So it's a bit hard to say if it's file system differences or something else. Changing settings like the preallocation scope might be a good idea to see if that works (like you said) or other tweaks. I don't have a Mac (only linux) so i can't test it locally in that environment.
On Thu, Jan 29, 2026 at 5:23 AM Jean-Louis Monteiro < [email protected]> wrote: > Correct. It is consistently failing on MacOS, but passing on Linux. > -- > Jean-Louis Monteiro > http://twitter.com/jlouismonteiro > http://www.tomitribe.com > > > On Thu, Jan 29, 2026 at 1:27 AM Christopher Shannon < > [email protected]> wrote: > > > So just to be clear, does the test ever fail in linux or only in MacOS > and > > Windows? > > > > On Wed, Jan 28, 2026 at 6:42 AM Jean-Louis Monteiro < > > [email protected]> wrote: > > > > > Hi all, > > > > > > For nightly builds we run a bigger matrix (including macos and > windows). > > > I've seen the following failure consistently. > > > > > > [INFO] ---org.apache.activemq.store.kahadb.JournalFdRecoveryTest - > 42.99 > > s > > > [INFO] +-- [XX] testRecoveryWithMissingMssagesWithValidAcks - 4.399 s > > > [INFO] [*] testRecoveryWithMissingMssagesWithValidAcks > > > [INFO] > > > [INFO] Stack trace > > > [INFO] java.lang.AssertionError: Empty? expected:<18> but was:<19> > > > at org.junit.Assert.fail(Assert.java:89) > > > at org.junit.Assert.failNotEquals(Assert.java:835) > > > at org.junit.Assert.assertEquals(Assert.java:647) > > > at > > > > > > > > > org.apache.activemq.store.kahadb.JournalFdRecoveryTest.testRecoveryWithMissingMssagesWithValidAcks(JournalFdRecoveryTest.java:244) > > > > > > > > > It seems that after stopping the broker and deleting db-4.log and > > db.data, > > > we expect 2 messages to be lost. > > > > > > There might be something in MacOS filesystem that makes only 18 to > > > disappear and 19 to remain. > > > Not sure how to fix it. We can probably set the preallocationScope to > > NONE > > > and that might make it to pass. Of we need to accept a range rather > than > > a > > > fix value (maybe not the goal for this test). > > > We can also look at the message IDs to determine the value instead of > > using > > > the hardcoded 18. > > > > > > Any thoughts? > > > > > > > > > > > > -- > > > Jean-Louis Monteiro > > > http://twitter.com/jlouismonteiro > > > http://www.tomitribe.com > > > > > >
