/test

Cc @jrife for visibility

Hmm I wonder if it's worth checking if SupportsScrub() is true in this probe? We already have a warning if it's not, but I'm thinking users might miss that.

Edit: then again, this would basically make netkit unusable in the meantime since the agent would just not start up on most kernels people would be using today.

Hmm I wonder if it's worth checking if SupportsScrub() is true in this probe? We already have a warning if it's not, but I'm thinking users might miss that.

Hm, for the case of per-endpoint-routes we could actually bail out in that case.. for the non-per-ep-routes we don't need to warn, wdyt?

That makes sense.

Edit: then again, this would basically make netkit unusable in the meantime since the agent would just not start up on most kernels people would be using today.

Ah okay, as I understand, without the patch users would at least have basic connectivity with per-ep-routes, right? If so then yes, we might keep the warning instead. Once the patch hits Linus tree on the next merge win, I'll ping Greg to backport to stable.

Ah okay, as I understand, without the patch users would at least have basic connectivity with per-ep-routes, right?

Yes. It's only when you apply network policies that connectivity issues may arise in some cases.

(CI was green, only changed error text as per @qmonnet's suggestion)

/test