fog-openstack
fog-openstack copied to clipboard
Auto Retry (Idempotency requests)
Auto Retry option for all services
Auto Retry is off by default but can be turned on with :idempotent option.
Default values are :retry_limit => 2 and :retry_interval => 5 (seconds)
Only GET and DELETE REST verbs are used for auto retry. More verbs can be added if that make sense.
Addresses https://github.com/fog/fog-openstack/issues/453
Build succeeded.
- fog-openstack-unittest-test : SUCCESS in 14m 56s
- fog-openstack-unittest-spec : SUCCESS in 14m 10s
Build succeeded.
- fog-openstack-unittest-test : SUCCESS in 12m 57s
- fog-openstack-unittest-spec : SUCCESS in 13m 42s
Build succeeded.
- fog-openstack-unittest-test : SUCCESS in 18m 48s
- fog-openstack-unittest-spec : SUCCESS in 13m 08s
Build succeeded.
- fog-openstack-unittest-test : SUCCESS in 12m 02s
- fog-openstack-unittest-spec : SUCCESS in 17m 38s
@gildub Hi Gilles, I know it is late, but I'm happy merging this PR when the https://github.com/fog/fog-openstack/pull/462/conflicts gets resolved.
@aufi, Hi Marek, sorry for slow response. I've addressed conflict. Thanks!
Build succeeded.
- fog-openstack-unittest-test : SUCCESS in 15m 42s
- fog-openstack-unittest-spec : SUCCESS in 16m 03s
@aufi any chance to merge this into the master? The Swift service used by our company is pretty unreliable and this should solve that issue.
@ShamoX this seems as a good and safe addition and ready to get in. Of course tests are missing, what is your preference here?
Yes, tests are missing. I don't really see how we can tests this except by implementing a complexe request/response scenario, which seems not easy to implement with VCRs (maybe with a sort of interaction with doubles - but seems really complicated).
Anyway, I can review it more precisely when I have more time and see if I can catch any statically analysable bad behaviours. But I doubt it.
We should at least test new configuration variables to see if they are used.