orangeCatDeveloper opened a new pull request, #4354:
URL: https://github.com/apache/hertzbeat/pull/4354
Part of #3210
Redfish monitoring of Dell iDRAC hardware fails every Fan, Battery and
PowerSupply metric, logging `StatusCode 404` for each chassis on every
collection cycle.
Redfish 2020.4 moved fans and power supplies out of the `Thermal` and
`Power` documents into standalone `ThermalSubsystem/Fans` and
`PowerSubsystem/PowerSupplies` collections. `RedfishCollectionSchema` hardcodes
only the new locations, while shipped iDRAC firmware serves only the older
embedded arrays — #2730 carries the raw response body from a reporter's
machine. A second defect sat behind it: once the paths resolve,
`parseRedfishResource` used a JsonPath parser that throws
`PathNotFoundException` for a field the firmware does not expose, and that
exception escaped `collect()`.
Each resource now carries a list of candidate collection uris, newest schema
first, probed until one yields members. Legacy members are an embedded array
rather than a collection, so the legacy candidate is written as a `#` fragment
(`.../Thermal#/Fans`) — the same JSON Pointer form iDRAC itself puts in
`@odata.id` — and resolving a fragment uri fetches the enclosing document and
slices into it, which leaves the member-detail path unchanged. Because those
member uris then share one document, a per-collect cache keeps a chassis with
sixteen fans from refetching the same Thermal document sixteen times. A failed
candidate keeps its error, so an expired token or a timeout is reported as
itself rather than flattened into `Get redfish resources uri error`, and a
field absent from the firmware becomes a null column instead of failing the
metric.
The second commit fixes an unrelated defect found while reading the same
class: `close()` prefixed the session location with the session service path,
building a URL no request could reach, so sessions were never released and
lingered on the BMC until it expired them.
Deliberately not covered: the `405` on session creation that opens #3210 is
a separate defect needing an HTTP Basic Auth fallback; Battery has no
pre-2020.4 equivalent resource, so it can only fail cleanly; legacy fan speed
lives in `Reading` while `app-redfish.yml` queries `SpeedPercent.*`, leaving
those two columns null until jsonPath can vary per schema; and candidate
selection is global rather than per chassis, which matters only for a host
whose chassis run different firmware generations.
Verified against a mock iDRAC that serves the pre-2020.4 layout from #2730
over TLS, driven through the real collector — session create, chassis walk,
member fetch, session delete — using the jsonPath list shipped in
`app-redfish.yml`.
before:
```
ERROR o.a.h.c.collect.redfish.RedfishCollectImpl -- Get redfish
/redfish/v1/Chassis/System.Embedded.1/ThermalSubsystem/Fans collection resource
error: Redfish session get resource error:StatusCode 404
ERROR o.a.h.c.collect.redfish.RedfishCollectImpl -- Get redfish
/redfish/v1/Chassis/Enclosure.Internal.0-2:RAID.Integrated.1-1/ThermalSubsystem/Fans
collection resource error: Redfish session get resource error:StatusCode 404
collect code = FAIL, msg = 'Get redfish resources uri error'
fan rows = 0
columns = [id, name, state, health, temperature, speed]
session release: FAILED - Redfish session close error:null
```
after:
```
collect code = SUCCESS, msg = ''
fan rows = 2
columns = [id, name, state, health, temperature, speed]
[/redfish/v1/Chassis/System.Embedded.1/Thermal#/Fans/0, System Board
Fan1A, Enabled, OK, , ]
[/redfish/v1/Chassis/System.Embedded.1/Thermal#/Fans/1, System Board
Fan1B, Enabled, OK, , ]
session release: OK
```
Requests the mock BMC received, showing the probe order and the session now
being released:
```
POST /redfish/v1/SessionService/Sessions
GET /redfish/v1/Chassis
GET /redfish/v1/Chassis/System.Embedded.1/ThermalSubsystem/Fans
GET
/redfish/v1/Chassis/Enclosure.Internal.0-2:RAID.Integrated.1-1/ThermalSubsystem/Fans
GET /redfish/v1/Chassis/System.Embedded.1/Thermal
GET /redfish/v1/Chassis/Enclosure.Internal.0-2:RAID.Integrated.1-1/Thermal
DELETE /redfish/v1/SessionService/Sessions/1
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
