hyc@symas.com wrote:

Howard Chu wrote:

...

michael@stroeder.com wrote:

...

FWIW: With back-mdb I ran 64 iterations without this error. Testing with back-bdb and back-hdb the failure occurs on my system after max. 10 iterations.

This sounds like it may be related to the issue I've been chasing down with ITS#7052. It appears that this bug was caused by the patch for ITS#6024. I'm about to revert that patch in master. Please re-test with d4b3c2af ...

Hm, no, that patch seemed to have no bearing here. But reverting ba4366eae appears to clear things up. Now I just need to re-read test054 to see if this is really what we want.

Just to sum up - this bug was a regression caused by ba4366eae. That patch was written in an attempt to resolve other bugs related to ITS#7052. But the actual cause of the remaining ITS#7052 issues was the ITS#6024 patch. Both of these patches have been reverted in git master.