DNS server not responding, lwresd not servicing incoming queries

Kory Keith korykeith.tx at gmail.com
Tue Dec 6 17:06:53 UTC 2016


I see in BIND 9.11.0, NTASKS and NRECV have been removed in place of a
configurable value with max of 32K - is this change to address issues such
as this? or for other reasons?

Thanks,

Kory

On Mon, Dec 5, 2016 at 5:06 PM, Kory Keith <korykeith.tx at gmail.com> wrote:

> My symptoms are like this: if I'm sending queries from my application at a
> slow, but steady rate, all is well for cached entries and when the server
> responds in a timely manner. When the server does not respond or not in a
> timely manner, my application receives no responses (even for cached
> entries) for sometimes seconds at a time - based on the configured
> timeout/retry.
>
> In lwdclient.c, I am using the default values of 2 tasks (NTASKS) and 2
> clients per task (NRECV).
>
> If I run a series of queries that all miss the cached entries and result
> in an external query, it appears from studying the code that those clients
> are blocked in the running state until a response is received or until the
> request times out.
>
> The only time I see the client returning to the IDLE list is after sending
> a reply (other than the few error scenarios such as failed socket read,
> etc.)
>
> If this is the intended behavior, am I correct that a network issue where
> a DNS server (or servers) are non-responsive or slow to respond could cause
> all traffic to back up coming into lwresd, even for the cached entries?
>
> Thank you,
>
> Kory
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.isc.org/pipermail/bind-workers/attachments/20161206/ef47a6ae/attachment.html>


More information about the bind-workers mailing list