https://bz.apache.org/SpamAssassin/show_bug.cgi?id=7335
--- Comment #23 from Bill Cole <[email protected]> --- (In reply to Reindl Harald from comment #21) > about which changes in SA in the last year are you talking about? The relevant bugs that come to mind are bug #7223, bug #7231, and bug #7265. > there was no release in the past two years Which does not mean that there have been no changes implemented in the current codebase. If this can be reproduced with the current code, rather than a vendor package of 3.4.1 with an unknown set of back-ported fixes and/or customizations. As has been noted, using SHORTCIRCUIT is problematic in connection with DNS-based rules, because unfinished queries are ignored. In the 2 collections of debug outputs provided, only the first of 20 runs fails to hit URIBL_BLACK and those 2 runs are cut short by SHORTCIRCUIT to less than 20ms from the start of the relevant DNS query to the harvesting of completed results. It is extremely unlikely to complete a recursive DNS query whose answer is not in a local cache in 20ms. In casual testing on a well-connected low-load host with the specific name being queried cleared from the cache between queries, I am unable to get an answer from any of the multi.uribl.com name servers in <35ms consistently and never <25ms. Given the available evidence, there is probably no bug here at all: only a known effect of SHORTCIRCUIT working as designed. If there IS anything more involved, it is not reasonable to diagnose a DNS issue using 3.4.1 rather than the current codebase due to the substantial code changes made in relation to DNS since the 3.4.1 release. -- You are receiving this mail because: You are the assignee for the bug.
