DPUs may manifest in a Systems collection.
If the caller does not specify sysurl, they are almost certainly wanting
to manage the 'main' system,
not an adapter.
Provide for this expectation by
retrying the collection with DPU
filtered out,
so that if there's exactly one non-DPU
system, then it will still be functional.
Change-Id: I300c47245111121440ae0f037d202b074407c45d
The change to make sure resources are freed replaced straightforward
message with a weird one. In the well known context, put the well known
message back in the reasonable path.
Change-Id: Iea17b8bbb7b14255debb05a90d5701f320b5d705
Avail ourselves of generic redfish push
update support, now that
we have had a few other redfish implementations
to see that it in fact does work genericly.
Change-Id: Id68f5774fba973c6c9b24047d89a3ef6ae90dff3
When doing a reseat, the target will tend to unplug
itself before a reply.
In such a scenario, be optimistic about the result.
Change-Id: Iae3aac15b48990e04dacabf94dd9149e9c0c1387
The 'base' device id in a multi-function
device may be considered to be the first,
rather than the last.
Change-Id: I4f39b81fb0225b4221aeffc108c12388cf7a7ea1
The XCC provides hints as to the valid
format of open ended text.
Apply this as a sanity check before accepting a changeset.
Change-Id: I4a97a5a85e4de69fa13894a46dca568cda447705
We may be given a file-like
that cannot seek.
For example, older 'ZipFileExt'
did not have tell/seek.
For such cases, fallback to memory
hungary behavior of using a BytesIO to
provide a trackable
upload.
Change-Id: I401e2fcc7369f31d0d413909854c8d9224b02023
A compound update may contain multiple updates with
json directions on processing.
Activate uxz unwrapping only
with singular payload that ends
in uxz.
Change-Id: I5fb3d158a33f696a7e2fb5143e703de4c9408e7a
XCC updates are now wrapped in .zip for newer platforms,
however the XCC is more robust at handling uxz.
Change-Id: I90d70ab167e44737bc785e4ed88b9f9890a545a7
SMM may reboot at any moment without warning after
finishing an update.
Assume good news if this should happen.
Change-Id: I841780bad9f9ca49b0fbc7c58652e569a3c0a638
OEM providers may provide better/faster health
information.
Allow the override, similar to how override is allowed for ipmi.
This paves the way for some more consistent health data as well as doing better than the redfish
health model, which has some gaps.
Change-Id: Ibb237ca10f724f9f8e54d3a446496d8b13234e5a
The event messages may convey more
specific information. The generic IPMI may
be too vague to know where to proceed next.
Change-Id: I6e0ce8a34d741fa8d4c6c975212f6845755b4f9b
A session being marked broken may
maintain a hold on logonwaiters.
Clear those out in
the process of marking it broken
Change-Id: I6e1ad7f34cd541863e8663cc6fed66ab90d6bf95
The spec declares that if reading type is not threshold and no thresholds declared,
then '0' means the same as '3'.
Change-Id: I1f02ad27af4f2dad286387ed6aa0bdbcc1475fcf
On v3 systems a job is created to handle the last part of the
firmware update process when using Redfish to update the firmware.
Follow the state of that job for the 'validating' phase.
Change-Id: I0a294b687f4e7665684239239cc2f37ca9c76f0c
It may be redundant with SDR
sensors, but stage it for
easier access should it come up
as a needed approach.
Change-Id: I765bbcbcc6c916da2573008bf53f2e88a2a35baa
Using a trial and error method to determine the format of
AddNewVolWithNaAsync causes issues on some server models. When
repeatedly deleting and creating the RAID array on Cyborg with
CDI3A2U XCC firmware we occasionally see hanged web sessions and
on rare occasions an incorrectly formatted AddNewVolWithNaAsync
call will return success and not create the array.
To reduce the number of AddNewVolWithNaAsync calls we determine
when GetDefaultVolProp requires the extra controller slot number
argument and in this case we directly use the
AddNewVolWithNaAsync call with the extra controller type argument
as these 2 changes go hand in hand in the XCC firmware. This logic
reduces the amount of AddNewVolWithNaAsync calls by one and the
issues described above can no longer be observed.
Change-Id: Icdaa5ef0fb5629ad965e37cbbecdcf892f6e315f
Power policy command has
changed. Settings
are now relegated
to the last two bytes,
with status for the first two.
Change-Id: I6505d99fef357d7e2023b7393ad9ef8920b620c7
The request is sufficiently specific that we
do not care about the larger state of the
resource. Explicitly declare this by
indicating a wildcard etag.
Change-Id: Ie0529ffea63516bf025f4ce146e39df40bcc768d
In the ipmi path, provide a null
dictionary to make the check
viable, and stop assuming that
an unknown system is always purley.
Change-Id: I13f7a9f70d83e5679d90acb83e0df1522f441f41
While it must not be used with redfish,
it is ok to use the fast path even though
setting requires IPMI.
Change-Id: I57de25e0ed0bb125adab42916429eca1e29f2392
There are non-PCI adapters that are modeled as 'adapter',
but without corresponding adapter data.
The inventory data is more limited, but we can at least report current
firmware.
Change-Id: I42d17befe1f95fde3b88304418743f1ac456eedd
Purley has an incomplete redfish implementation for uefi settings,
blacklist it from the new way of pushing settings.
Reading settings seems fine, however.
Change-Id: Ie8baf50e11b05e2cafa644c414970ceeaaa35dbf
While IPMI has well defined
roles and redfish has some well-defined
starter roles, permit management
of users for custom roles.
Change-Id: Id6ff87a8ccbb12491adc37b460765d4eadf2c92d
For newer systems, provide access
to the richer configuration in
the proprietary Lenovo interface.
Fall back for standard for older platforms that still require IPMI
for that configuration.
Change-Id: Id66515f00116ce3c22fcba80c0003cfb9e6b1a5d
