Re: blocklistd[...]: Key size mismatch 296 != 288

In article <CAGN_6pYHuV89SbfeGzegcmA1zAXb-5WOHu8Q9Gh0hQqZY_5jdg@mail.gmail.com>,
David Brownlee  <[email protected]> wrote:
>On Mon, 17 Nov 2025 at 20:15, Christos Zoulas <[email protected]> wrote:
>>
>> In article
><CAGN_6pbyP154V4Cc6i0omN01x1mndEc9FK9nkW7FrEu4ECiwMA@mail.gmail.com>,
>> David Brownlee  <[email protected]> wrote:
>> >I'm getting a stream of these from several Internet visible NetBSD-11 system
>> >
>> >It implies that *something* is trying to interact with blocklistd and
>> >failing (or something malicious is afoot).
>> >
>> >Does anyone have any thoughts?
>>
>> Hmm, struct conf is indeed 296 bytes on x86_64 and 292 on i386 but I don't
>> see anything that is 288. This file (conf.h) has not changed nice 2025.
>> I would save the old db file into .old and restart blocklistd and see if
>> that happens again.
>
>That seems to have cleared the "Key size mismatch" issue - would it be
>possible to add some additional check to blocklistd to help people who
>have upgraded know that they would wipe their old db file?
>
>Now I'm seeing the following, which I'm assuming is "better"? :-p
>
>Nov 25 10:01:12 iris blocklistd[3946]: no message (Message too long)
>Nov 25 10:01:12 iris blocklistd[3946]: no message (Message too long)
>Nov 25 11:20:14 iris blocklistd[3946]: no message (Message too long)
>Nov 25 11:20:31 iris blocklistd[3946]: no message (Message too long)
>Nov 25 12:00:20 iris blocklistd[3946]: no message (Message too long)
>Nov 25 12:00:20 iris blocklistd[3946]: no message (Message too long)

That is weird, because this comes when bl_recv() fails,  but in all
cases where bl_recv() fails, it also logs so there should be more info.

christos