Repository navigation
New rules cause dnsmasq to be restarted instead of reloaded #3613
Description
When adding/configuring new dhcp/dns rules, the dnsmasq process is restarted:
https://github.com/apache/cloudstack/blob/master/systemvm/debian/opt/cloud/bin/cs/CsDhcp.py#L71
Explore if it's possible to simply reload the dnsmasq process than restart it.
(If dnsmasq restarts at the same time monitoring service check kicks in, it might show to recover the dnsmasq again)
ISSUE TYPE
- Improvement Request
COMPONENT NAME
VR
Activity
onitake commented on Oct 1, 2019
Restarting should not be necessary for reloading DNS and DHCP records, according to the man page: http://www.thekelleys.org.uk/dnsmasq/docs/dnsmasq-man.html#lbAG
Quote:
The systemd unit in Debian buster sends a SIGHUP on reload, so it should suffice:
andrijapanicsb commented on Oct 2, 2019
it WAS necessary some days (years?) ago since there was no way to reload the dnsmasq (or it was buggy) - but today - you are right.
yadvr commented on Jan 28, 2020
This will require large changes, moved to 4.15
yadvr commented on Aug 9, 2021
It may be worth discussing and move it out if the scope/changes are huge. cc @nvazquez @weizhouapache
pdion891 commented on Aug 16, 2021
Reloading dnsmasq when creating a new vm would probably work, but when you introduce a new network tier in a VPC, you will probably have to restart dnsmasq so it introduce the new network interface of the VR in dnsmasq configuration.
weizhouapache commented on Aug 16, 2021
@pdion891
yes, this is why I keep the following lines in CsDhcp.py in #5319
if self.conf.commit(): restart_dnsmasq = Trueweizhouapache commented on Sep 3, 2021
fixed by #5319