Roy Dahan
Roy Dahan
@soyacz are you looking into where the issue is? Why the docker image is picked incorrectly and why the fallback is to a newer/newest image?
According to the comment we have in SCT it shouldn't pick the latest, but the latest of each release (at least this is what I understand). Also, there is a...
@sylwiaszunejko can you please take a look at this one?
@sylwiaszunejko do you mean this is the python-driver issue? https://github.com/scylladb/python-driver/issues/604
Moved the issue to Gemini, need to investigate a memory leakage.
We need to avoid name changes or handle it here so comparison would match. We can't lose history.
I'm fine with either approach as long as we have the stats and able to compare releases after the change
IIUC the thread above we can summarize it with "there is regression with handling enospc in scylla nodes" so we worked around it by disabling testing and ignoring the possibility...
> Since 2025.4 we have the OoS prevention mechanisms (among others, the node rejects writes), if one of the replicas used for read was the node with the critical disk...
I don't see where it can be a driver issue. It's a Scylla issue, one node that reach enospc or critical low space should not prevent QUORUM. Stress should continue...