[Starlingx-discuss] fmClientCli and fm_core usage without keystone authentication

Cordoba Malibran, Erich erich.cordoba.malibran at intel.com
Tue Oct 15 21:00:16 UTC 2019


Hi,

I was playing around with fault components and I noticed that I'm able
to query, create and delete alarms with fmClientCli and the fm_core
python module.

As I understand the `fm` cli (from python-fmclient) creates a request
to keystone before connecting to the fm-rest-api to perform the
operation. However this is not the case for fmClientCli, I can create
an alarm with the following command. 

$ fmClientCli -c "blabla###100.114###set###host###host=controller-0.ntp###2019-10-14 02:28:09.5432###major###This is a dummy alarm created by hand###communication###not-applicable###Monitor and if condition persists, we are sorry###False###True###False###"

and delete it with:

$ fmClientCli -d "###100.114###host=controller-0.ntp"

This without doing the `source /etc/platform/openrc`, I'm even allowed
to create alarms from a non sysadmin user.

The same case with the `fm_core` python module, which links with
libfmcommon as fmClientCli does.

>>> fm_core.get_by_eid("host=controller-0")
['c0d94f45-bc6d-4709-9cc6-6a937659c593###100.114###set###host###host=controller-0.ntp###2019-10-14 02:28:09.005432###major###This is a dummy alarm created by hand###communication###not-applicable###Monitor and if condition persists, we are sorry###False###True###False###']


So, I'm wondering if this an expected behavior or something is missing
in libfmcommon to enable authentication.

Also, do you know if the fmClientCli tool is really used?

Thanks!

-Erich






More information about the Starlingx-discuss mailing list