[bmcweb] Device Level Logging Services
Bills, Jason M
jason.m.bills at linux.intel.com
Thu May 27 08:12:45 AEST 2021
On 5/26/2021 1:50 PM, Willy Tu wrote:
> Hi all,
>
> Main discussion:
> https://lore.kernel.org/openbmc/CAHwn2XknKWyv6Bsq48tkvA52rHKZRFoEJgUuZUmMj6uZb584JQ@mail.gmail.com/ (a
> bit outdated)
>
> This will be a sub-thread about logging services that will be used to
> support Drive (and could support others).
>
> For bmcweb, I am wondering if we can support custom logging services
> other than the static ones that we have in
> https://github.com/openbmc/bmcweb/blob/master/redfish-core/lib/log_services.hpp.
>
> I am thinking about supporting custom log service at a device level that
> can be exported by different hardwares log entries that are more
> specific. Then at each device resource can link the Status.Condition to
> the LogEntries.
>
> Initial Idea,
>
> For example, I want to have log entries for Drive devices.
>
> ComputerSystemId could still be system or bmc and the Redfish
> endpoint will be something like.
>
> ```
> /redfish/v1/Systems/system/LogServices/Drive0LogService/Entries
> ```
Specifically regarding the endpoint, we have considered a similar
approach for publishing a LogService for a memory device. I don't know
if we have finalized our proposal to the DMTF yet, but we were thinking
it would be better to add a LogService to the individual resource rather
than keep everything under Systems/system.
If the same idea is applied for your drive devices, each drive resource
in Redfish would also contain its own LogService.
>
> On the D-bus side, the service exporting the log can add log entry
> <https://github.com/openbmc/phosphor-dbus-interfaces/blob/master/xyz/openbmc_project/Logging/Entry.interface.yaml> under.
>
> We can have
> ```
> /xyz/openbmc_project/logging/device/drive_0
> |__/xyz/openbmc_project/logging/device/drive_0/entry/...
> ```
>
> We can parse the `xyz.openbmc_project.ObjectMapper` at
> `/xyz/openbmc_project/logging/device` with depth of 1 to get all the new
> device log services and so on.
>
> Please let me know if there are any questions about it. If no one is
> against it, I'll work on implementing it.
>
> Best,
>
> Willy Tu
>
>
>
>
> Best,
>
> Willy Tu
More information about the openbmc
mailing list