On Wed, 20 Mar 2019 08:10:22 +0100 c5244714@gmail.com wrote:
Hello, I'm using NAV 4.9.3 and have a big issue with the trend of interface traffic historical data(trend). I configured to monitor over 100 cisco switches and +2000 interfaces and the trend is consuming lot of data and deplete my storage space. the historical is a great feature with a 1 year history but its too much for us with our limited resources.
My question is if there is anyway to control or limit the period of the data , lets say only 2 weeks with retention
This would be an adjustment to Graphite.
First of all, you will need to change carbon's default storage schemas in storage-schemas.conf [1]. NAV's suggested default here is to keep data for most metrics for 600 days [2]. However, this will only affect new Whisper files that are created, when new metrics appear.
All your pre-existing metrics, backed by Whisper files, already have been pre-allocated 600 days of storage because of the existing storage-schema rules. You will need to resize all of the existing files, using the whisper-resize command line tool, if you want to reclaim your storage space. There are some interesting usage examples for whisper-resizing whole directory trees at [3].
Be aware, however, that you should not resize every whisper file to use the same schema - you have to look at store-schemas.conf to match up the different file trees - it is very important to keep the interval of the first retention archive the same as in the original NAV config values, since these correspond to the NAV's collection interval for these metrics (i.e. changing port traffic metric to have a first retention archive with a 60s interval will cause your traffic graphs to go blank, since the port counters are only collected every 5 minutes).
[1] https://graphite.readthedocs.io/en/latest/config-carbon.html#storage-schemas... [2] https://github.com/Uninett/nav/blob/4.9.x/python/nav/etc/graphite/storage-sc... [3] https://gist.github.com/kirbysayshi/1389254