Hi all,
We are pleased to announce the release of DNS-STATS Compactor version 1.2.1-beta2.
Updates since 1.2.1-beta1:
* Bump the private version because schema extended with new stats.
* Add new stats to C-DNS: sniffer drops, matcher drops and sampling discards,
and 3 libpcap stats. The libpcap stats should not be considered reliable and
provided for information only.
* Fix file durations for files with more than one C-DNS block in them so high
traffic rates can be correctly calculated from the statistics.
* Constrain size of matcher queue to reduce memory consumption when
dropping/sampling.
* Add `-D` option (--relaxed-mode). With this option a warning (not an error) is
issued for unrecognized command line and config file options.
*Please note the change of PPA - Bionic packages are available for this release*
For more information, please see the project wiki:
https://github.com/dns-stats/compactor/wiki.
Packages available from: ppa:dns-stats/compactor-unstable-bionic
Source code available at: https://github.com/dns-stats/compactor
Documentation: https://github.com/dns-stats/compactor/wiki/user-guide-1.2.1-beta2.pdf
Sara.
Hi,
I'm running NSD and Unbound on Kubernetes, and I wanted to run the DNS
Stats Collector alongside them. To this end I created a Docker image for
Collector (https://github.com/SeanBurford/dns-stats-docker) I'm sharing
it here in case others find it useful.
There are README.md files throughout the tree recording my thoughts, but
here are the main points:
- Dockerfile (
https://github.com/SeanBurford/dns-stats-docker/blob/main/compactor/Dockerf…)
contains the directions used to build the binaries and then the container.
- bin/start_compactor.sh adds the hostname to the output file template,
which is handy when you're running multiple instances on Kubernetes (they
all have unique hostnames).
- I use a Unix socket to pass dnstap messages from the name servers to
the compactors, with one compactor instance per name server instance.
- I'm using NFS to mount a shared log directory across the instances,
though there are plenty of options for cluster filesystems.
- I would have preferred to use Alpine Linux, however various
dependencies made Ubuntu (Focal) an easier choice. This is fine, Alpine
Linux containers run alongside Ubuntu containers in a single Kubernetes pod
(sharing the dnstap Unix socket).
Feel free to send pull requests or suggestions.
--
Thanks,
Sean Burford
