We had a power outage few days ago, and I got the servers up and running but I was informed to look into, if there is a way to bring up the servers automatically/defaultly.
I was told the windows admin has their server set up where the servers are up automatically if there is a power outage.. manager told me "look in to the Bios". I didn't know what that meant, but only stuff I found was on windows servers....
I found this link but it talks about dell server.
There is usually a BIOS setting for "turn on after power loss", yes. It's a setting in the motherboard itself which you get into via a menu before the OS even boots. This makes it highly system-specific as well, some systems use F10 on boot to get to that menu, some use F2, some use DEL, etc, and where the option is found also varies. You'll have to research and explore.
Unfortunately, since the BIOS settings are very often corrupted and reset to defaults in power loss situations, the BIOS "power on after power loss" setting is very untrustworthy anyway.
There was one model of PC back in 1997 which had a switch on the motherboard to configure behavior after power loss. They're the only computers I've encountered which I could actually trust to do so.
I hope I have uploaded two files - one titled autostartPolicypng and the other is startpolicy.png
These are screencaptures from the web-based ASMI - Advanced System Management Interface. I do not have an HMC, so I use the web interface. HMC would have a different way to change these.
Approach ASMI as the system firmware IP address as https://IP_addressOrHostname and login.
Now you should see - on the left - an area titled "Power/Restart Control".
The panel entitled "Power On/Off System" has a selection entitled "System firmware start policy". This is (mainly) for a normal startup.
The choice you want to look at is in the left menu as: "Auto Power Restart". Make sure this is enabled.
HOWEVER, if you system is managed by an HMC all partitions get started, more or less in parallel - so the ones that are dependent on a VIOS may not start as expected. When you do not have an HMC - the VIOS - as IVM (integrated Virtualization Manager) has that role and will start client partitions AFTER it has started itself - and so the system returns as much as possible (depending mainly on the integrity/availability of external storage).
Short advice: get on the ASMI web interface and make you adjustments. Or use the HMC to do the same.
That helps yes, but is ASMI web interface a IBM tool or Third party free source? how can i get this? can i get this even if i have HMC? stupid question but how can I bring this up to my manager...?
might all be easy questions to answer but i just want to not look stupid when bringing it up to my manager.
Once you find a way to automatically restart the system, you still have to decide if you should.
Assuming you have a UPS.
Does the UPS trigger the system to shutdown, and if so after what percentage of battery life is used?
Does the system just continue to run until the battery is dead?
Does the UPS restore power to the system if the battery is less than xx% charged?
Do you know how many power failures you have per year, and how many of these exceed the battery life?
If you do decide to automatically restart the system, you should make sure that you wait until the power supply is stable.
I don't expect, that you can convince your boss not to do it, but I personally will never start an AIX server with a lot of LPARs automatically. And I will never suggest it to anyone.
How would you propose to boot the HMC anyway? You can define partitions to be auto-boot at power on, however I think that requires the chassis to be powered on from the HMC.
I agree with agent.kgb that it is not good practice anyway. If your servers are so critical, it is far better to have good procedures and a 24-hour operations team.
Doing things because it seems easy might not be the right way to do it. Surely your servers have some dependencies that might need to be confirmed anyway after a power loss and you may have filesystems and/or database to check over to ensure they have recovered.
You want to explain something technical to a manager and expect him to understand? Good luck with this! If he would be able to understand something technical he would be administrator, not manager, no?
Having said this: a POWER system is a lot more complex than the PC-crap Windoze-people use to clutter up datacenters with. Understand that the following is stretching the truth a bit to accomodate the PC-jargon. It is probably as close to the truth as your manager is going to understand it anyway, so perhaps no harm is done:
There is something remotely analogous to a BIOS, but it is controlled from the HMC mostly. The system is connected to the outside (this means primarily: the HMC) via one or two so-called "Service Processor"s, which - in functionality - are similar to the "ILO-boards" of rack-mounted PC-servers. These service processors have a web interface (in fact they have little specialized web servers built in as firmware), which is called ASMI. You can access the ASMI directly on a new system by connecting a network device to the service processor board, configuring your interface into the same address range as the default addresses of the interfaces (consult this page for details) or you can access it via the HMC once the system is connected to one.
[/quote]
This i can wholeheartedly concur to. If a system goes down there is usually a reason for this and there should be an administrator present to assess which this reason was and if it is safe to bring the system back online again. Whatever the reason for going down was it usually doesn't go away with the reboot so bouncing back online is a potentially hazardous proposition. Automatically starting a system after a forced down might initiate a case of "system ping-pong" which may be even more destructive than the initial outage.
I strongly advise against rebooting systems automatically and to my knowledge it is nowhere done in professional shops.
ASMI is built-in to the firmware. You need to know the address of the service processor.
Are you using an HMC? Yes, or No. If so, the easiest way to find the IP address is to use the HMC to ask for it. If you are not using an HMC, it probably still has a default address - and these may be dependent on the model (I have not worried about default addresses for a long time, so I shall trust someone else will provide a link for the data and/or a table with the defaults).
Since you are talking about auto-start I assume the system is already installed.
What version of AIX and/or Linux (or even IBM i) are you using? Is it bare metal, or are you using VIOS?
---------- Post updated at 08:36 PM ---------- Previous update was at 08:29 PM ----------
And in reply - rather addition to Bakukin.
When the system has gone down on it's own - definitely not wise to try to autostart everything. If it has been a powerfailure - it should be okay to let the VIOS restart (and the option mentioned here is only for restart after a Powerfail) - I do not worry myself much about having the VIOS restart. VIOS is meant to be "black-box" once it is configured. If the reboot fails, no, nada, zero, reasons to allow it continuing the start without an admin to watch it.
Autostart of partitions that depend on a VIOS, basically, fails all the time when the system is managed by an HMC - because the client partition may start in parallel - and be passed it's boot attempts before the VIOS is ready to service it.
In short, IF you are going to use autostart - limit it to VIOS partitions.
