We use a custom start script for the pod that does a git pull and checkout of the environment-specific branch - can't recall why but we didn't use the Registry built-in pull mechanism (might have been branch related or something else I've forgotten).
Cheers, Chris Sampson On Mon, 14 Jun 2021, 14:43 Chris McKeever, <cgmckee...@gmail.com> wrote: > Hey Chris, this may be a silly RTFM manual question, but I couldnt find a > way to set the branch .. now that I write this, I wonder if you use the > *?ref=BRANCH* trick > > On Mon, Jun 14, 2021 at 8:30 AM Chris Sampson <chris.samp...@naimuri.com> > wrote: > >> We use a similar approach. >> >> We only allow our dev environment to write to the git repo (this is >> controlled by using different tokens to authenticate with our git server - >> one is read/write, others read-only). >> >> We have each environment using its own branch of the git repo (e.g. dev, >> staging, prod) and have a manual step to merge any changes between branches >> as part of our deployment process. >> >> Our Registry instances use the default h2 database and essentially treat >> it as ephemeral - every time an instance is restarted, it needs to rebuild >> its metadata database from the git repo. This is straight forward for us as >> we're deploying into a kubernetes environment, so the h2 files are on a >> filesystem that doesn't persist through a Pod restart. >> >> >> Cheers, >> >> Chris Sampson >> >> On Mon, 14 Jun 2021, 14:16 Sim, Yoosuk, <yoosuk....@bell.ca> wrote: >> >>> Hello everyone, >>> >>> >>> >>> Well, I have a little bit more updates to the issue. >>> >>> I did find that I was mistaken about pulling: nifi registry’s local git >>> does pull from remote git repo. When I checked the nifi-registry’s local >>> git storage, multiple ones were fully synced with remote git. However, even >>> though they were synced, nifi-registry itself did not recognize buckets >>> that were not created by the nifi-registry (I think), which meant only a >>> subset of the folders in the repo was shown as buckets by the nifi import >>> process. >>> >>> >>> >>> The only way to resolve it yet was to do something similar to >>> Hesselmann’s method: we dropped the metadata backend database in postgresql >>> and redeploy nifi-registry to repopulate the metadata. This worked, but it >>> certainly does introduce complications in the development process. >>> >>> >>> >>> At the moment, I am more inclined to give one git dedicated to one nifi >>> registry, unless I can find a simpler way to handle nifi-registry sync >>> issue. >>> >>> >>> >>> Any suggestions? >>> >>> >>> >>> Cheers, >>> >>> >>> >>> Tony Sim >>> >>> >>> >>> *From:* Hesselmann, Brian <brian.hesselm...@cgi.com> >>> *Sent:* June-07-21 3:58 PM >>> *To:* users@nifi.apache.org >>> *Subject:* [EXT]Re: Nifi Registry GitFlowPersistenceProvider >>> >>> >>> >>> The easiest way I found to sync nifi registry from Git is by deleting >>> the registry database file(should be in /nifi-registry/database/*.db) and >>> restarting nifi-registry. After that it should fully reflect all changes in >>> the git repository. Basically our process is something like: push changes >>> to nifi registry on dev branch, merge from dev to master branch, delete >>> nifi registry database on master and restart nifi registry. >>> >>> I'm sure there must be a nicer way to do this, but so far this has >>> worked for us until we can spent more time on implementing the registry. >>> ------------------------------ >>> >>> *Van:* Chris McKeever <cgmckee...@gmail.com> >>> *Verzonden:* maandag 7 juni 2021 16:49:21 >>> *Aan:* users@nifi.apache.org >>> *Onderwerp:* Re: Nifi Registry GitFlowPersistenceProvider >>> >>> >>> >>> EXTERNAL SENDER: Do not click any links or open any attachments unless >>> you trust the sender and know the content is safe. >>> EXPÉDITEUR EXTERNE: Ne cliquez sur aucun lien et n’ouvrez aucune pièce >>> jointe à moins qu’ils ne proviennent d’un expéditeur fiable, ou que vous >>> ayez l'assurance que le contenu provient d'une source sûre. >>> >>> >>> >>> Tony - did you ever get an answer on this? >>> >>> >>> >>> On Fri, Jun 4, 2021 at 9:04 AM Chris McKeever <cgmckee...@gmail.com> >>> wrote: >>> >>> oooh, this is interesting ... I know only one registry could/should be >>> the authoritative WRITER .. >>> >>> maybe there is a fetch hook that you can schedule to refresh ... >>> following >>> >>> >>> >>> On Fri, Jun 4, 2021 at 8:50 AM Sim, Yoosuk <yoosuk....@bell.ca> wrote: >>> >>> Hello everyone, >>> >>> >>> >>> I am currently setting up Nifi Registry. In our setup, we wanted >>> multiple Nifi Registry to talk to the same remote git repository. (say, one >>> in DEV, and another in QA, etc.) >>> >>> Over time, I found that not all Nifi Registry retained the same >>> information, even though the remote git repository had the latest >>> information. >>> >>> >>> >>> Does Nifi Registry ever pull from the remote git aside from when it >>> clones? >>> >>> What might be the best way to resolve this issue? >>> >>> >>> >>> Cheers, >>> >>> >>> >>> [image: logo_bell_120dpi_0_84_154] >>> >>> Yoosuk (Tony) Sim >>> >>> Dev, Machine Learning Engineering >>> >>> ------------------------------ >>> >>> *External Email:** Please use caution when opening links and >>> attachments / **Courriel externe:** Soyez prudent avec les liens et >>> documents joints * >>> >>
