| GDNSD-PLUGIN-MULTIFO(8) | gdnsd | GDNSD-PLUGIN-MULTIFO(8) |
gdnsd-plugin-multifo - gdnsd plugin for multi-address, all-active failover groups
Example plugin config:
plugins => {
multifo => {
up_thresh => 0.3
service_types => up,
v4www => {
lb01 => 192.0.2.200,
lb02 => 192.0.2.201,
lb03 => 192.0.2.202,
}
v6smtp => {
service_types => [ smtp ],
up_thresh => 0.1,
lb01_v6 => 2001:DB8::1,
lb02_v6 => 2001:DB8::2,
lb03_v6 => 2001:DB8::3,
}
pubwww => {
up_thresh => 0.5
service_types => corpwww_type
addrs_v4 => [ 192.0.2.100, 192.0.2.101, 192.0.2.102 ]
addrs_v6 => {
service_types => [ up ],
up_thresh => 0.7
lb01_v6 => 2001:DB8::1,
lb02_v6 => 2001:DB8::2,
lb03_v6 => 2001:DB8::3,
}
}
}
}
Example zonefile RRs:
web4 180 DYNA multifo!v4www smtp 180 DYNA multifo!v6smtp www 180 DYNA multifo!pubwww
gdnsd-plugin-multifo is designed to do multi-address all-active failover grouping. Basically, for each configured resource name, you supply a labeled list of addresses. multifo monitors these addresses according to "service_types", and answers "DYNA" address queries using the non-"DOWN" subset. The core gdnsd code will round-robin rotate the records on the way out, as it does for all address RR-sets.
At the top level of the plugin's configuration stanza, three special parameters "up_thresh", "service_types", and "ignore_health" are supported. These set default per-resource options of the same name for any resources which do not define them explicitly.
The rest of the hash entries at the top level are the names of the resources you define. Each resource gets a configuration hash of its own for containing resource-specific parameters as well as the actual address data.
Within a resource, you have two basic options. You can either directly specify a set of "label => address" pairs which are all the same family (IPv4 or IPv6), or you can use the sub-stanzas "addrs_v4" and/or "addrs_v6" to specify one or both families in the same resource.
The "up_thresh", "service_types", and "ignore_health" parameters are inherited through every level, and can be overridden at any level (even per-address-family):
If you have no parameters (service_types, up_thresh, ignore_health) to configure in a given stanza (single-family direct resource config, or addrs_v[46]), and do not care about the descriptive per-address labels used in monitoring, you can replace the hash with an array of addresses. The labels will be generated for you as a series of integers starting with 1. For example, the following are equivalent:
res1 => { addrs_v4 => [ 192.0.2.1, 192.0.2.2 ] }
res1 => { addrs_v4 => { 1 => 192.0.2.1, 2 => 192.0.2.2 } }
All of the addresses for all of the resources are monitored using the per-address-family inherited "service_types" specified (default would be the static virtual monitor "up"). When the core daemon requests a lookup for address records of a given family on one of this plugin's resources, it goes through essentially the following process to determine the set of response addresses for that address family: 1) Add all non-DOWN addresses to the result set. 2) If the set of non-DOWN addresses fail the up_thresh check, add *all* addresses to the result set as a fallback. 3) If any address is in the DOWN state, cut the zonefile-specified TTL in half.
If "ignore_health" is true, all addresses are added to the result set regardless of health, but the up_thresh and TTL effects still happen, and the final resource-level state still reflects the overall state as it would without "ignore_health".
This process is repeated independently for each of the IPv4 and IPv6 address subsets, in the case that a resource has both address families configured (the TTL is only cut in half once of course). Details on the up_thresh check follow:
If there are not enough UP addresses to pass the threshold (per address family), all addresses (of a given address family) will be returned as a fallback.
The threshold is implemented mathematically as in the following pseudo-code "if(non_down >= ceil(thresh * total)) threshold_passed;". For example, if thresh is at the default value of 0.5, and there are 3 total IPv4 addresses, then 2 of them must be non-down to pass the threshold. The net result is that with the default threshold, the plugin will never return an isolated single address from a set of 3. It will either return all 3, or it will return 2/3 if a single address from the set has failed.
When the threshold check fails (and all addresses are returned) for either address family, resource-level total failure will also be signaled to any applicable upstream meta-plugins such as metafo or geoip.
General rules for the results of the up_thresh formula:
Intermediate value examples: (threshold: non-down/total required to pass threshold):
gdnsd.config(5), gdnsd.zonefile(5), gdnsd(8), gdnsd-plugin-simplefo(8)
The gdnsd manual.
Copyright (c) 2012 Brandon L Black <blblack@gmail.com>
This file is part of gdnsd.
gdnsd is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.
gdnsd is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with gdnsd. If not, see <http://www.gnu.org/licenses/>.
| 2021-02-24 | gdnsd 3.5.2 |