BIG-IP failed load license and comes to un-configured system after perform reboot ?


If you found your BIG-IP system failed to load license or it was on un-configured system accidentally.
I found that problem too, and that my opinion about that.

Main root causes :
1. SOL7036: Linux uptime counter wrap after 497 days back set to zero.
This is not a critical issue. This issue was discovered in 1998, and is due to the 32bit counter overflow. Currently, there is no Linux kernel patch available to resolve this issue.

2. SOL7071: The SCCP kernel driver timer wrap causes an I2C read failure.
This is the result of a known issue. A timer wrap within the SCCP kernel occurs after 497 days, and resets the timer to O (zero).
The timer wrap causes I2C (Inter-Integrated Circuit bus) driver read/write failures because the I2C is unable to correctly handle the kernel timer wrapping.
The SMBus then enters a lockup condition, which causes all subsequent I2C read/write communications to fail until the SCCP is rebooted.
As a result, the SCCP cannot gather data from the host system on chassis fan speeds and link status until the SCCP is rebooted.

Summary :
1. Why system perform reboot accidentally ?
We should analyze /var/log/install.log and /var/log/user.log
--------------------
boot_marker: ---===[ HD1.1 - BIG-IP 9.2.3 Build 34.3 ]===---
boot_marker: ---===[ HD1.1 - BIG-IP 9.2.3 Build 34.3 ]===---
-----------------

As we see, last system power up on . And this system accidentally restart at .

Difference between and more than 487 days uptime.
Both Main root causes have triggered.

2. Why after reboot system at in-operational state ?
BIG-IP 3400 use SCCP module.
SCCP module can't be restart with soft-reboot (ex: "reboot" command ), it need hard-power down or hard-reboot.
SCCP also can restarted by

If SCCP module can't start well, below the results:
# more /var/log/kern.log
-----------------
i2c-piix4.o: Unusual config register value
i2c-piix4.o: Try using fix_hstcfg=1 if you experience problems
-----------------

If i2c fail, it can't load hardware platform to kernel.
Hardware platform used for loading "license" and it used by all of BIG-IP daemons.
It also affect all network interfaces except Management Network Interface.

Here results if loading license on system was failed :
# more /var/log/ltm
--------------------
sccp sbmand[4300]: 012a0003:3: Failed to get/verify data on this slave chip
sccp sbmand[4300]: 012a0005:5: Reading system_type from CPLD
sccp sbmand[4300]: 012a0002:2: Interface map not initialized - platform may not be supported
tmm tmm[1006]: 01230017:2: Unable to attach to PCI device 00:0f.00 for Interface 0.10
tmm tmm[1006]: 01230017:2: Unable to attach to PCI device 00:13.00 for Interface 0.11
tmm tmm[1006]: 01230017:2: Unable to attach to PCI device 00:14.00 for Interface 0.12
mcpd[1083]: 012a0003:3: hal_get_dossier: dossier_storage_info error = 2
sccp bcm56xxd[4305]: 012c0001:3: BCM56xxd failed to init hardware bcm56xxd.cpp(959)
---------------------

Different results if licence was successfully loaded :
# more /var/log/ltm
---------------------------
sccp bcm56xxd: 012c0013:6: Connecting to HAL
sccp sbmand[225]: 012a0005:5: Reading platform_type from SEEPROM = C62a
--------------------------

I think there is no human-error here, it just Linux kernel and SCCP module limitation. We just need hard-reboot or hard-power down to solve this.

Popular posts from this blog

Howto configure boot device order on ILOM

Installing Depot files on HP-UX 11.23 ia64

SAN Switch Config Command