Possible memory leak when monitoring large number of prefixes
See original GitHub issueHello,
Before anything, thanks for this project. I’ve been watching it for a while, and now finally deployed it… I hope I’ll be able to contribute to at some point, for now, at least, with some reports.
Describe the bug
Well, it is not exactly a bug (or maybe it is), but I’ve noticed the memory is steadily increasing. In 24 hours after deploying the app, the memory footprint grew from 150MB to over 800MB, and it looks like it’s only going to be getting worse - see screenshot below:
The CPU usage seems pretty stable however.
I’m yet to play with Node Js, but in general a pattern like that means there’s a possibility of a memory leak somewhere.
Provide an example Provide an example in terms of prefixes and BGP messages. Possibly provide a snippet of config.yml and prefixes.yml.
Nothing really specific, pretty much all the prefixes from https://bgp.he.net/AS14061#_prefixes and https://bgp.he.net/AS14061#_prefixes6, e.g.,
"5.101.96.0/21":
description: "Some description"
asn: 14061
ignoreMorespecifics: false
ignore: false
3581 prefixes in total.
Expected behavior
Given the high number of prefixes, I’d expect to see a pretty important memory footprint, but I find an always growing figure more worrying because it’s not easily predictable how much resources you’re going to need on the long term.
Are you using the binary or the source code?
A custom Docker image based on nttgin/BGPalerter, so source code.
Your information Provide your name and your AS/company (see Bert Hubert’s post https://berthub.eu/articles/posts/anonymous-help/).
AS14061, DigitalOcean.
Cheers!
Issue Analytics
- State:
- Created 4 years ago
- Comments:19 (19 by maintainers)
I just want to say you guys rock 😃
On Tue, Jan 21, 2020 at 23:03 Massimo Candela notifications@github.com wrote:
To be clear, the frequent re-announcements are real independently from the duplication. So there there are 2 phenomena:
The RIS team is working on the second one!