kcsbridge: Intermittently not working

[leiyu-bytedance]

We tried to enable the kcsbridge in this repo on the meta-g220a system.
It works better than the openbmc/kcsbridge.
But intermittently we get the below errors:

Apr 16 09:03:26 g220a kcsbridged[626]: Canceling outstanding request
Apr 16 09:03:34 g220a kcsbridged[626]: Canceling outstanding request
Apr 16 09:03:39 g220a kcsbridged[626]: Canceling outstanding request
Apr 16 09:03:44 g220a kcsbridged[626]: Canceling outstanding request
Apr 16 09:03:49 g220a kcsbridged[626]: Canceling outstanding request
Apr 16 09:03:54 g220a kcsbridged[626]: Canceling outstanding request
Apr 16 09:03:59 g220a kcsbridged[626]: Canceling outstanding request
Apr 16 09:04:04 g220a kcsbridged[626]: Canceling outstanding request
Apr 16 09:04:09 g220a kcsbridged[626]: Canceling outstanding request
Apr 16 09:04:14 g220a kcsbridged[626]: Canceling outstanding request
Apr 16 09:04:19 g220a kcsbridged[626]: Canceling outstanding request
Apr 16 09:04:24 g220a kcsbridged[626]: Canceling outstanding request
Apr 16 09:04:29 g220a kcsbridged[626]: Canceling outstanding request

And when the issue occurs, the inband IPMI from OS is blocked.
The issue is recovered after a while though.

• When the issue occurs

# time ipmitool mc info
No data available
Get Device ID command failed
Get Device ID command failed: Invalid command

real    0m21.777s
user    0m0.000s
sys     0m0.000s

• When it's recovered:

# time ipmitool mc info
Device ID                 : 32
Device Revision           : 0
Firmware Revision         : 0.00
IPMI Version              : 2.0
Manufacturer ID           : 46045
Manufacturer Name         : Unknown (0xB3DD)
Product ID                : 0 (0x0000)
Product Name              : Unknown (0x0)
Device Available          : yes
Provides Device SDRs      : no
Additional Device Support :
    Sensor Device
    SEL Device
    FRU Inventory Device
    Chassis Device
Aux Firmware Rev Info     :
    0x00
    0x00
    0x00
    0x00

real    0m0.099s
user    0m0.001s
sys     0m0.000s

[leiyu-bytedance]

@wak-google If you need additional information, please let me know.

[williamspatrick]

@leiyu-bytedance when we opened the Google-misc repo we said that no other meta layers than Google ones should be pulling it into recipes. If you are thinking this is better than kcsbridge, can you and the Google team work at merging those improvements back in? I’m kind of disappointed we have effectively forks of code we already maintain because that was not what was said when we opened Google-misc.

[wak-google]

My plan was to send this code out for review once I had the necessary bits in sdbusplus that shouldn't really exist as part of this rewrite. This is something we should merge back in to the kcsbridge repo.

[leiyu-bytedance]

@leiyu-bytedance when we opened the Google-misc repo we said that no other meta layers than Google ones should be pulling it into recipes. If you are thinking this is better than kcsbridge, can you and the Google team work at merging those improvements back in? I’m kind of disappointed we have effectively forks of code we already maintain because that was not what was said when we opened Google-misc.

The original issue was created at openbmc/kcsbridge#1, there is no feedback from the maintainer, and @wak-google responded that Google had this re-write.
So we tested the re-write, found an issue and created the issue here.

Anyway, I did propose to submit this to openbmc/kcsbridge (see openbmc/kcsbridge#1 (comment))