Bryan, Thanks for the update. I will reach out to Hortonworks to see what could be happening. I'll update this thread with what I find.
-Chad On Tue, Feb 19, 2019 at 3:49 PM Bryan Bende <[email protected]> wrote: > Chad, > > I suspect it may be an issue with the version of NiFi registry in the > vendor distribution. > > As far as I can tell, it is working correctly on apache releases of > NiFi 1.8.0/1.9.0-RC2 and registry 0.3.0. > > -Bryan > > On Tue, Feb 19, 2019 at 3:27 PM Chad Woodhead <[email protected]> > wrote: > > > > Bryan, > > > > I also just tested with a new flow not in version control with NiFi > Registry and then started version control with it, and same behavior on my > side. No load balance connection properties in the snapshot. > > > > Thanks, > > Chad > > > > On Tue, Feb 19, 2019 at 3:21 PM Chad Woodhead <[email protected]> > wrote: > >> > >> Bryan, > >> > >> No the snapshot in my git repo (we use GitFlowPersistenceProvider) does > not have the load balance connection properties. If I create a template of > the same PG on NiFi, I do see the load balance connection properties in the > xml. So for some reason the load balance connection properties are not > being sent to NiFi Registry. > >> > >> I am running HDF 3.3.0.0-165 > >> > >> Thanks, > >> Chad > >> > >> > >> On Tue, Feb 19, 2019 at 1:08 PM Bryan Bende <[email protected]> wrote: > >>> > >>> Chad, > >>> > >>> Using the 1.9.0-RC2 build of NiFi and the 0.3.0 release of registry, I > >>> haven't been able to reproduce the issue. > >>> > >>> I don't know of anything that would have fixed the issue between 1.8.0 > >>> and 1.9.0, so I'm not sure what you are running into. > >>> > >>> Can you look in your registry flow_storage directory (or git repo) and > >>> find the flow snapshot file for the latest snapshot that you believe > >>> has the load balanced connection, and then look for something like: > >>> > >>> "connections" : [ { > >>> "backPressureDataSizeThreshold" : "1 GB", > >>> "backPressureObjectThreshold" : 10000, > >>> "bends" : [ ], > >>> "componentType" : "CONNECTION", > >>> "destination" : { > >>> "comments" : "", > >>> "groupId" : "3320ad51-dbb1-388e-8457-38e1fe226e2e", > >>> "id" : "8a6087bd-d422-331e-9bc7-d7d2f533cfc0", > >>> "name" : "LogAttribute", > >>> "type" : "PROCESSOR" > >>> }, > >>> "flowFileExpiration" : "0 sec", > >>> "groupIdentifier" : "3320ad51-dbb1-388e-8457-38e1fe226e2e", > >>> "identifier" : "ea1b818e-1d2e-3c42-a956-1796392e85be", > >>> "labelIndex" : 1, > >>> "loadBalanceCompression" : "DO_NOT_COMPRESS", > >>> "loadBalanceStrategy" : "ROUND_ROBIN", > >>> > >>> If using the file based storage then the paths in flow_storage are > >>> <bucket_id>/<flow_id>/<version>/<version>.snapshot. > >>> > >>> If you don't see those last two load balanced related fields then that > >>> would cause the issue, but not sure why they wouldn't be populated > >>> from the NiFi side. > >>> > >>> Also, can you clarify if you are using apache releases, or a vendor > >>> release such as HDF? > >>> > >>> Thanks, > >>> > >>> Bryan > >>> > >>> On Mon, Feb 18, 2019 at 7:49 AM Chad Woodhead <[email protected]> > wrote: > >>> > > >>> > I am running NiFi 1.8.0 and NiFi Registry 0.3.0. I have noticed load > balance strategies on queues aren't coming through versioned flows in NiFi > Registry. Here are the steps I am performing: > >>> > > >>> > 1. Have existing flow running on latest flow version in Dev and Cert > (flow has already been developed and in version control) > >>> > 2. Add load balance strategy to queue in Dev flow > >>> > 3. NiFi shows local changes and I commit the changes to NiFi Registry > >>> > 4. Cert NIFi shows new flow version > >>> > 5. Pull latest version down to Cert. The load balance strategy for > the queue doesn't come with it. I then have to edit flow on Cert to add the > load balance strategy for the queue which causes NiFi to see local changes > which I then have to commit to Registry again but this time from Cert. > >>> > > >>> > I saw this JIRA https://issues.apache.org/jira/browse/NIFIREG-194 > which made me think I shouldn't be experiencing the behavior I am seeing. > >>> > > >>> > Thanks, > >>> > Chad >
