On 19/11/18 21:30 +0000, jose.delar...@dell.com wrote:
Do we need to use a CLI tool to implement a fence agent? I rather stay away
from redfishtool to avoid having it as a dependency (and Python v3) and would
rather just use the requests library directly.
No need to use CLI tool. We actually prefer using libraries where
available.
I created a fence agent for redfish, the agent can take as argument the Systems
URI or can programmatically find it as it will be different across OEMs:
https://github.com/jose-delarosa/fence-agents/blob/add_redfish_fence_agent/agents/redfish/fence_redfish.py
Looking for initial feedback please.
You're welcome to create a PR against our ClusterLabs repo, where we
can discuss regarding improvements. You should rename it to .py.py, so
that make replaces @PYTHON@ and other variables detected during
configure.
It's GPLv2 by default, so add a license header if you want it licensed
under another license.
I usually don’t use tabs to indent lines, is this a requirement?
We just stay consistant with what indentation the code comes with, so
no need to worry about tab vs space for indentation.
Jose Delarosa
Linux Engineering
Dell EMC|Infrastructure Solutions Group
-----Original Message-----
From: Developers [mailto:developers-boun...@clusterlabs.org] On Behalf Of
Andrei Borzenkov
Sent: Thursday, November 15, 2018 10:09 PM
To: developers@clusterlabs.org
Subject: Re: [ClusterLabs Developers] [fence-agents] How to "login" other than
ssh or telnet?
[EXTERNAL EMAIL]
16.11.2018 0:31, jose.delar...@dell.com<mailto:jose.delar...@dell.com> пишет:
All,
I am looking at writing a fence agent for Redfish, which uses REST APIs to run
"commands" like power off/on/status. There is no ssh login per se, as the APIs
are accessed through HTTP methods (GET/POST/PATCH, etc.), so none of these can be used:
https://github.com/ClusterLabs/fence-agents/blob/master/lib/fencing.py
.py#L968-L976
<https://github.com/ClusterLabs/fence-agents/blob/master/agents/autode
tect/fencing.py#L888-L896> I'm new to this project, so looking for
insightful feedback from the project maintainers or more experience developers here as to
the best way to implement a fence "login" for Redfish.
python redfishtool.py -r <rhost> -u <username> -p <password> Systems reset
ForceRestart
At which point you can take any of existing fencing/stonith agent as template
and fill in actual command.
https://github.com/DMTF/Redfishtool
Also, is anyone here also looking at developing a Redfish fence agent?
I see https://github.com/ClusterLabs/fence-agents/issues/230, but not much
activity, happy to collaborate.> Thank you all.
Jose Delarosa
_______________________________________________
Developers mailing list
Developers@clusterlabs.org<mailto:Developers@clusterlabs.org>
https://lists.clusterlabs.org/mailman/listinfo/developers
_______________________________________________
Developers mailing list
Developers@clusterlabs.org<mailto:Developers@clusterlabs.org>
https://lists.clusterlabs.org/mailman/listinfo/developers
_______________________________________________
Developers mailing list
Developers@clusterlabs.org
https://lists.clusterlabs.org/mailman/listinfo/developers
_______________________________________________
Developers mailing list
Developers@clusterlabs.org
https://lists.clusterlabs.org/mailman/listinfo/developers