Yeah, i considered that as well. It seems that if there's an SSVM starting it should try to wait for that, or perhaps there should be a configuration option to choose between LocalEndpoint or SSVM.
On Thu, Oct 17, 2013 at 9:43 AM, Darren Shepherd <darren.s.sheph...@gmail.com> wrote: > I haven't looked at this at all, but could it be related to an issue I > brought up before that commands go to the "LocalEndpoint" if the SSVM is > unavailable. I've seen this too where before the ssvm starts the management > stack tries to do the download instead of waiting for the ssvm to become > available and sending the request there. > > Darren > >> On Oct 17, 2013, at 8:07 AM, Marcus Sorensen <shadow...@gmail.com> wrote: >> >> In writing some full buildout tests I'm running into some intermittent >> issues. If I register a template after zone is created, but before >> secondary storage vm starts, the template is pretty much guaranteed to >> be deleted. If I watch for the SSVM to come up, and its agent to >> start, then there's maybe a 50% chance that the template will be >> deleted, even if it successfully downloads. The odds may go down if I >> wait awhile after the SSVM comes up. >> >> I noticed this ticket, but it talks about template from snapshot. I'm >> not sure if it's the same, but the root cause may be. I notice there's >> no fix release set. If it only ever happens when the zone is first >> being created, then it's not such a big deal, but if it happens >> whenever the SSVM is stopped or just started (or randomly as the issue >> implies), then it should be a blocker. If any template I register will >> get successfully downloaded and then deleted if my SSVM ever happens >> to be down, that's not good. >> >> https://issues.apache.org/jira/browse/CLOUDSTACK-3305