[ovirt-devel] Re: Recovering from power failure

Moving the discussion to devel list on this scenario. Il giorno mar 12 feb 2019 alle ore 11:16 Hetz Ben Hamo <hetz(a)hetz.biz&gt; ha scritto: > Hi, > > Well, there is a severe bug that I complained about it on 4.2 (or 4.1? I > don't remember) and it's regarding "yanking the power cable". > Basically I'm performing a simple test: kill all hosts immediately to > simulate a power loss without UPS. > > For this test I have 2 nodes, and 4 storage domains: hosted_storage > (that was setup during the HE installation), 1 iSCSI domain, 1 NAS domain > and 1 ISO domain. > > After all the nodes loose power, I power them on and the following > procedure happens: > 1. The node with HE finishes booting, and it takes few minutes until the > HE is up. > 2. When the HE is up, all the storage domains comes back to life as > online and VM's with high availability starting to boot. > 3. Few minutes later, *all* (with the exception of hosted_storage) > storage domains are going down > 4. After about 5 minutes, all the other storage domains which went down, > are coming up, but by then, and VM's without high availability that are not > hosted on hosted_storage remains down, you'll need to power them manually > back. > > This whole procedure takes about 15-25 minutes after booting the nodes, > and this issue is always repeatable, just kill the power to the nodes, > power them up again and see for yourself. > > The solution would be to change the code and if a storage domain is up - *leave > it up*, skip the check. > > Tal, Nir, what do you think about this?

> Thanks > > > On Tue, Feb 12, 2019 at 11:56 AM Sandro Bonazzola <sbonazzo(a)redhat.com&gt; > wrote: > >> Hi, >> We are planning to release the first candidate of 4.3.1 on February >> 20th[1] and the final release on February 26th. >> Please join us testing this release candidate right after it will be >> announced! >> We are going to coordinate the testing effort with a public Trello board >> at https://trello.com/b/5ZNJgPC3 >> You'll find instructions on how to use the board there. >> >> If you have an environment dedicated to testing, remember you can setup >> a few VMs and test the deployment with nested virtualization. >> To ease the setup of such environment you can use Lago ( >> https://github.com/lago-project) >> >> The oVirt team will monitor the Trello board, the #ovirt IRC channel on >> irc.oftc.net server and the users(a)ovirt.org mailing list to assist with >> the testing. >> >> [1] >> https://www.ovirt.org/develop/release-management/releases/4.3.z/release-m... >> >> -- >> >> SANDRO BONAZZOLA >> >> MANAGER, SOFTWARE ENGINEERING, EMEA R&D RHV >> >> Red Hat EMEA <https://www.redhat.com/> >> >> sbonazzo(a)redhat.com >> <https://red.ht/sig> >> _______________________________________________ >> Users mailing list -- users(a)ovirt.org >> To unsubscribe send an email to users-leave(a)ovirt.org >> Privacy Statement: https://www.ovirt.org/site/privacy-policy/ >> oVirt Code of Conduct: >> https://www.ovirt.org/community/about/community-guidelines/ >> List Archives: >> https://lists.ovirt.org/archives/list/users@ovirt.org/message/URIGV3LPTE2... >> > -- SANDRO BONAZZOLA MANAGER, SOFTWARE ENGINEERING, EMEA R&D RHV Red Hat EMEA <https://www.redhat.com/> sbonazzo(a)redhat.com <https://red.ht/sig>