fault is left null in certain cases about openstack4j HOT 9 CLOSED

So I found a change in behavior :
if i do a
List<? extends Server> servers = os.compute().servers().list(true);
Then parse the list the servers that have faults are still having fault == null
but if I do :
Server srv = os.compute().servers().get(serverId);
then fault is properly updated.

The strange thing is that the first case works on another region ...
any idea ?

from openstack4j.

@bmaury-bouet, thanks for the report. Please turn on the wire logging and verify whether the field is part of the server response in both cases you mention. You are calling the list with details on, so that should not be the cause.

from openstack4j.

Hi @olivergondza,
I enabled the wire, and when I compare the 2 calls ( the json from the full list and the json from the single server) there indeed a difference and the only field missing is "Fault" (at least in this case).

from openstack4j.

@olivergondza please note that on the other region (main region) I do see the "faut" in the json when calling list() (even with detailed==false)

from openstack4j.

@bmaury-bouet, that sure is interesting. Any difference in endpoint version in the other region? Different configs?

How does the python client present the fault in case it is not being served via REST?

from openstack4j.

Hi @olivergondza unfortunately i'm don't have the full details on the different region I'll try to poke my colleagues in charge of the IT.
With python cli using "server list" or "server list --long" i never see the fault on both regions it doesn't show the fault I have to use server show {id} on both region to get the "fault" information

from openstack4j.

@olivergondza just got word from our IT departement
So our "main" region is running on Newton version
and the new region runs on Stein.
So there might be a difference in behavior between the 2

from openstack4j.

Ok, so newton[1] have nova 14.1.0, stein[2] have nova 19.3.1. Which is quite a difference. I also understand it is the outdated version where this misbehaves. I have found no indication in the changelog of why would this change in between the versions.

I am closing this as not a problem of openstack4j - the server simply does not serve the needed data for whatever reason. Mind that Newton went EOL 2017-10-25.

[1] https://releases.openstack.org/newton/index.html
[2] https://releases.openstack.org/stein/index.html

from openstack4j.

Sure no problem, we can easily bypass the issue.
I just wanted to make sure if could come from a bug on your end.
Thanks for the support @olivergondza !

from openstack4j.