Hi,
the log line you posted got a bit mangled in the copy/paste, so let me first try to put it back together, then tell you what I think is behind it. I'm fairly confident about one part, but for the other we'd really need the complete line to avoid guessing.
My read is that "SHARADA" is simply your server's hostname that ended up in the middle of the message when the line got reassembled (in AIX syslog format the hostname sits before daemon:err). The interesting bit is "interfaceite_create", which is missing an "r": the actual function is isc_interfaceiter_create(), which in BIND is the routine that enumerates the machine's network interfaces to figure out which addresses to listen on. So what the message is telling you is: every time named runs its periodic interface scan, the enumeration fails. WHY it fails is sitting exactly in the part of the line that got cut off after the colon, if you can post the full line (and maybe the output of named -V) we can dig further.
As for the "every minute" part, I think I can give you a fairly precise answer, because the timing lines up too well to be a coincidence. All 9.18 releases up to .37 carry a known bug (GL #5246): when ISC changed the interface-interval parser, the numeric default of 60 went from meaning 60 minutes to meaning 60 seconds. Result: named rescans the interfaces every minute instead of every hour. On Linux hardly anyone noticed because the scan succeeds silently; on your box it fails, so you get one error line per minute. The fix landed in 9.18.38 (July 2025):
• ISC merge request: https://gitlab.isc.org/isc-projects/bind9/-/merge_requests/10680
• 9.18.38 release notes (look for GL #5246): https://bind9.readthedocs.io/en/v9.18.38/notes.html
Here's what I'd do, in order:
1. First check that named is actually listening on all the IPs you expect (netstat -an | grep .53). If it is, the practical impact today is mostly log noise, with the caveat that if you ever add an IP alias or a new interface, named might not pick it up until a reload.
2. As a stopgap I'd put an explicit interface-interval in named.conf, with the suffix: interface-interval 1h; careful NOT to write the bare number 60, because on these versions it would again be interpreted as seconds. With 1h the error drops from 1440 to 24 occurrences a day. You could also set interface-interval 0; to silence it completely, but then the scan only happens at startup/reload (the mechanics are well explained in the ISC KB: https://kb.isc.org/docs/aa-00420), on a DNS server with static IPs that may be perfectly acceptable, your call.
3. The real fix is getting to 9.18.38 or later. Here it depends on where your binary comes from: if it's the named shipped with AIX (check with lslpp -L | grep -I bind, on recent AIX levels IBM ships 9.18 as named918, there's a "named daemon - BIND version 9.18" page on IBM Docs), then the right channel is a case with IBM Support, since they build and support that binary: I'd attach the complete log line and reference GL #5246. If it's compiled in-house, just rebuild a recent 9.18.
One last note for honesty's sake: the .38 fix addresses the frequency (you'd be back to one error per hour instead of per minute), but the fact that interface enumeration fails on your AIX at all is a separate issue that this fix doesn't touch. To pin that down we'd really need the full line, plus a couple of details about the box: how many interfaces/aliases you have (ifconfig -a) and whether IPv6 is active. I couldn't find any IBM APAR or public ISC issue matching this specific message on AIX, so you may well be among the first to run into it: post the details and we can try to narrow it down.
Regards
Roberto