| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
|
|
|
|
| |
On very large clusters dnsmasq performance suffers due to
limited cache-size and dns-forward-max values.
bug: 1482847
https://bugzilla.redhat.com/show_bug.cgi?id=1482847
|
|
|
|
|
|
|
|
| |
Currently, we have to set the listen ip for dnsmasq via ansible
during installation.
This commit enables dnsmasq to bind-interfaces + exclude interfaces
to ensure dnsmasq doesn't listen on lo interface.
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
When running headless services as pods on OpenShift, no ClusterIP is
assigned. In these cases, peer pods are relying on DNS to locate the
service endpoints. When a pod is deleted or another trigger causes the
endpoint to change, the OpenShift DNS is updated immediately. However,
dnsmasq has a default TTL of 30s, so the wrong response is returned
on name resolution. Removing negative caching and turning the TTL to
a very short 1s should resolve the issue.
Signed-off-by: Steve Kuznetsov <skuznets@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
| |
strict-order forces dnsmasq to iterate through nameservers in order. If one of
the nameservers is down this will slow things down while dnsmasq waits for a
timeout. Also, this option prevents dnsmasq from querying other nameservers if
the first one returns a negative result. While I think it's odd to have a
nameserver that returns negative results for a query that another returns
positive results for this does seem to fix the issue in testing.
Fixes Bug 1399577
|
|
|