I am only submitting this bug report after getting confirmation that someone else has seen the bug. One of my own 7.0 installs with updates had its named die a couple of weeks ago. Upon inspection, the logs show the following: May 19 22:35:14 kanga named[28855]: stale: impossible condition May 19 22:35:14 kanga named[28855]: stale: impossible condition Also, another box running 6.2 with updates had its named die today, the log showed: 06-Jun-2001 18:16:21.162 panic: critical: savedata: memget This issues are very sporadic and extremely serious.
I can't reproduce this. Please check if this still happens in the current version from rawhide (9.1.3-2). Marking RESOLVED RAWHIDE for now; if you can still reproduce this with the current bind, please reopen this bug.
Please let me know if the current version fixes this for you; I can't reproduce this on my testbox (and my production DNS is running 7.1).
It takes a month of running a bunch of nameservers to reproduce this; you can't tell me it's resolved by a package that hasn't been around that long.
alternatively, if 9.1.3-2 actually does fix the bug, please release it as an errata for 6.x and 7.x
I am certain that this particular problem does not exist in 9.* (since it's a rewrite) - but there might be other problems causing similar problems, and I'd rather not release a may-or-may-not-fix-a-bug-I-could-never-reproduce update. QA people, can you find a way to reproduce this and/or make sure the current version is not affected?
*** Bug 59905 has been marked as a duplicate of this bug. ***
closing because of inactivity. Seems to be resolved with bind-9