exit type vs bios versions



  • hello guys!

    We just found a disturbing issue. Lab PCs are set to have boot order starting with pxe boot normally, then chained back to hdd. Here came the problem. We just discovered that some of the PCs are stuck at this point:
    0_1516197128251_2018. 01. 17.  PXE.jpg

    After a bit of investigations we found out that the fail and the success turns on the bios versions of the motherboards.

    fail and stuck:
    MB model: H61M-DS2
    Bios vers: F9
    Bios date: 2013.04.17.
    BIOS id: 8a01ag07

    success:
    MB model: H61M-DS2
    Bios vers: F9
    Bios date: 2011.12.21.
    BIOS id: 8a01ag07

    I think it is something with bios exit type (general default is sanboot). Host is not regged in fog database (veryvery few pcs are perma regged) as the have tasks only at cloning. What can be the solution here? To set general exit type other from sanboot may result in inverted, but still existing problem, maybe a lot more issued computer.

    What if i register pc but give zero task. Will the "host specific exittype work? or is it only applied at end of task?

    Downgrading the bios in mass is ofc not an option too :D and i guess pc manufacturers wont give us a few new bios just for this.

    Server type is 1.4.4., on debian 9.3…



  • Oh, good point. We do tests on it. As testing i did reset, not cold start. Will check and report.


  • Developer

    @Foglalt Maybe there is a difference between restart (warm start) and complete shutdown/boot up (cold start)? You might even want to try switching off the machine, unplugging the power cable, pushing the power button twice, then replugging the power and turn it on. Sometimes there is a different behavior between those different ways of restart/turn off/start.



  • @george1421 :) ok, sorry. i try to be more clear.

    as it was suggested, i tried to use other exit type to see what is a working solution. it came out that grub is ok for that host (i did many test to see). we sat back and enjoy. a later time my colleague reported that problem is there again (actually i was no investigating personally as he was the same guy who reported it first, so i trusted his eyes to see it or not :D

    i sent him back to make test restarts and it worked ok during tests. but if it worked on tests what may cause the random stuck, like on first ocasion? here we have a saying like “at dentists office no tooth hurts”. so when i am standing there and see restart test it is ok, but when my back is turned it stucks. ofc not all time, as i would kick that pc away to trash. but sometimes.

    we will wait a few days to see if it is there or not again (btw it can still be a failing buggy bios)


  • Moderator

    @foglalt Sorry maybe it friday, or I need more coffee this AM, but I don’t understand what you are saying here. Pictures and screen shots of settings and the console of the computer in error may help us understand what is going on.

    What boot up issues? Its stuck at boot from sanboot. If you change the host specific exit mode to grub, it should then use grub as the exit type not sanboot.



  • we have something strange. the pc with the bootup issue produced same error, stuck at “boot from sandivice”, even when it is set to grub exit type. when we saw it, we did couple of restarts to reveal something, but all time it was booting ok. so, if it is not being inspected, it produced error.

    anyone can guess what happens? if i did change exit type to and fro it does its things as expected. on “failtype” it fails, on “succestype” it succeeds. strange.


  • Moderator

    @foglalt Then, without the fog client working for you, you can scale your network well into the 10,000s without any issue.



  • @george1421 it is not a problem at all, as we… well, never ever used fog client on pcs. we do use fog to replicate and deploy installed pcs. :)


  • Moderator

    @foglalt said in exit type vs bios versions:

    o you happen to have performance statistic about how much speed reduction may come if all host (like 2k of host) are registered?

    With 2K hosts, you might want to change the client check in time to something longer than the default of 5 minutes. Maybe something in the area of 10 to 15 minutes. Consider in an ideal condition with 2k hosts and a 5 min check in window you would have ~7 client check-ins per second. But nothing is ever ideal, with 2K hosts you may have a bunch of check-ins then nothing the next second then a bunch more in the following second. All of these check-ins do consume CPU resources on your FOG server. I say all of this to that you can be mindful and watch your FOG CPU usage when you have all of the FOG clients loaded. If your CPU usage is too high, you can then change the check-in time to make it easier in your FOG CPU.


  • Senior Developer

    @foglalt there should be minimal impact of performance. The only slowdown I can imagine would come from the GUI when you do list all. No worries here in the very near future however. We have finally figured out how to fix this too.



  • Test was ok, the machine in question successfully continued boot with bios exit type set to grub. Thanks for the option we have in fog, as with older versions i guess it would be a full stop with those bios (btw manufacturer does not provide working solution for that)



  • ah, sounds nice! ok, will go to see what exit type works for that bugged crap :) thanks for help (btw do you happen to have performance statistic about how much speed reduction may come if all host (like 2k of host) are registered?)


  • Developer

    @foglalt said in exit type vs bios versions:

    Does it mean if i make a host registration (mac, os, anything) and i specify the exittype to anything other than default, it will boot and exit with that setting, not the fog system default, even if i give no task for that host? This is the true question, i can play with exit types, but i cant have set all host to the same exittype for testing

    Exactly that!



  • Does it mean if i make a host registration (mac, os, anything) and i specify the exittype to anything other than default, it will boot and exit with that setting, not the fog system default, even if i give no task for that host? This is the true question, i can play with exit types, but i cant have set all host to the same exittype for testing :)


  • Moderator

    Well it sounds like a firmware issue. Is there a newer version of the firmware for this system than 2013?

    Its possible that SANBOOT will not work for this specific model anymore because of the firmware update. Can you try another exit mode like GRUB or GRUB_FIRST_HDD on the newer firmware model? You can do this from the host management page. If GRUB works, then we need to try GRUB on a system that worked with SANBOOT. If both versions work then we can take the next step


 

561
Online

5.4k
Users

12.6k
Topics

118.9k
Posts