Github user MikeThomsen commented on the issue:
https://github.com/apache/nifi/pull/2175
So yeah, +1 on merge.
---
Github user MikeThomsen commented on the issue:
https://github.com/apache/nifi/pull/2175
Ok. I just went back and tried out PutElasticSearchHttp (don't have time to
retry the record one right now), and the behavior was as expected. It treats
any delete that doesn't result in a
Github user mattyb149 commented on the issue:
https://github.com/apache/nifi/pull/2175
This is my PR. If you give it a +1 we can have a committer merge it, thanks!
---
Github user MikeThomsen commented on the issue:
https://github.com/apache/nifi/pull/2175
@mattyb149 Do you want to keep reviewing this or close it out?
---
Github user MikeThomsen commented on the issue:
https://github.com/apache/nifi/pull/2175
> Are you seeing "not-found-updates" as failures, and "not-found-deletes"
as successful? If so then are you ok with that behavior? I'm tending towards
keeping it that way.
Absolutely. I
Github user mattyb149 commented on the issue:
https://github.com/apache/nifi/pull/2175
Failure in the case of not-found-while-deleting is subjective IMO. You can
look at it as "the processor did not perform the task I issued", or "the
processor accomplished the end goal of the task
Github user mattyb149 commented on the issue:
https://github.com/apache/nifi/pull/2175
So it is Elasticsearch that is treating "not found while deleting"
differently than "not found while updating", they return different status codes
with the "error" message in different fields in
Github user MikeThomsen commented on the issue:
https://github.com/apache/nifi/pull/2175
Going forward, I think < 5.X support should be marked as deprecated. By v7
they've claimed that "types" are going to be gone completely and it's going to
resemble Solr in being a flat kv pair
Github user MikeThomsen commented on the issue:
https://github.com/apache/nifi/pull/2175
I'm using 5.5.2
---
Github user mattyb149 commented on the issue:
https://github.com/apache/nifi/pull/2175
Hmm, that's interesting, deleting with ID not found was my whole point of
using "result" instead of "reason". What version of ES are you using to test
against? I wonder if the behavior has changed
Github user MikeThomsen commented on the issue:
https://github.com/apache/nifi/pull/2175
I forgot to mention that when you run into that scenario, it actually sends
the output to Success, which it shouldn't since most of the documents failed to
be deleted.
---
Github user MikeThomsen commented on the issue:
https://github.com/apache/nifi/pull/2175
I found one area that this doesn't handle properly at all, and that's
deleting documents where the ID cannot be found. The reason your patch fails
there is that the check around the declaration
Github user mattyb149 commented on the issue:
https://github.com/apache/nifi/pull/2175
If you need to raise the logging level, you can add that processor to
logback.xml with a setting of INFO, otherwise you can change all processors to
INFO with the following sed statement (works on
Github user mattyb149 commented on the issue:
https://github.com/apache/nifi/pull/2175
Before the fix, you would probably see a log statement for each failed doc.
Also for some "not found" errors, such as maybe deleting a doc that doesn't
exist? you would see an empty reason. The fix
Github user MikeThomsen commented on the issue:
https://github.com/apache/nifi/pull/2175
I actually had a bad batch go through, and it sent them out properly. What
log settings need to be set in the configuration files to make sure I see your
changes?
---
Github user mattyb149 commented on the issue:
https://github.com/apache/nifi/pull/2175
This PR was about error handling, could you try 10k bad
documents/operations?
---
16 matches
Mail list logo
