Categories

  • All news regarding Xen and XCP-ng ecosystem

    146 Topics
    5k Posts
    C
    @benapetr @bvitnik Yeah, saving (even vertical) screen real estate is always a good thing - at least having the option (as it means more is visually and functionally available at a glance - at the same time).
  • Everything related to the virtualization platform

    1k Topics
    16k Posts
    poddingueP
    I haven't tried Duo on a host myself, so this comes from reading an 8.3 dom0 rather than from experience. The first thing I'd watch is that XAPI checks the root password through PAM as well. /etc/pam.d/xapi just includes system-auth, so my guess is that a Duo module placed in system-auth would make every password login to XAPI (XO's connection to the pool, xe -pw, XO Lite) ask for a factor it can't give. The physical console is less simple than it looks too: tty1 runs xsconsole, and on my 8.3 host it checks root against the password-auth PAM service by handing over just the password, so I'd expect a module that asks a second question to make that login fail rather than prompt. Also worth knowing before an audit: software from outside the XCP-ng repos isn't supported in dom0 (https://docs.xcp-ng.org/installation/requirements), and as far as I can tell from rpm, /etc/pam.d/xapi isn't marked as a config file, so an update would put it back. Anyone who has been through this for real will know more than me.
  • 3k Topics
    30k Posts
    MarieGardeM
    Thanks @poddingue for the detailed answer, that’s right on both points. The split advice comes from XO 6 being designed around one schedule per backup job, but it doesn’t fit your case: as said above, two jobs would mean two separate chains. You can keep your current job (one chain) for now, and run it manually from XO 5. The upcoming XO 6 backup job form will let you choose on which days/hours the full backup happens, on a single schedule. You’ll then be able to merge your two schedules into one: deltas every day, fulls on the chosen days, all in the same chain.
  • Our hyperconverged storage solution

    57 Topics
    839 Posts
    C
    HI, I have a test cluster which has exhibited a repeated failure pattern. After a week or two of operation, all actions on the XOSTOR SR fail with messages like this: SR_BACKEND_FAILURE_47(, The SR is not available [opterr=Error: Unable to connect to linstor://82.147.85.101: [Errno 111] Connection refused], ) Note that this IP address is not part of my cluster. Once the error message appears, it will complain about the same IP address until I fix the issue (usually by some fairly drastic cluster reboots and/or recreating the SR). This has now happened three times, with a different apparently-random IP each time. Obviously this looks very suspicious. Thankfully there's no serious data on this cluster at the moment, and I have extensive offsite backups if needed. I can keep it running like this for a little while if anyone is able to do some debugging. Hopefully this is just a very odd bug, of course, but I'd like to be sure. Especially because it is now recurring. Thanks,
  • 37 Topics
    137 Posts
    S
    @jqueiroz said: Bom dia a todos, Estou com um problema em uma VM, ela está ligando/desligando sozinha, repetidamente. Não encontro nenhuma mensagem de erro explicando o motivo. Alguém pode me dar alguma dica sobre como solucionar isso? Uso o XCP-NG 8.0 em um cluster de dois servidores HP Proliant DL-360, com o HA-Lizard. Tudo vinha funcionando bem até alguns dias atrás, quando houve uma queda de energia no datacenter da localidade. Tenho o mesmo arranjo de VMs/servidores em mais seis localidades, nenhuma apresenta esse problema. Analisar esse comportamento instável do cluster é quase como prever resultados complexos, parecendo a análise de riscos que a malta faz ao dar palpites no https://mostbetbrasil.lat. Grato. Como o problema começou depois da queda de energia e só acontece nessa localidade, eu começaria verificando os logs do XCP-ng e do HA-Lizard para entender quem está disparando os reinícios. Pode ser alguma política de HA tentando recuperar a VM repetidamente, algum problema de armazenamento ou até inconsistência no estado do host após a interrupção. Também vale conferir o estado do pool e os logs do Xen Orchestra, se estiver usando. Se possível, desative temporariamente a automação de recuperação dessa VM, com cuidado para não comprometer a disponibilidade, e veja se ela continua reiniciando sozinha. Isso ajuda a isolar a causa.