On Fri 25 Sep 2020 at 12:28:31 (+0000), Andy Smith wrote: > On Fri, Sep 25, 2020 at 07:49:19AM -0400, Greg Wooledge wrote: > > On Fri, Sep 25, 2020 at 07:44:25AM +0000, Andy Smith wrote: > > > "hostid" tends to return a hexadecimal representation of the first > > > IPv4 address (but isn't guaranteed to). > > > > unicorn:~$ hostid > > 007f0101 > > > > Doesn't look very useful. That's just 127.0.1.1 in a 16-bit little > > endian format. > > Oh, none of mine do that, it seems to pick the other IP address for > me. But if it's a problem there are other sources of "machine" ID as > I mentioned. There's some more here: > > http://0pointer.de/blog/projects/ids.html
IIRC Debian's recommendation is that a machine's own hosts entry should be: 127.0.0.1 localhost […] 127.0.1.1 axis.corp axis # 192.168.1.14 (The comment at the end is there because I generate my hosts file with a script that puts the 127.0.1.1 into place.) That might explain the monotonous 007f0101. Perhaps you don't set you hosts files that way. > > You know what else works really well? Just putting a different start > > time in each system's crontab. > > If that works for you, great, but I have quite a few machines, VMs > and containers provisioned identically and would rather not have to > change the scripts or configuration on a per-host basis. I don't know what scaling you require, nor the time resolution you can detect, but the last octet of the IPv4 address × 10 seconds gives you delays of up to ~42 minutes, unique on what was once called a "Class C" network. (15 seconds will go just over the hour.) Cheers, David.
