Failure in the actual browbeat run about ovn-heater HOT 13 CLOSED

What fails is this:

rally-ovs deployment destroy ${deployment}

Line 442 in ./do.sh

I am going to debug it further and see what is the root cause.

from ovn-heater.

It fails to destroy it - tested on both F32 and F33 with the same error. If we step through to the next step in do.sh, that fails because it finds an existing deployment.

from ovn-heater.

I can't figure out exactly what generates it and rally somehow manages to disable normal python error output. There are no files and no line numbers when it encounters an error.

This is reproducible with both F32 and F33. I have played with all the usual suspects - dns, selinux, etc. It is not related to them and it happens every time. Ideas?

from ovn-heater.

@kot-begemot-uk, this should be fixed with the following PRs:
ovn-org/ovn-scale-test#217
dceara/rally#2

Could you, please, try the following out:

rm -rf runtime/
RALLY_OVS_REPO=https://github.com/dceara/ovn-scale-test RALLY_OVS_BRANCH=fix-deployment-task ./do.sh install
./do.sh rally-deploy
./do.sh browbeat-run browbeat-scenarios/switch-per-node-low-scale.yml test

from ovn-heater.

@kot-begemot-uk On Fedora 33, you might want to make sure that the SSH keys you use are secure enough, e.g., the following works:

ssh-keygen -t ed25519 -a 64 -N '' -f ~/.ssh/id_ed25519
cat ~/.ssh/id_ed25519.pub >> ~/.ssh/authorized_keys

from ovn-heater.

It is still broken from a clean install.

1. Starting from clean freshly installed F32 - the insecure registry config to the nodes is not deployed. ./do.sh install fails with an error
2. After deploying it manually install and deploy succeed. However, after that the actual rally run has an error. Here is what can be seen in the report:

running docker container ls on all test test nodes shows an empty list. Master node shows only registry

from ovn-heater.

ON SSH keys - I already ran into that. It did not like the authorized keys I transferred from my setup.

In addition to generating the key locally, the other option is to disable selinux altogether (keys transferred via scp do not have the correct selinux attributes).

from ovn-heater.

from ovn-heater.

It fails with the attached browbeat file which is the scale I need to run it on.

Hmm, something probably got lost on the way, I don't see any attached file. Could you, please, add it inline here or drop me a direct email?

Thanks,
Dumitru

from ovn-heater.

switch-per-node-12-600.zip

from ovn-heater.

switch-per-node-12-600.zip

You need the following change to your browbeat file:

--- browbeat-scenarios/switch-per-node-12-600.yml       2021-04-13 08:11:34.000000000 -0400
+++ browbeat-scenarios/switch-per-node-12-600.yml.new   2021-06-01 04:41:03.615420017 -0400
@@ -67,6 +67,6 @@
         max_farm_nodes: 12
         farm_nodes: 12
         ports_per_network: 600
-        cluster_cmd_path: /root/ovn-scale-testing/runtime/ovn-fake-multinode
+        cluster_cmd_path: /root/ovn-heater/runtime/ovn-fake-multinode
         ovn_monitor_all: true
         ovn_central_ip: "192.16.0.1-192.16.0.2-192.16.0.3"

from ovn-heater.

from ovn-heater.

ovn-org/ovn-scale-test#217 got merged, and based on #26 (comment) closing.

from ovn-heater.