[GitHub] [lucene-solr] mikemccand commented on pull request #2088: LUCENE-9617: Reset lowestUnassignedFieldNumber in FieldNumbers.clear()

2020-12-15 Thread GitBox
mikemccand commented on pull request #2088: URL: https://github.com/apache/lucene-solr/pull/2088#issuecomment-745373852 I beasted this PR for 36 iterations and uncovered only an [unrelated issue](https://issues.apache.org/jira/browse/LUCENE-9638) -- I think this is ready! I'll push soon.

[GitHub] [lucene-solr] mikemccand commented on pull request #2088: LUCENE-9617: Reset lowestUnassignedFieldNumber in FieldNumbers.clear()

2020-12-14 Thread GitBox
mikemccand commented on pull request #2088: URL: https://github.com/apache/lucene-solr/pull/2088#issuecomment-744542544 > > I think this method serves an important purpose/use-case: it lets you efficiently clear the whole index, transactionally, so that e.g. searchers can see the old index

[GitHub] [lucene-solr] mikemccand commented on pull request #2088: LUCENE-9617: Reset lowestUnassignedFieldNumber in FieldNumbers.clear()

2020-12-01 Thread GitBox
mikemccand commented on pull request #2088: URL: https://github.com/apache/lucene-solr/pull/2088#issuecomment-736781955 > I am not sure we can guarantee anything on an OOM? Yeah I suppose that is essentially nearly hopeless, sigh -- this is why `IndexWriter` aborts on OOME. >

[GitHub] [lucene-solr] mikemccand commented on pull request #2088: LUCENE-9617: Reset lowestUnassignedFieldNumber in FieldNumbers.clear()

2020-11-29 Thread GitBox
mikemccand commented on pull request #2088: URL: https://github.com/apache/lucene-solr/pull/2088#issuecomment-735405583 Great, thank you for adding the new "user space impact" test case @msfroh (in addition to the low-level unit test)! @dsmiley does that address your concern about the low

[GitHub] [lucene-solr] mikemccand commented on pull request #2088: LUCENE-9617: Reset lowestUnassignedFieldNumber in FieldNumbers.clear()

2020-11-27 Thread GitBox
mikemccand commented on pull request #2088: URL: https://github.com/apache/lucene-solr/pull/2088#issuecomment-734887851 > Curiously, the test wasn't failing on the OOME itself, but rather we fail an assertion when trying to rollback on tragedy (where the tragedy was the OOME). Here's that

[GitHub] [lucene-solr] mikemccand commented on pull request #2088: LUCENE-9617: Reset lowestUnassignedFieldNumber in FieldNumbers.clear()

2020-11-20 Thread GitBox
mikemccand commented on pull request #2088: URL: https://github.com/apache/lucene-solr/pull/2088#issuecomment-731398054 OK thanks @msfroh. This indeed sounds like a very inefficient way to test for this particular bug! But maybe you could mark it `@Monster`, like e.g. `Test2BBinaryDocVal

[GitHub] [lucene-solr] mikemccand commented on pull request #2088: LUCENE-9617: Reset lowestUnassignedFieldNumber in FieldNumbers.clear()

2020-11-20 Thread GitBox
mikemccand commented on pull request #2088: URL: https://github.com/apache/lucene-solr/pull/2088#issuecomment-731226197 Maybe we could add a simple unit test here that keeps re-indexing the same small set of docs with many unique fields, then calls `IW.deleteAll`, many times? Such that th

[GitHub] [lucene-solr] mikemccand commented on pull request #2088: LUCENE-9617: Reset lowestUnassignedFieldNumber in FieldNumbers.clear()

2020-11-20 Thread GitBox
mikemccand commented on pull request #2088: URL: https://github.com/apache/lucene-solr/pull/2088#issuecomment-731222134 > > As long as FieldNumbers.clear() is only called from IndexWriter.deleteAll(), my understanding is that the safety is provided by the try (Closeable finalizer = docWrit