[atlas] RIPE Atlas and nagios status-checks

Hello,

Update: about half an hour ago we re-enabled this configuration. If you
suddenly get 403 errors from the status check service using nagios/icinga,
then you'll need to change your configuration.

Regards,
Robert Kisteleki


On 2016-11-18 13:39, Robert Kisteleki wrote:
> Dear All,
> 
> The following message may be of interest to you if you're using Nagios (or
> variations like icinga) to query RIPE Atlas for status-checks.
> 
> Yesterday morning our operations team made a change to the service
> configuration such that atlas.ripe.net only answers queries if they have
> proper Host: headers set. This is arguably a good operational practice, so
> we'd like to enforce it.
> 
> However, one of the users running said nagios checks noticed that this
> causes 403 answers instead of the expected behaviour. We concluded that this
> is because the check_http plugin doesn't always set the Host: header.
> According to our statistics in total 12 users are affected, and since we
> could not foresee this problem, we temporarily reverted our configuration to
> the old settings.
> 
> The fix on the client side is easy: one needs to add "-H atlas.ripe.net"
> parameter (or a similar change depending on forks and versions) to the
> checks. We'd like to ask our user to apply this change in the next two weeks
> in order to send well-formatted requests to our service.
> 
> Thank you,
> Robert Kisteleki
> RIPE NCC