wait_for_network waits longer than two minutes

[jrha]

The current code appears to assume that nslookup fails instantly, however the default behaviour is to make three requests, each with a timeout of 5 seconds, so a host with completely broken networking will take 120 * (15 + 1) seconds (32 minutes) to error out. This has started to become an issue for us as we make more use of more dynamic environments.

We could force the behaviour the code expects by adding -retry=1 -timeout=1 to nslookup, although the sleep would still make the loop take longer than two minutes.

Or we could remove the loop completely and just use something like nslookup -retry=12 -timeout=10

Current code:

# Wait for functional network up by testing DNS lookup via nslookup.
wait_for_network () {
    # Wait up to 2 minutes until the network comes up
    i=0
    while ! nslookup \$1 > /dev/null
    do
        sleep 1
        let i=\$i+1
        if [ \$i -gt 120 ]
        then
            fail "Network does not come up (nslookup \$1)"
        fi
    done
}