immediately fills up disk
Closed this issue · 2 comments
I have an 80g disk that I'm running this on a VM with. When I do the polling against a Cisco core switch in our infrastructure, the disk immediately fills up. What data structures is it creating when pollerp.php runs?
The default RRA data structure can be seen (and modified) here:
https://github.com/lozzd/FITB/blob/master/functions.php#L46
The detail level is rather high by default... 60 second accuracy for 2 weeks. The resulting RRD files are 37mb for bits/sec, packets/sec and 74mb for errors & discards/sec.
So if you have a lot of ports, and have chosen a lot of graphtypes, you will definitely fill up the disk extremely quickly.
Of course, by the nature of RRD files, they have to be created to their final size even when they have no data in them. On the upside, they will never grow, on the downside, you have to have a lot of space from the beginning to store the eventual data.
If you don't mind losing accuracy (certainly 2 weeks at 1 second accuracy is rather detailed, perhaps just 1 day would suffice and have averages after that) just change the RRA definition link above in functions.php, and delete the existing rrd files. They'll be recreated much smaller.
Hope this helps!
Sounds like what I was looking for!
Thank you very much!
Ernie A Oporto|Create Your Badge
From: Laurie Denness reply@reply.github.com
To: Ernie Oporto ernieoporto@yahoo.com
Sent: Saturday, November 19, 2011 7:00 AM
Subject: Re: [FITB] immediately fills up disk (#2)
The default RRA data structure can be seen (and modified) here:
https://github.com/lozzd/FITB/blob/master/functions.php#L46
The detail level is rather high by default... 60 second accuracy for 2 weeks. The resulting RRD files are 37mb for bits/sec, packets/sec and 74mb for errors & discards/sec.
So if you have a lot of ports, and have chosen a lot of graphtypes, you will definitely fill up the disk extremely quickly.
Of course, by the nature of RRD files, they have to be created to their final size even when they have no data in them. On the upside, they will never grow, on the downside, you have to have a lot of space from the beginning to store the eventual data.
If you don't mind losing accuracy (certainly 2 weeks at 1 second accuracy is rather detailed, perhaps just 1 day would suffice and have averages after that) just change the RRA definition link above in functions.php, and delete the existing rrd files. They'll be recreated much smaller.
Hope this helps!
Reply to this email directly or view it on GitHub:
#2 (comment)