ipmi_exporter: "Failed to parse bmc-info" when the system does not provide "System Firmware Version"

On my Supermicro servers, I’m getting the following error with the bmc collector enabled:

ERRO[0004] Failed to parse bmc-info data from [local]: Could not find value in output: Device ID             : 32
Device Revision       : 1
Device SDRs           : unsupported
Firmware Revision     : 1.52
Device Available      : yes (normal operation)
IPMI Version          : 2.0
Sensor Device         : supported
SDR Repository Device : supported
SEL Device            : supported
FRU Inventory Device  : supported
IPMB Event Receiver   : supported
IPMB Event Generator  : supported
Bridge                : unsupported
Chassis Device        : supported
Manufacturer ID       : Super Micro Computer Inc. (10876)
Product ID            : 6937
Auxiliary Firmware Revision Information : 00000000h

Device GUID : 00000000-0000-0000-0000-000000000000

System GUID : 00000000-d43a-46ef-ec3c-534d31303136
…

As you can see the BMC on SMCI doesn’t always provide a System Firmware Version - can we make this optional?

About this issue

Original URL
State: closed
Created 4 years ago
Comments: 21 (20 by maintainers)

Commits related to this issue

Add a `bmc-device-id` collector When the `bmc` collector was extended to attempt to collect the system firmware revision the resulting IPMI request broke on some older systems. This collector can be ... — committed to prometheus-community/ipmi_exporter by bitfehler 3 years ago
Try to parse bmc-info even if command failed This is a workaround for an issue described in #57. The bmc-info command might produce usable output minus the system firmware revision, but then choke on... — committed to prometheus-community/ipmi_exporter by bitfehler 3 years ago

Most upvoted comments

Thanks for the report! This is certainly a bit problematic, I had not thought of this scenario. I will try to release a fixed version ASAP

bitfehler on Sep 24, 2020

Oh my. Thanks for the detailed report. Give me a moment to figure out how to solve this mess somewhat gracefully…

bitfehler on Oct 26, 2020

Just because we love IPMI so much… I have an old SMC host that does not seem to like running bmc-info without the --get-device-id param, which was removed in https://github.com/soundcloud/ipmi_exporter/commit/2c6a73b1510cb1b6095feb5f18ed5d1da9823616#diff-f1c80b360812fd52bef43d56f3c4ef40853f02711810857332ccedfd9f463f79L291

$ bmc-info -l user -u MONITOR -p $password -h tank-ipmi
Device ID             : 32
Device Revision       : 1
Device SDRs           : unsupported
Firmware Revision     : 3.29
Device Available      : yes (normal operation)
IPMI Version          : 2.0
Sensor Device         : supported
SDR Repository Device : supported
SEL Device            : supported
FRU Inventory Device  : supported
IPMB Event Receiver   : supported
IPMB Event Generator  : supported
Bridge                : unsupported
Chassis Device        : supported
Manufacturer ID       : Nedam ENG. Co., ltd. (47488)
Product ID            : 43025

Device GUID : 00000000-0000-0000-0000-000000000000

System GUID : 00000000-dfad-2990-2500-534d31303037

ipmi_cmd_get_system_info_parameters_system_firmware_version_first_set: privilege level insufficient

The fact that bmc-info exits with status code 1 means that the collector is treating it as a failed command, which is fair enough. But it results in the ipmi_bmc_info metric disappearing for that host. We don’t really want to use a higher privilege level for monitoring.

Edit: Running bmc-info without --get-device-id, or running it explicitly with --get-system-info will trigger the failure. Also, using a higher privilege level user fails, with the same error, e.g.

$ bmc-info -l ADMIN -u ADMIN -p $adminpw -h tank-ipmi --get-system-info
ipmi_cmd_get_system_info_parameters_system_firmware_version_first_set: privilege level insufficient

Versions of freeipmi tested: 1.6.3, 1.6.4.

dswarbrick on Oct 22, 2020