We’re giving away 1,500 more DJI Tello drones. Enter to win ›
by David Tansley | Published May 12, 2014
In this demonstration, I will be using hdisk0 in the examples provided. The checks I provide are not an exhaustive list, but rather common issues I have come across over the years. When getting bosboot errors, I always find it best to go by the numbers, that is, a checklist I tick off until the error is resolved. Getting bosboot errors are not a show stopper, of course easy to say perhaps, but not so easy to be convinced by that statement when you are by yourself and it is late into the evening doing some IBM® AIX® maintenance. The errors typically occur due to some recent change on the disks, this could be migration or accidental user error. I have always been able to resolve bosboot issues and reboot a system, even if AIX shouts at me not to reboot.
The most common errors issued by the bosboot command that I have come across are:
The following tasks are in no particular order as a bosboot error can crop up in different situations. The first port of call is to check your current bootlist and find whether you are booting from the correct disk. You can check this using the following commands.
# lspv |grep rootvg
hdisk0 00cd94a60f01c745 rootvg active
# bootlist -m normal -o
What was the disk you last booted off? Is it what you think it was? Both of the following commands will return the information you require.
# bootinfo -b
# getconf BOOT_DEVICE
If no output is returned in the above example, this is typically due to a recent rootvg migration, and this in itself is not a real problem. So let’s carry on with the tasks to check.
As a point of interest, cross check the date of the last hardware migration with the last time it was rebooted using the following command.
# who -b
. system boot Dec 04 10:57
Confirm whether the bootable disks that AIX knows about are the disks listed and are contained in your current bootlist.
# ipl_varyon -i
PVNAME BOOT DEVICE PVID VOLUME GROUP ID
hdisk0 YES 00cd94a60f01c7450000000000000000 00cd94a600004c00
hdisk1 NO 00cd94a6e0bd72af0000000000000000 00cd94a600004c00
If you find that your disk is not a bootable device in the output from the ipl_varyon command, ask yourself whether you issued a : chpv –c hdisk0 command by mistake? If you did clear the boot records of the hdisk, it will not be displayed as bootable. If this is the case, rerun a bosboot command on that disk with the following command:
: chpv –c hdisk0
# bosboot –ad /dev/hdisk0
Check that your boot logical volume which is typically hd5 resides on the bootable disk.
# lslv -m hd5
LP PP1 PV1 PP2 PV2 PP3 PV3
0001 0181 hdisk0
If it is not, then migrate it across to the bootable disk. If hd5 is damaged and you cannot migrate it, then simply remove and re-create it afresh.
# rmlv hd5
# mklv -y hd5 -t boot -a e rootvg 1 hdisk0
Very rarely when running a bosboot command, AIX might complain about hd5 not being contiguous across the partitions it has been allocated. I have experienced this only one time. If this is the case, you have no choice but to remove hd5 and re-create it, as noted above. Then run the bosboot to the bootable disk and ipldevice, as described further on.
If AIX states that it cannot run a bosboot command because hd5 existed on a previous hdisk or it cannot find ipldevice, then the following should fix it: Confirm whether the bootable disk has the same major and minor number as the ipldevice and the ipldevice is a hard link to the bootable disk.
# ls -l *hdisk0*
brw------- 1 root system 16, 2 Jan 15 11:23 hdisk0
crw------- 2 root system 16, 2 Oct 31 14:58 rhdisk0
# ls -l ipldevice
crw------- 2 root system 16, 2 Oct 31 14:58 ipldevice
In the above output, the major and minor numbers 16, 2 on hdisk0 match with that of the ipldevice, and therefore, no problems here. Issue bosboot on both: the ipldevice and the bootable disk.
# bosboot –a /dev/ipldevice
# bosboot –a /dev/hdisk0
The host can then be rebooted.
If the major and minor numbers are different on the bootable disk to ipldevice, then fix it by removing the ipldevice and relinking.
# rm /dev/ipldevice
# ln /dev/rhdisk0 /dev/ipldevice
Now that ipldevice has been relinked to the bootable disk, run bosboot, on both the ipldevice and the boot disk.
# bosboot -ad /dev/ipldevice
# bosboot -ad /dev/hdisk0
If your ipldeivce is not present at all, re-create the link as describe above.
In my experience, all conditions have been met. If you can run a bosboot command on the ipldevice and the hdisk and the major and minor numbers match with the ipldevice, the host can be rebooted. When the host comes up back, you will find that the bosboot –a command works fine.
The savebase command stores information in the Object Data Manager (ODM) as such and is closely linked with the bosboot command. Typically, savebase errors occur after a hardware migration or a alt_disk migration. A common error is when lspv reports that the bootable disk is not part of rootvg, but you can see it is when issuing a lsvg –p rootvg command. The savebase errors can be confusing sometime because the errors just pop up and can mask another issue. However, for this demonstration, let’s assume it is a straight-forward savebase problem. First, check whether the error is not due to space issues and run a verbose savebase command to confirm this.
lsvg –p rootvg
# savebase –v
Next , let’s assume it is a mirrored volume group and confirm that the volume group is indeed synchronized correctly using the following command.
syncvg –v rootvg
Then force a rebuild of the logical control blocks so that they are in sync with the volume group descriptors area on the disks.
synclvodm –Pv rootvg
That should be sufficient and the savebase command should now work. Finally, run the savebase –v command, and all should be good. Then, to complete the process, run bosboot on both: the ipldevice and the bootable disk, as noted earlier.
I have found that running the checks described in this article are sufficient to resolve the common bosboot issues. Getting savebase and bosboot errors is an inconvenience for sure. But, you can overcome the issue by following the checks discussed in this article. This can also give you the confidence that a reboot is good to go.
March 14, 2019
This article provides a way to implement a kernel module on Linux, compile it, and explore ways in which a…
May 8, 2019
Artificial intelligenceData science+
Back to top