Dan Kenigsberg has submitted this change and it was merged.
Change subject: sp: fix spm start when failing to produce domain ......................................................................
sp: fix spm start when failing to produce domain
During spmStart we attept to go over all the domains and set their role to be regular in case it's not to avoid situation in which there are two domains with the master role. If we fail to produce any of the domains (which might happen, the domains returned by getDomains(activeOnly=True) we shouldn't fail, as we attempt to produce all the domains that are marked as active in the metadata, but it doesn't mean that they are actually reachable and that calling produce() for the domain won't fail.
In case the we fail to call produce() on domain with StorageDomainDoesNotExist exception, it doesn't mean that spmStart should fail - otherwise we might never succeed to start spm.
I believe that we should ignore any exception to avoid the same case in different scenarios (for example - failure to alter the metadata), but as reviewers requested, let's start with StorageDomainDoesNotExist exception only.
Change-Id: Ia4f34360770ca8c8741a50956d0cab92bcd9a810 Bug-Url: https://bugzilla.redhat.com/show_bug.cgi?id=1072900 Signed-off-by: Liron Aravot laravot@redhat.com Reviewed-on: http://gerrit.ovirt.org/25424 Reviewed-by: Federico Simoncelli fsimonce@redhat.com --- M vdsm/storage/sp.py 1 file changed, 6 insertions(+), 1 deletion(-)
Approvals: Federico Simoncelli: Looks good to me, approved Liron Ar: Verified