Peerdiscovery tests are failing.
Closed this issue ยท 17 comments
The Jenkins job for bootstrap server liveness is failing:
[Test_BootstrapServers] $ /bin/sh -xe /tmp/jenkins[...].sh
+ /usr/bin/python3 py-ipv8/stresstest/bootstrap_rtt.py
Traceback (most recent call last):
File "py-ipv8/stresstest/bootstrap_rtt.py", line 23, in <module>
from ipv8.util import create_event_with_signals
ImportError: cannot import name 'create_event_with_signals' from 'ipv8.util' (/home/jenkins/.local/lib/python3.8/site-packages/ipv8/util.py)
We are now flying blind. This should be fixed ASAP.
EDIT: The bootstrap job is now running again. The peerdiscovery jobs are still broken.
The same error appears for job/performance_peer_discovery and performance_peer_discovery_defaults.
I tried wiping out the workspace but the job is not even running anymore. We may be out of disk space again.
The badges have also disappeared from our readme.
Maybe it's using the IPv8 from PyPI which doens't have this new function?
I can confirm that the file exists with the appropriate function in the workspace.
At this point, I think its a Jenkins issue.
Sure /home/jenkins/ci/workspace/ipv8/performance_peer_discovery/ipv8/util.py
exists, but it seems to be looking for create_event_with_signals
in /home/jenkins/.local/lib/python3.9/site-packages/ipv8/util.py
.
Ah, good catch. I guess someone installed IPv8 on bbq
recently?
Regarding the badges disappearing: thanks @devos50 for cleaning up some experiment data. That part of the issue now seems solved.
Judging by the logs, between 3 and 4 PM yesterday this failure started.
Ah, it's not just bbq
: himba-linux-builder
has the same issue.
If you set the Python path during the job, shouldn't everything be fine? Something like export PYTHONPATH=$(pwd):${PYTHONPATH}
@egbertbouman that works, I'll do that for all failing jobs.
Alright sitrep:
- ๐ Badges
- ๐ PR tests
- ๐ Periodic
Test_BootstrapServers
- ๐ซ Periodic
performance_peer_discovery
: cannot apply configuration changes. - ๐ซ Periodic
performance_peer_discovery_defaults
: cannot apply configuration changes.
At least our bear minimum critical infrastructure is now operational. I'll scale this issue down to high.
According to StackOverflow the last two jobs being "unconfigurable" is because of outdated plugins. We would have to upgrade and restart Jenkins to fix this.
Many thanks to @kozlovsky for figuring out which plugins needed an upgrade.
We are back up and operational! Thanks to everyone who helped fix this!