Opened 16 months ago

Last modified 16 months ago

#3981 new change

(Re-) Configure per-server traffic monitoring

Reported by: matze Assignee: matze
Priority: P2 Milestone:
Module: Infrastructure Keywords:
Cc: fhd, fred, darkue Blocked By: #286
Blocking: Platform: Unknown / Cross platform
Ready: yes Confidential: no
Tester: Unknown Verified working: no
Review URL(s):

Description

After migrating to a more advanced DNS backend in the context of #286 (amongst others), which includes better monitoring, notifications and advanced balancing features, we discovered and tackled several traffic and routing issues already (each of which was obscured by proprietary implementation without monitoring or notification on our former provider's side).

In the same context, and with respect to recent upgrades in our server provider's network and monitoring, we realized that we have to find and configure individual traffic boundaries for each server ourselves. Otherwise there won't be any notification in case of peaks, but occasional network up-link shutdowns (which we sometimes even misinterpreted as server hardware issues).

This measure will help us detecting similar issues in the future before they become a visible to the user (see i.e. #3540). Note, however, that these values need to be configured with two distinct resources, our provider's GUI/API and our own Nagios-related deployments. There is no facility for sharing these bits in automated fashion yet, but a future introduction of such may follow a proof of concept using this very example..

Change History (1)

comment:1 Changed 16 months ago by matze

The balancing weights have been adjusted, the actual thresholds in monitoring are about to follow.

Note: See TracTickets for help on using tickets.