Side effects of --no-ip-validation

[lenzls]

Hello,

I want to see if I manage without logging IP addresses on my server.

I'm planning to use ipscrub, to log only hashes of IP addresses. To do that, I assume I have to use --no-ip-validation as indicated in #1354.

I wonder, though, what are the costs of --no-ip-validation. I guess that geolocation doesn't work any more (unless with a workaround like #1089).
But will features like "unique visitors" still work with only the hash instead of the IP address?
Are there any other draw-backs of using --no-ip-validation?

Kind regards,
Simon

[allinurl]

Great question! If using --no-ip-validation, you would still need a placeholder for the %h field. The trade-off is accuracy. Very likely you will get a smaller number of visitors. I'd say give it a shot using a smaller number of requests, with and without IPs. Let me know what you get.

[lenzls]

Thanks for your reply. Small Update:

For now, I decided against hashing of the IP. Reason is, that there is an open issue in ipscrub which I don't understand masonicboom/ipscrub#11.

Instead, I went with masking the last few bits of the IP address. As those are still valid IPs, I don't have to use the --no-ip-validation flag.

Note, I was surprised that the --anonymize-ip flag seems to only affect the rendered report, and not the persisted data (--persist). Since I don't want to indefinitely save non-masked IPs, I'm now doing the masking already on nginx side when writing the access.logs. But that is another topic 🙂