Fixed.
For future reference, these ASF VMs don't allow home directory
~/.ssh/id_rsa.pub keys. You have to have root and stick the authorized
keys in at /etc/ssh/ssh_keys/.pub .
root@couchdb-vm:/etc/ssh/ssh_keys# ls -la
total 24
drwxr-xr-x 2 rootroot 4096 Jul 26 20:57 .
drwxr-xr-x 5 root
Ah, that's a missing pubkey in the jenkins user on repo-nightly, not
that the hostkey changed. I thought I got that, but apparently not. I'll
fix it today.
-Joan
On 2020-07-26 7:05 a.m., Jan Lehnardt wrote:
Hey all,
we are getting CI fails on master because the host key for repo-nightly
Hey all,
I’m happy to present the first PR worth sharing for introducing per-doc-access
control to the 3.x codebase.
https://github.com/apache/couchdb/pull/3038
There are few odds and ends left to do, but this is in good enough shape to get
wider review on approach and implementation so
Both approaches will collate keys the same way, it’s a showstopping bug if they
don’t match existing couch collation order.
I’d probably not even document that right now. We can document what aegis does
and doesn’t cover when we make the first 4.0 RC.
--
Robert Samuel Newson
Heya Bob,
this sounds like a good path forward that allows us to validate the remaining
pieces of this.
Just two side notes:
- do we expect that the map-keys-in-fdb approach would sort differently from
map-keys-in-ebtree? (I assume no)
- I further assume that we’ll find a nicer UX for “make
I'd like to find consensus on a path forward and, with that in mind, I have a
suggestion.
Before that, I think it's useful to lay out the points of contention, which
I've ranked in I hope an uncontroversial order;
The principal concern with using ebtree for map as well as reduce is that it
Hey all,
we are getting CI fails on master because the host key for repo-nightly changed.
I’m not sure how to fix this:
https://ci-couchdb.apache.org/blue/organizations/jenkins/jenkins-cm1%2FFullPlatformMatrix/detail/master/209/pipeline
[2020-07-26T10:55:22.484Z] + rsync -avz -e ssh -o