While confluent does pull in dateutil, pyghmi
itself depends upon it, so it should pull it in
as well.
Change-Id: I3af963f14451bf6efac0c74981c6644df9950d05
Large images take some time before
they can be mounted. Track
the size field to see when the
file is actually done copying.
Change-Id: I23c921798d6be4370e7314822d37fba1cb9faad3
Set flake8-import-order version to 0.17.1
Full py3 compatible version.
Add all Python3 modules to stdlib list.
Change-Id: I3ef01e49c0e6822368146434f056ac7a1aa7ce4d
During the course of applying
configuration changes, the pending set
may change for some reason.
Repeat the logic if etag triggers an exception.
Change-Id: I3783b4ba45dd0dd57fdd3af80af16223e39fce05
For redfish, store the MessageId
for programattic processing.
For identify, use wildcard for
the identify etag. While an implementation required
etag for set identify, it permitted wildcard.
It doesn't matter to check etag on something as trivial as
LED control, so just wildcard it.
Change-Id: If3cf600e46f38858551c5f744388a57393b45123
In python2, cannot directly copy a memoryview to a ctypes
string. Catch and try with a copy of the memoryview
data for python2.
Change-Id: I6dc0183921e4e888aca131fe7968d820b25dc5d5
Commands without data were causing problems
for python2 ctypes interface. Skip the
copy attempt when there is no data to
copy.
Change-Id: If115a1145c277992aa4ffbe420a582dc08d0926e
The new version enables a lot of standard flake8 checks, so a few
fixes are required. W503 is disabled as it conflicts with W504
and the latter seems to be preferred nowadays.
Change-Id: I0d2fa8c88f081d553ebefb5f764397b2761dc2b5
With support for non-zero LUN,
update portions of event log
to be consistent. PET does not
seem to have LUN in the data,
so for the time being just assume 0 there.
Change-Id: I24b6b27f81fe1c24a516d28788a2c0efc2f9ccbb
When get sensor reading says
a sensor is disabled without
also saying it is unavailable,
mark it as unavailable.
Change-Id: I00c61824f3539373b48223da9b37acfed97f6576
Some redfish implementations endlessly mutate
their etag. This makes it impossible to
fully establish atomicity.
Move the logic as much server side as possible.
There is still a window where entries change
between fetching the entries and the loginfo,
but there is little hope.
Change-Id: I9135b96acf82f06bc8622efb4d53d0bc695d1758
So of course it turns out that in redfish
multiple firmware developers implemented
another exactly identical behavior with
different requirements.
Accomodate by trying to detect at least two
variants seen in the industry.
Change-Id: I5aec979f12fd775bcf8570fde411eaee787ba9c3
If the caller provides no custom validation, use normal validation.
Historically, this webclient has been used for internal networks where
custom validation was more feasible than normal TLS validation.
However, it is starting to be used to communicate with internet sites,
so a change is required to work with that model sanely.
Change-Id: I31dae103bfbe534aa326a648f9207ad86b86d550
While XCC is aggressive about expiring tokens, sometimes it isn't.
In such a case, throw a logout request to the 'stale' session in case it
isn't to avoid consuming too many session slots.
Change-Id: I01819786755d5d5d29014c58d4dcbe5468c63607
Newest ipmitool expects a reply to get cipher suites. For
now give a static reply that only cipher suite 3 is supported.
Change-Id: I31f77818ffa746954eb86fa95269b1e5a7e09879
The sentinal value would not get hit with
python3 ('' versus b''). Change to use
an easier to use while loop.
Change-Id: I4d7f4b763aff1fdf13e5a53377146606de199a11
Honor rqlun from client, if not zero. There are BMC implementations
that fail to follow the spec and do rqlun of zero, but pyghmi
implements it correctly. In practice I have never seen an
rqlun of non-zero except as an experiment.
Closes-Bug: #1865987
Change-Id: I73e59672ce5e146cf24386c2228fa8dae3bc579c
Some versions of python cannot take bytes
for the json payload. Workaround by explicitly
decoding from utf8 in such a case.
Change-Id: Idb86003fcf6643e5c67e13202de41be6c35607a0
After adding support for non-zero lun, some reports came back of
dropping responses. However, there is at least one platform that
incorrectly puts 0 in rsLun in replies to packet that had 0 in rsLun.
Workaround by masking out the rsLun value in replies before checking for a
match.
Change-Id: I23a46fce23327ce90d2be845a806d6ba0c4b975e
The IO thread is meant to be a background task. Mark it
as such to be treated properly on interpreter exit.
Change-Id: I73003625d014f823d26c55bdd22217e19f6bf1c9
simplesession represents a stripped down session.
For now it doesn't support many bells and whistles, but it
should suffice for one-off command usage.
Caller has to opt-in to the experimental simplesession by
calling select_simplesession() method in command:
import pyghmi.ipmi.command
pyghmi.ipmi.command.select_simplesession()
Change-Id: I3884762621306d4a895f4b56d173fc80526a095c
The supported wheezy platform we are now adding
pyca to. Additionally, the options were wrong for
such old debian build utilites.
Change-Id: If92de1356cd45c5f4a260b6df72144dbbfe70c3d
XCC convention is to zero-pad the minor number
if less than ten. Support this and make it look less odd.
Change-Id: I093019d68d5c47ffe73d49eba736da5c1bd9ce32
