CentOS - Oct 2014 - tinydns exceeds "holdoff time" on startup under CentOS 7

Hello listmates,

Somehow or other my DNS services that are part of
the ndjbdns-1.06-1.el7.x86_64 package would not start properly at startup.
When I then start them up using systemctl:

systemctl start dnscache
systemctl start tinydns

they start just fine.
>From the log I got the following for tinydns:
Oct 24 15:01:43 ns99 tinydns[1867]: tinydns: version 1.06: starting: Oct-24
2014 15:01:43 EDT
Oct 24 15:01:43 ns99 tinydns[1867]: tinydns: DEBUG_LEVEL set to `1'
Oct 24 15:01:43 ns99 tinydns[1867]: tinydns: DATALIMIT set to `300000' bytes
Oct 24 15:01:43 ns99 tinydns[1867]: tinydns: could not bind UDP socket
Oct 24 15:01:43 ns99 systemd[1]: tinydns.service holdoff time over,
scheduling restart.

Any idea why that would happen? Any idea how to increase the holdoff time
in the configuration?

The config for the service looks as follows:

[root at ns99 etc]# more /usr/lib/systemd/system/tinydns.service
[Unit]
Description=A DNS server daemon
Documentation=man:tinydns(8)
Requires=network.target
After=network.target

[Service]
Restart=always
PIDFile=/var/run/tinydns.pid
ExecStart=/usr/sbin/tinydns

[Install]
WantedBy=multi-user.target
[root at ns99 etc]#

I can't possibly spot anything wrong there.

Any help much appreciated.

Cheers,

Boris.

Hello again,

I think I have resolved this issue by adding the following line to my
relevant service startup files:

RestartSec=60s

I presume the line forces a restart within 60 seconds (or with the time
allowance of 60 seconds). Actually according to this source:

http://www.dsm.fordham.edu/cgi-bin/man-cgi.pl?topic=systemd.service&ampsect=5

it is the former - the sleep time before attempting a restart.

I put the line directly below the "Restart=..." line. See
my dnscache.service for example:

[root at ns99 system]# more /usr/lib/systemd/system/dnscache.service
[Unit]
Description=An iterative DNS resolver daemon
Documentation=man:dnscache(8)
Requires=network.target
After=network.target

[Service]
Restart=always
RestartSec=60s
PIDFile=/var/run/dnscache.pid
ExecStart=/usr/sbin/dnscache

[Install]
WantedBy=multi-user.target
[root at ns99 system]#

Cheers,

Boris.



On Mon, Oct 27, 2014 at 2:26 PM, Boris Epstein <borepstein at gmail.com>
wrote:
> Hello listmates,
>
> Somehow or other my DNS services that are part of
> the ndjbdns-1.06-1.el7.x86_64 package would not start properly at startup.
> When I then start them up using systemctl:
>
> systemctl start dnscache
> systemctl start tinydns
>
> they start just fine.
>
> From the log I got the following for tinydns:
>
> Oct 24 15:01:43 ns99 tinydns[1867]: tinydns: version 1.06: starting:
> Oct-24 2014 15:01:43 EDT
> Oct 24 15:01:43 ns99 tinydns[1867]: tinydns: DEBUG_LEVEL set to `1'
> Oct 24 15:01:43 ns99 tinydns[1867]: tinydns: DATALIMIT set to `300000'
> bytes
> Oct 24 15:01:43 ns99 tinydns[1867]: tinydns: could not bind UDP socket
> Oct 24 15:01:43 ns99 systemd[1]: tinydns.service holdoff time over,
> scheduling restart.
>
> Any idea why that would happen? Any idea how to increase the holdoff time
> in the configuration?
>
> The config for the service looks as follows:
>
> [root at ns99 etc]# more /usr/lib/systemd/system/tinydns.service
> [Unit]
> Description=A DNS server daemon
> Documentation=man:tinydns(8)
> Requires=network.target
> After=network.target
>
> [Service]
> Restart=always
> PIDFile=/var/run/tinydns.pid
> ExecStart=/usr/sbin/tinydns
>
> [Install]
> WantedBy=multi-user.target
> [root at ns99 etc]#
>
> I can't possibly spot anything wrong there.
>
> Any help much appreciated.
>
> Cheers,
>
> Boris.
>