recursive zone calls hang when at least 1 child API call hangs

[Chris Behrens]

If a child zone API call hangs for some reason (I've had it happen due to another bug.. I think possibly lp:803168), the parent will wait forever. Parent needs to deal with this more gracefully.. either by timing out queries to children. If the scheduler is doing a select query to figure out available hosts for a build request, it should ignore zones that aren't responding. Other queries might need to return an error.

[Chris Behrens (cbehrens)]

Attempting to fit this into my team's Sprint 9 work (due 7/15/2011)

[Chris Behrens (cbehrens)]

I believe this was fixed by adding timeouts to nova-client. Going to change this to 'Invalid'. If there's a problem in the future, we'll open a new bug.

[Ed Leafe (ed-leafe)]

Just a heads-up: timeout exceptions were added to nova-client, but a subsequent merge removed them. You might want to double-check your "beliefs" before rejecting this.

[Chris Behrens (cbehrens)]

Ah, thanks. Fair point.

I knew you had fixed it in novaclient... at one time. I did not know it got removed, however. Interesting. Was there a specific reason for the removal from novaclient as far as you're aware... Ie, people felt it didn't belong there... or ?

Did you want to leave this bug assigned to yourself or want one of us to take it?

[Ed Leafe (ed-leafe)]

I don't believe that it was deliberately removed; at least no one said anything about doing so. Rather, it seems as though it was accidentally removed in a later pull request, and none of the approvers noticed.

I won't have time to work on this, so you should assign it to someone else.

[Chris Behrens (cbehrens)]

doesn't apply anymore now that zones code is removed.