Chuck,
It's interesting, I will try to reproduce it. Could you please provide me:
- Approximate number of zones.
- Approximate size of the zones.
- How often the zones are updated.
- Is the server fully utilized or waits for something.
- Some log outputs could help (in private if necessary).
Thanks,
Daniel
On 01/30/2018 09:05 PM, Chuck Musser wrote:
On Jan 30, 2018, at 8:37 AM,
daniel.salzman(a)nic.cz wrote:
Hi Chuck,
I'm unaware of the culprit at the moment, but I would try:
- Increasing the zone file synchronization delay via
https://www.knot-dns.cz/docs/2.6/singlehtml/index.html#zonefile-sync
- Changing the journal database mode to asynchronous via
https://www.knot-dns.cz/docs/2.6/singlehtml/index.html#journal-db-mode
- Switching off the zone journal via
https://www.knot-dns.cz/docs/2.6/singlehtml/index.html#journal-content
Regards,
Daniel
No luck so far. We already had zonefile-sync set to -1 (no sync). I tried running
with journal-db-mode: asynchronous for a while, then switched to journal-content: none. In
both cases, the slave fell behind again after several minutes of operation.
Chuck