RE: BMC can’t get watchdog status by command “mc watchdog get” when system power is off

[Wayne Tung/WYHQ/Wiwynn]

Vishwa,

Thanks for replying!

Here is our thought and response.

- Could you help show which section demands this service to be mandatory available per IPMI v2.0 ?
-> This command is in section 27.7 Get Watchdog Timer Command in IPMI v2.0.

- What do you expect IPMI to tell when the host is off ?
We expect that IPMI can tell user the last status of watchdog because we want to let users know which stage of booting failed when host power on failed.

The information we want to know is as follow:

  1.  Watchdog Timer Use: Which stage of booting use the watchdog timer.
  2.  Watchdog Timer Actions: Watchdog timeout action.
  3.  Initial Countdown: Initial time of watchdog timer.

Here is an example of IPMI mc watchdog get response
Watchdog Timer Use:     Reserved (0x80)
Watchdog Timer Is:      Stopped
Watchdog Timer Actions: Hard Reset (0x01)
Pre-timeout interval:   0 seconds
Timer Expiration Flags: 0x00
Initial Countdown:      30 sec
Present Countdown:      30 sec



From: vishwa <vishwa at linux.vnet.ibm.com>
Sent: Tuesday, November 5, 2019 5:18 PM
To: Wayne Tung/WYHQ/Wiwynn <Wayne_Tung at wiwynn.com>; openbmc at lists.ozlabs.org
Subject: Re: BMC can’t get watchdog status by command “mc watchdog get” when system power is off


Wayne,

Thanks for writing. Here is my thought on this.

- phosphor-watchdog was meant to time the Host booting, and hence it is getting started where ever it is.
- There are other external users for watchdog ( like bmcweb ). So we need to see if this change affects those.
- Could you help show which section demands this service to be mandatory available per IPMI v2.0 ?
- What do you expect IPMI to tell when the host is off ?
- Lastly, please help push a gerrit review, so we could add comments there.

Thanks..

!! Vishwa !!
On 11/5/19 7:09 AM, Wayne Tung/WYHQ/Wiwynn wrote:
Hi,

We met an error that BMC can’t get watchdog status by command “mc watchdog get” when system power is off.

Cause:
"ipmitool mc watchdog get" gets dbus properties in path: /xyz/openbmc_project/watchdog/host0 interface: xyz.openbmc_project.State.Watchdog and this dbus object is registered by phosphor-watchdog.service.
Because phosphor-watchdog.service will not be started before system power-on and will be stopped after system power-off, therefore "ipmitool mc watchdog get" command gets nothing when system power is off.

Follow the standard, BMC should still report watchdog status even if system power is off.

So we modified phosphor-watchdog’s bb file and service file to
1. Start phosphor-watchdog after multi-user.target during BMC boot period
2. Do not stop phosphor-watchdog when system power off.
3. Always restart phosphor-watchdog.service.

Test Result:
wayne at TGA:~$ ipmitool -H 10.248.42.74 -U root -P 0penBmc -I lanplus chassis power status
Chassis Power is off
wayne at TGA:~$ ipmitool -H 10.248.42.74 -U root -P 0penBmc -I lanplus bmc watchdog get
Watchdog Timer Use:     BIOS FRB2 (0x81)
Watchdog Timer Is:      Stopped
Watchdog Timer Actions: No action (0x00)
Pre-timeout interval:   0 seconds
Timer Expiration Flags: 0x00
Initial Countdown:      0 sec
Present Countdown:      0 sec
wayne at TGA:~$ ipmitool -H 10.248.42.74 -U root -P 0penBmc -I lanplus chassis power on
Chassis Power Control: Up/On
wayne at TGA:~$ ipmitool -H 10.248.42.74 -U root -P 0penBmc -I lanplus bmc watchdog get
Watchdog Timer Use:     BIOS FRB2 (0xc1)
Watchdog Timer Is:      Started/Running
Watchdog Timer Actions: Hard Reset (0x01)
Pre-timeout interval:   0 seconds
Timer Expiration Flags: 0x00
Initial Countdown:      600 sec
Present Countdown:      597 sec
wayne at TGA:~$ ipmitool -H 10.248.42.74 -U root -P 0penBmc -I lanplus bmc watchdog get
Watchdog Timer Use:     BIOS FRB2 (0xc1)
Watchdog Timer Is:      Started/Running
Watchdog Timer Actions: Hard Reset (0x01)
Pre-timeout interval:   0 seconds
Timer Expiration Flags: 0x00
Initial Countdown:      600 sec
Present Countdown:      582 sec
wayne at TGA:~$ ipmitool -H 10.248.42.74 -U root -P 0penBmc -I lanplus bmc watchdog get
Watchdog Timer Use:     BIOS FRB2 (0x81)
Watchdog Timer Is:      Stopped
Watchdog Timer Actions: No action (0x00)
Pre-timeout interval:   0 seconds
Timer Expiration Flags: 0x00
Initial Countdown:      0 sec
Present Countdown:      0 sec

Thanks


---------------------------------------------------------------------------------------------------------------------------------------------------------------

This email contains confidential or legally privileged information and is for the sole use of its intended recipient.

Any unauthorized review, use, copying or distribution of this email or the content of this email is strictly prohibited.

If you are not the intended recipient, you may reply to the sender and should delete this e-mail immediately.

---------------------------------------------------------------------------------------------------------------------------------------------------------------
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ozlabs.org/pipermail/openbmc/attachments/20191120/2ae899dc/attachment-0001.htm>