SEL Event logs in Redfish

[Lei Yu]

On Wed, Jun 2, 2021 at 3:05 AM Bills, Jason M
<jason.m.bills at linux.intel.com> wrote:
>
>
>
> On 5/28/2021 1:58 AM, Jayashree D wrote:
> > Classification: Public
> >
> > Hi Jason,
> >
> > In journalctl log, I found that the sensor value is crossed the threshold value.
> > But when I tried in Redfish API, I am not able to see any logs, whether anything I am missing here ?
> > Could you please provide suggestions on this .
> Unfortunately, we ran into trouble trying to persist the journal to
> flash, so we have one additional layer where rsyslog watches for any
> journal entries that have a "REDFISH_MESSAGE_ID" metadata and logs them
> to '/var/log/redfish' where bmcweb pulls them for the Redfish EventLog.
>
> https://github.com/openbmc/openbmc/blob/master/meta-intel-openbmc/meta-common/recipes-extended/rsyslog/rsyslog/rsyslog.conf#L66
>
> If you have the journal event, I'd check if you have 'var/log/redfish'
> populated with that event.  If not, you may need to add the above config
> to rsyslog to create it and that will hopefully fix the event in Redfish.

There is a limitation about the way to log SEL event to
/var/log/redfish, that it only appears in redfish event logs, but not
in ipmi sel.

On g220a it still uses the logging service to represent the event log
(see [1]), and ipmid is updated to consume the logging entries to get
the SEL entries.
The missing part is in bmcweb that it does not support the event log
entries from DBus. We could probably add some new code to make it
support consuming data from logging service, and it could be
implemented under a config, so that the user could config which data
source to get event log at compile time.

[1]: https://github.com/openbmc/phosphor-dbus-interfaces/blob/master/xyz/openbmc_project/Logging/SEL.metadata.yaml

-- 
BRs,
Lei YU