zishuo/nip
1
0
Fork 0
mirror of https://github.com/cunnie/sslip.io.git synced 2025-10-04 07:16:33 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

Replace ambiguous metric, "Answered Queries"

Browse Source 
I've always been uncomfortable with the metric "Answered Queries" — it
implies that we don't answer all the queries. But we do answer all the
queries!

What the metric meant is "the number of DNS responses that we send that
have one or more records in the ANSWER section".

The new metric is "Answer ≥ 1". Not great, but better than before.
This commit is contained in:
Brian Cunnie
2024-11-19 09:25:10 -08:00
parent 3ed466bc74
commit 0fc3c81641
3 changed files with 25 additions and 26 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
integration_metrics_test.go
View File

@@ -205,12 +205,13 @@ func getMetrics(port int) (m xip.Metrics) {
Expect(err).ToNot(HaveOccurred())
var uptime int
var junk string
var greaterThanOrEqualsUnicode string
_, err = fmt.Sscanf(string(stdout),
"\"Uptime: %d\"\n"+
"\"Blocklist: %s %s %s\n"+
"\"Queries: %d (%s\n"+ // %s "swallows" the `/s"` at the end
"\"TCP/UDP: %d/%d\"\n"+
"\"Answered Queries: %d (%s\n"+ // %s "swallows" the `/s"` at the end
"\"Answer %s 1: %d (%s\n"+ // %s "swallows" the `/s"` at the end
"\"A: %d\"\n"+
"\"AAAA: %d\"\n"+
"\"TXT Source: %d\"\n"+
@@ -222,7 +223,7 @@ func getMetrics(port int) (m xip.Metrics) {
&junk, &junk, &junk,
&m.Queries, &junk,
&m.TCPQueries, &m.UDPQueries,
&m.AnsweredQueries, &junk,
&greaterThanOrEqualsUnicode, &m.AnsweredQueries, &junk,
&m.AnsweredAQueries,
&m.AnsweredAAAAQueries,
&m.AnsweredTXTSrcIPQueries,

44
k8s/document_root_sslip.io/index.html
View File

@@ -125,8 +125,8 @@ src="https://oss.maxcdn.com/respond/1.4.2/respond.min.js"></script> <![endif]-->
accomplish this, set the following three DNS servers as NS records for the subdomain “xip.example.com”</p>
<div class="alert alert-danger" role="alert">
<b>2024-11-16</b> <code>ns-aws.sslip.io</code> and <code>ns-azure.sslip.io</code> are deprecated. Please update
your nameservers to the nameservers
below. <code>ns-aws</code> and <code>ns-azure</code>will be shut down on <b>2024-12-25</b>.<br>
your nameservers to the nameservers below. <code>ns-aws</code> and <code>ns-azure</code>will be shut down on
<b>2024-12-25</b>.<br>
<br>
In October 2024, AWS charged me $113.88 for bandwidth for 1,265.3 GB at $0.09 / GB, and I am loath to spend
$1,366 on yearly bandwidth when other vendors, such as OVH and Hetzner, are much more reasonable.
@@ -147,13 +147,13 @@ src="https://oss.maxcdn.com/respond/1.4.2/respond.min.js"></script> <![endif]-->
</tr>
<tr class="even">
<td><code>ns-hetzner.sslip.io.</code></td>
<td>5.78.115.44<br />
<td>5.78.115.44<br>
2a01:4ff:1f0:c920::</td>
<td>USA</td>
</tr>
<tr class="odd">
<td><code>ns-ovh.sslip.io.</code></td>
<td>51.75.53.19<br />
<td>51.75.53.19<br>
2001:41d0:602:2313::1</td>
<td>Poland</td>
</tr>
@@ -267,7 +267,7 @@ dig @ns-azure.sslip.io metrics.status.sslip.io txt +short
"Blocklist: 2023-10-04 07:37:50-07 3,6"
"Queries: 14295231 (86.3/s)"
"TCP/UDP: 5231/14290000"
"Answered Queries: 4872793 (29.4/s)"
"Answer ≥ 1: 4872793 (29.4/s)"
"A: 4025711"
"AAAA: 247215"
"TXT Source: 57"
@@ -294,24 +294,23 @@ dig @ns-azure.sslip.io metrics.status.sslip.io txt +short
<dt>TCP/UDP</dt>
<dd>This is the number of queries received on the TCP protocol versus the UDP protocol. The sum should equal
the number of queries. DNS typically uses the UDP protocol</dd>
<dt>Answered Queries</dt>
<dt>Answer ≥ 1</dt>
<dd>This consists of two numbers: the first is the number of queries we responded to with at least one record
in the answer section, and the second is the first number divided by the uptime (i.e. queries/second). Note
that the number of answered queries is typically a third or fourth the size of the overall queries. This is
normal. One reason for this disparity is that often both the IPv4 (A) and IPv6 (AAAA) records will be checked,
but only one reply will have a record in the answer section . For example, browsing to "127.0.0.1.sslip.io"
generates two lookups, one with an answer (IPv4), and one without (IPv6). Another reason is that lookups
follow
a chain, e.g. looking up "127.0.0.1.sslip.io" may generate up to four queries for A records ("1.sslip.io",
"0.1.sslip.io", "0.0.1.sslip.io" and "127.0.0.1.sslip.io"), only the last of which returns a record in the
answer section. Pro-tip: if you want to shave milliseconds off name resolution, use dashes not dots in your
hostname (e.g. "10-9-9-30.sslip.io" instead of "10.9.9.30.sslip.io")</dd>
that the number of responses with an answer record is typically a fourth the size of the overall responses.
This is normal. One reason for this disparity is that often both the IPv4 (A) and IPv6 (AAAA) records will be
checked, but only one reply will have a record in the answer section . For example, browsing to
"127.0.0.1.sslip.io" generates two lookups, one with an answer (IPv4), and one without (IPv6). Another reason
is that lookups follow a chain, e.g. looking up "127.0.0.1.sslip.io" may generate up to four queries for A
records ("1.sslip.io", "0.1.sslip.io", "0.0.1.sslip.io" and "127.0.0.1.sslip.io"), only the last of which
returns a record in the answer section. Pro-tip: if you want to shave milliseconds off name resolution, use
dashes not dots in your hostname (e.g. "10-9-9-30.sslip.io" instead of "10.9.9.30.sslip.io")</dd>
<dt>A</dt>
<dd>The number of responses which included an A (IPv4) record since starting operation (e.g. "dig
127.0.0.1.sslip.io")</dd>
<dd>The number of responses which included an A (IPv4) record in the answer section since starting operation
(e.g. "dig 127.0.0.1.sslip.io")</dd>
<dt>AAAA</dt>
<dd>The number of responses which included an AAAA (IPv6) record since starting operation (e.g. "dig
--1.sslip.io aaaa")</dd>
<dd>The number of responses which included an AAAA (IPv6) record in the answer section since starting operation
(e.g. "dig --1.sslip.io aaaa")</dd>
<dt>TXT Source</dt>
<dd>The number of responses which included a TXT record of the querier's IP address since starting operation
(e.g. "dig @ns.sslip.io ip.sslip.io txt")</dd>
@@ -339,8 +338,8 @@ dig @ns-azure.sslip.io metrics.status.sslip.io txt +short
<a href="http://nip.io">nip.io</a>: similar to xip.io, but the PowerDNS backend is written in elegant Python
</li>
<li>
<a href="https://letsencrypt.org/">Let's Encrypt</a>: A Certificate Authority providing TLS certificates; they
have never failed to increase our rate limits when asked. If you can, <a
<a href="https://letsencrypt.org/">Let's Encrypt</a>: A Certificate Authority providing TLS certificates;
they have never failed to increase our rate limits when asked. If you can, <a
href="https://www.abetterinternet.org/donate/">donate</a>.
</li>
</ul>
@@ -349,8 +348,7 @@ dig @ns-azure.sslip.io metrics.status.sslip.io txt +short
<p><a id="status"><sup>[Status]</sup></a> A status of “build failing” rarely means the system is failing. It’s
more often an indication that when the servers were last checked (currently every six hours), the CI (continuous
integration) <a href="https://ci.nono.io/teams/main/pipelines/sslip.io">server</a> had difficulty reaching one
of
the three sslip.io name servers. That’s normal. <sup><a href="#timeout" class="alert-link">[connection timed
of the three sslip.io name servers. That’s normal. <sup><a href="#timeout" class="alert-link">[connection timed
out]</a></sup></p>
<p><a id="timeout"><sup>[connection timed out]</sup></a></p>
<p>DNS runs over <a href="https://en.wikipedia.org/wiki/User_Datagram_Protocol">UDP</a> which has no guaranteed

2
xip/xip.go
View File

@@ -985,7 +985,7 @@ func TXTMetrics(x *Xip, _ net.IP) (txtResources []dnsmessage.TXTResource, err er
len(x.BlocklistCIDRs)))
metrics = append(metrics, fmt.Sprintf("Queries: %d (%.1f/s)", x.Metrics.Queries, float64(x.Metrics.Queries)/uptime.Seconds()))
metrics = append(metrics, fmt.Sprintf("TCP/UDP: %d/%d", x.Metrics.TCPQueries, x.Metrics.UDPQueries))
metrics = append(metrics, fmt.Sprintf("Answered Queries: %d (%.1f/s)", x.Metrics.AnsweredQueries, float64(x.Metrics.AnsweredQueries)/uptime.Seconds()))
metrics = append(metrics, fmt.Sprintf("Answer ≥ 1: %d (%.1f/s)", x.Metrics.AnsweredQueries, float64(x.Metrics.AnsweredQueries)/uptime.Seconds()))
metrics = append(metrics, fmt.Sprintf("A: %d", x.Metrics.AnsweredAQueries))
metrics = append(metrics, fmt.Sprintf("AAAA: %d", x.Metrics.AnsweredAAAAQueries))
metrics = append(metrics, fmt.Sprintf("TXT Source: %d", x.Metrics.AnsweredTXTSrcIPQueries))