Configurable timeout for Hostname Resolver

Description

The provisioning(sync) of requisitions takes a long time because of the DNS reverse lookup if your domain controllers are faulty configured or not reachable.

The DNS reverse lookup takes about 36 seconds per node if your DNS server is faulty configured (tested with Meridian 2017.1.9) and up to 1 minute if the dns servers are not reachable (tested with Meridian 2018.1.11).

It would be helpful if you could set the timeout for the hostname resolver yourself.
I have attached a file with an excerpt of the provisioning problem.

Environment

Red Hat Enterprise Linux 7 x86_64

Acceptance / Success Criteria

None

Attachments

1

Lucidchart Diagrams

Activity

Show:

Jesse White January 6, 2020 at 2:49 PM

The new DNS resolver is only used by telemetryd, details here: http://docs.opennms.org/opennms/releases/25.1.1/guide-admin/guide-admin.html#ga-dnsresolver

Modules including provisiond have not been updated to use it yet - this could be an opportunity to do so though.

David Hustace January 6, 2020 at 12:00 PM

A lot of work was done making enhancing OpenNMS' capability to handle thousands of DNS queries per second in H24. Since Java uses the system's resolver libraries for host name resolution, the application is therefore dependent on the options settings in resolv.conf: attempts, timeout.

Unsure if the updates in H24 are of any help here, perhaps you can test with H24 since M2017 is based on a much much older code base.

Details

Assignee

Reporter

Components

Affects versions

Priority

PagerDuty

Created December 20, 2019 at 12:22 PM
Updated September 21, 2021 at 9:27 PM