent
a single entry posted to elasticsearch once every 60 seconds, with "_"
representing not in a build and "A", "B" and "C" for the build steps
themselves.
What we would like to achieve is for timings between not only the first and
last entry for a goal, so 3 for "A", 4 for "B" and 6 for "C", but
potentially the first and last entry for a build, start of "A" to end of
"C", so 13 in this example. The other complication is around having
multiple builds and separating them. Using the collective example, two
builds might appear as "AAABBBBCCCCCC______AAABBBBCCCCCC" (each
character being a single snapshot every 60 seconds), so being able to split
the first build from the second and so on.
Regards
On Thursday, 23 April 2015 02:23:13 UTC+1, Mark Walkom wrote:
You ideally want to restructure your data and split the fields out.
If you can't do it in your code then Logstash would be able to do
something.
On 22 April 2015 at 23:05, KT SSP <kevin...@ssp-uk.com <javascript:>>
wrote:
Hello
We have a build process using ant that is externally monitored (web page)
and shows the current goal in a group of goals. The active goal is being
uploaded to elasticsearch every 60 seconds and is allowing us to monitor a
builds progress. So for three goals (A,B & C) we would see something like
"AAABBBBCCCCCC" with A taking 3 minutes, B taking 4 and C
taking 6.
Using this, we are able to manually/visually determine how long each goal
is taking. What we would like to be able to do is query and ascertain
trends/expected times from this data to, for example, allow us to flag when
builds/goals take 20% longer than the average. Is this possible with the
current data set or do we need to change what we upload?
Appreciate any/all help.
(Note: currently using elasticsearch 1.1.2-1 on CentOS 6)
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.