[knot-dns-users] Knot 1.6.1 and full journal

Marek Vavruša marek.vavrusa at nic.cz
Mon Dec 15 13:04:48 CET 2014


Just a follow-up - as Jan wrote, there's no file size shrinking
implemented in the journal. Moreover, the file does not cap exactly at
the configured limit, for example a journal may be treated as full if
the journal size is 9M, the configured limit is 10M, and the next
update requests >= 1M of space, the journal can't satisfy such result,
requests flush and then attempts to vacate old entries.
You can configure a slightly larger journal size to mitigate this, as
I wrote in the previous email.

Best,
Marek



On 15 December 2014 at 13:01, Jan Kadlec <jan.kadlec at nic.cz> wrote:
> Hello Anand,
>
> Knot will discard as many old records as needed to store the incoming entry,
> exactly how many depends on size of the incoming record and state of the
> journal. Also, when a zone is flushed to a text file, all journal entries
> are marked 'clean' and can later be discarded without flushing, but they
> will stay in the journal, as long as there is enough space for incoming
> records. One note about the journal file size: once the journal file reaches
> its maximum size, it will stay at that size indefinitely, although there
> might be free space in the journal.
>
> I hope this helps,
>
> Regards,
>
> Jan
>
>
> On 12/15/2014 11:22 AM, Anand Buddhdev wrote:
>>
>> Hi Knot developers,
>>
>> I have another question about journals. I've noticed that for one zone,
>> the journal size is 9M (with my configured limit at 10M).
>>
>> Now, I see this each time in the logs:
>>
>> 2014-12-15T07:56:02 notice: [103.in-addr.arpa] journal is full, flushing
>> 2014-12-15T08:13:09 notice: [103.in-addr.arpa] journal is full, flushing
>> 2014-12-15T08:16:35 notice: [103.in-addr.arpa] journal is full, flushing
>> 2014-12-15T08:20:27 notice: [103.in-addr.arpa] journal is full, flushing
>>
>> It looks like once a journal is close to the maximum size, then it just
>> remains at that size, with the result that each time an IXFR comes in,
>> and overflows the journal, Knot wants to flush the changes to disk
>> immediately. Does Knot discard old records from the journal at this point?
>>
>> Anand
>> _______________________________________________
>> knot-dns-users mailing list
>> knot-dns-users at lists.nic.cz
>> https://lists.nic.cz/cgi-bin/mailman/listinfo/knot-dns-users
>>
>
> _______________________________________________
> knot-dns-users mailing list
> knot-dns-users at lists.nic.cz
> https://lists.nic.cz/cgi-bin/mailman/listinfo/knot-dns-users


More information about the knot-dns-users mailing list