An old ddns script that worked fine on FC6/bind-9.3.4 now fails on F7/9.4.1 with this error: mem.c:866: INSIST((((ctx->pools).head == ((void *)0)) ? isc_boolean_true : isc_boolean_false)) failed. I *think* this is because it can no longer handle leading blank lines on stdin. In my case, it was an easy-to-read bash script using a here string: nsupdate -k $1 <<<" server $DDNSSERVER zone $DDNSZONE update delete $LHOSTNAME A update add $LHOSTNAME $TTL A $IPADDR send " The confusing and misleading error is the bug.
(In reply to comment #0) > I *think* this is because it can no longer handle leading blank lines on stdin. > The confusing and misleading error is the bug. It's memory-leak checking (if ctx->pools->head isn't NULL that means some memory have been allocated but not dealocated). It's low-severity bug because records are updated as expected (yes, in the end of nsupdate is some memory isn't dealocated but kernel take care about it ;) ). Already fixed in rawhide. I don't think that this has to be fixed in F7. If you want fix in F7, please reopen. Regards, Adam
Agreed. The F7 workaround is to clean up the input and thus appease nsupdate's persnicketiness.