apitech
/
loki
mirror of https://github.com/grafana/loki
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

Display replicas, cores and nodes in sizing tool. (#8088)

Browse Source 
**What this PR does / why we need it**:
No every users wants to deploy Loki with Helm. That's why we are going
to display the required replicas as well.


![image](https://user-images.githubusercontent.com/2418440/212642958-17bc3fcd-0f6a-4ea8-8366-a3c28ddc6622.png)

This change also fixes two bugs in the algorithm and introduces some
tests for its invariant.

**Checklist**
- [ ] Reviewed the `CONTRIBUTING.md` guide
- [x] Documentation added
- [x] Tests updated
- [ ] `CHANGELOG.md` updated
- [ ] Changes that require user attention or interaction to upgrade are
documented in `docs/sources/upgrading/_index.md`

Signed-off-by: dependabot[bot] <support@github.com>
Signed-off-by: Kaviraj <kavirajkanagaraj@gmail.com>
Signed-off-by: Danny Kopping <danny.kopping@grafana.com>
Signed-off-by: Ashwanth Goli <iamashwanth@gmail.com>
Co-authored-by: Dimitar Dimitrov <dimitar.dimitrov@grafana.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Danny Kopping <danny.kopping@grafana.com>
Co-authored-by: Kaviraj Kanagaraj <kavirajkanagaraj@gmail.com>
Co-authored-by: liam-howe-maersk <118169897+liam-howe-maersk@users.noreply.github.com>
Co-authored-by: JordanRushing <rushing.jordan@gmail.com>
Co-authored-by: Trevor Whitney <trevorjwhitney@gmail.com>
Co-authored-by: Periklis Tsirakidis <periklis@redhat.com>
Co-authored-by: Ashwanth Goli <iamashwanth@gmail.com>
pull/8182/head
Karsten Jeschkies 2 years ago committed by GitHub
parent da76b1394d
commit f24da4a194
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
7 changed files with 105 additions and 30 deletions
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Show Stats Download Patch File Download Diff File
  1. 1
      cmd/logql-analyzer/main.go
  2. 47
      docs/sources/installation/sizing/index.md
  3. 27
      pkg/sizing/algorithm.go
  4. 25
      pkg/sizing/algorithm_test.go
  5. 4
      pkg/sizing/helm.go
  6. 29
      pkg/sizing/http.go
  7. 2
      pkg/sizing/node.go

1
cmd/logql-analyzer/main.go
Unescape View File

@ -52,6 +52,7 @@ func createServer(cfg server.Config, logger log.Logger) (*server.Server, error)
s.HTTP.Handle("/api/sizing/helm", http.HandlerFunc(sizingHandler.GenerateHelmValues)).Methods(http.MethodGet, http.MethodOptions)
s.HTTP.Handle("/api/sizing/nodes", http.HandlerFunc(sizingHandler.Nodes)).Methods(http.MethodGet, http.MethodOptions)
s.HTTP.Handle("/api/sizing/cluster", http.HandlerFunc(sizingHandler.Cluster)).Methods(http.MethodGet, http.MethodOptions)
s.HTTP.HandleFunc("/ready", func(w http.ResponseWriter, _ *http.Request) {
http.Error(w, "ready", http.StatusOK)

47
docs/sources/installation/sizing/index.md
Unescape View File

@ -20,18 +20,18 @@ This tool helps to generate a Helm Charts `values.yaml` file based on specified
<div id="app">
<label class="icon question" v-on:mouseover="help='node'" v-on:mouseleave="help=null">Node Type</label>
<label>Node Type<i class="fa fa-question" v-on:mouseover="help='node'" v-on:mouseleave="help=null"></i></label>
<select name="node-type" v-model="node">
<option v-for="node of nodes">{{ node }}</option>
</select>
<label class="fa fa-question" v-on:mouseover="help='ingest'" v-on:mouseleave="help=null">Ingest</label>
<label>Ingest<i class="fa fa-question" v-on:mouseover="help='ingest'" v-on:mouseleave="help=null"></i></label>
<input v-model="ingest" name="ingest" placeholder="Desired ingest in GiB/day" type="number" max="1048576" min="0"/>
<label class="fa fa-question" v-on:mouseover="help='retention'" v-on:mouseleave="help=null">Log retention period</label>
<label>Log retention period<i class="fa fa-question" v-on:mouseover="help='retention'" v-on:mouseleave="help=null"></i></label>
<input v-model="retention" name="retention" placeholder="Desired retention period in days" type="number" min="0"/>
<label class="fa fa-question" v-on:mouseover="help='queryperf'" v-on:mouseleave="help=null">Query performance</label>
<label>Query performance<i class="fa fa-question" v-on:mouseover="help='queryperf'" v-on:mouseleave="help=null"></i></label>
<div id="queryperf" style="display: inline-flex;">
<label for="basic">
<input type="radio" id="basic" value="Basic" v-model="queryperf"/>Basic
@ -42,6 +42,23 @@ This tool helps to generate a Helm Charts `values.yaml` file based on specified
</label>
</div>
<div v-if="clusterSize">
<table>
<tr>
<th>Read Replicas</th>
<th>Write Replicas</th>
<th>Nodes</th>
<th>Cores</th>
</tr>
<tr>
<td>{{ clusterSize.TotalReadReplicas }}</td>
<td>{{ clusterSize.TotalWriteReplicas }}</td>
<td>{{ clusterSize.TotalNodes}}</td>
<td>{{ clusterSize.TotalCoresRequest}}</td>
</tr>
</table>
</div>
<a v-bind:href="helmURL" class="primary-button">Generate and download values file</a>
<blockquote v-if="help">
@ -97,14 +114,18 @@ createApp({
ingest: null,
retention: null,
queryperf: 'Basic',
help: null
help: null,
clusterSize: null
}
},
computed: {
helmURL() {
return `${API_URL}/helm?${this.queryString}`
},
queryString() {
const bytesDayIngest = this.ingest * 1024 * 1024 * 1024
return `${API_URL}/helm?node-type=${encodeURIComponent(this.node)}&ingest=${encodeURIComponent(bytesDayIngest)}&retention=${encodeURIComponent(this.retention)}&queryperf=${encodeURIComponent(this.queryperf)}`
return `node-type=${encodeURIComponent(this.node)}&ingest=${encodeURIComponent(bytesDayIngest)}&retention=${encodeURIComponent(this.retention)}&queryperf=${encodeURIComponent(this.queryperf)}`
}
},
@ -117,7 +138,21 @@ createApp({
async fetchNodeTypes() {
const url = `${API_URL}/nodes`
this.nodes = await (await fetch(url,{mode: 'cors'})).json()
},
async calculateClusterSize() {
if (this.node == 'Loading...' || this.ingest == null || this.retention == null) {
return
}
const url = `${API_URL}/cluster?${this.queryString}`
this.clusterSize = await (await fetch(url,{mode: 'cors'})).json()
}
},
watch: {
node: 'calculateClusterSize',
ingest: 'calculateClusterSize',
retention: 'calculateClusterSize',
queryperf: 'calculateClusterSize'
}
}).mount('#app')
</script>

27
pkg/sizing/algorithm.go
Unescape View File

@ -5,9 +5,10 @@ import (
)
type ClusterSize struct {
totalNodes int
totalReadReplicas int
totalWriteReplicas int
TotalNodes int
TotalReadReplicas int
TotalWriteReplicas int
TotalCoresRequest float64
expectedMaxReadThroughputBytesSec float64
expectedMaxIngestBytesDay float64
@ -23,7 +24,7 @@ const (
func calculateClusterSize(nt NodeType, bytesDayIngest float64, qperf QueryPerf) ClusterSize {
// 1 Petabyte per day is maximum. We use decimal prefix https://en.wikipedia.org/wiki/Binary_prefix
bytesDayIngest = math.Min(bytesDayIngest, 1e12)
bytesDayIngest = math.Min(bytesDayIngest, 1e15)
bytesSecondIngest := bytesDayIngest / 86400
numWriteReplicasNeeded := math.Ceil(bytesSecondIngest / nt.writePod.rateBytesSecond)
@ -33,7 +34,7 @@ func calculateClusterSize(nt NodeType, bytesDayIngest float64, qperf QueryPerf)
replicasOnLastNode := math.Mod(numWriteReplicasNeeded, writeReplicasPerNode)
coresOnLastNode := 0.0
if replicasOnLastNode >= 0.0 {
if replicasOnLastNode > 0.0 {
coresOnLastNode = math.Max(float64(nt.cores)-replicasOnLastNode*nt.writePod.cpuRequest, 0.0)
}
@ -44,12 +45,13 @@ func calculateClusterSize(nt NodeType, bytesDayIngest float64, qperf QueryPerf)
readReplicasOnFullyPackedWriteNodes := readReplicasPerNode * fullyWritePackedNodes
readReplicasOnPartiallyPackedWriteNodes := math.Floor(coresOnLastNode / nt.readPod.cpuRequest)
basicQperfReadReplicas := readReplicasOnFullyPackedWriteNodes + readReplicasOnPartiallyPackedWriteNodes
// Required read replicase without considering required query performance.
baselineReadReplicas := readReplicasOnFullyPackedWriteNodes + readReplicasOnPartiallyPackedWriteNodes
scaleUp := 0.25
additionalReadReplicas := 0.0
if qperf != Basic {
additionalReadReplicas = basicQperfReadReplicas * scaleUp
additionalReadReplicas = baselineReadReplicas * scaleUp
}
readReplicasPerEmptyNode := math.Floor(float64(nt.cores) / nt.readPod.cpuRequest)
@ -58,15 +60,16 @@ func calculateClusterSize(nt NodeType, bytesDayIngest float64, qperf QueryPerf)
actualNodesAddedForReads := calculateActualReadNodes(additionalNodesNeededForReads)
actualReadReplicasAdded := actualNodesAddedForReads * readReplicasPerEmptyNode
totalReadReplicas := actualReadReplicasAdded + basicQperfReadReplicas
totalReadReplicas := actualReadReplicasAdded + baselineReadReplicas
totalReadThroughputBytesSec := totalReadReplicas * nt.readPod.rateBytesSecond
totalNodesNeeded := nodesNeededForWrites + actualNodesAddedForReads
totalCoresLimit := numWriteReplicasNeeded*nt.writePod.cpuRequest + totalReadReplicas*nt.readPod.cpuRequest
return ClusterSize{
totalNodes: int(totalNodesNeeded),
totalReadReplicas: int(totalReadReplicas),
totalWriteReplicas: int(numWriteReplicasNeeded),
TotalNodes: int(totalNodesNeeded),
TotalReadReplicas: int(totalReadReplicas),
TotalWriteReplicas: int(numWriteReplicasNeeded),
TotalCoresRequest: totalCoresLimit,
expectedMaxReadThroughputBytesSec: totalReadThroughputBytesSec,
expectedMaxIngestBytesDay: (nt.writePod.rateBytesSecond * numWriteReplicasNeeded) * 86400,

25
pkg/sizing/algorithm_test.go
Unescape View File

@ -7,7 +7,7 @@ import (
"github.com/stretchr/testify/require"
)
func Test_AlgorithmTest_Algorithm(t *testing.T) {
func Test_Algorithm(t *testing.T) {
f := func(ingest float64) bool {
if ingest < 0 {
ingest = -ingest
@ -16,12 +16,12 @@ func Test_AlgorithmTest_Algorithm(t *testing.T) {
for _, cloud := range NodeTypesByProvider {
for _, node := range cloud {
size := calculateClusterSize(node, ingest, Basic)
postiveReplicas = size.totalNodes > 0.0 && size.totalReadReplicas > 0.0 && size.totalWriteReplicas > 0.0
postiveReplicas = size.TotalNodes > 0.0 && size.TotalReadReplicas > 0.0 && size.TotalWriteReplicas > 0.0
require.Truef(t, postiveReplicas, "Cluster size was empty: ingest=%d cluster=%v node=%v", ingest, size, node)
require.InDelta(t, size.totalReadReplicas, size.totalWriteReplicas, 5.0, "Replicas have different sizes: ingest=%d node=%s", ingest, node.name)
require.InDelta(t, size.TotalReadReplicas, size.TotalWriteReplicas, 5.0, "Replicas have different sizes: ingest=%d node=%s", ingest, node.name)
size = calculateClusterSize(node, ingest, Super)
postiveReplicas = size.totalNodes > 0.0 && size.totalReadReplicas > 0.0 && size.totalWriteReplicas > 0.0
postiveReplicas = size.TotalNodes > 0.0 && size.TotalReadReplicas > 0.0 && size.TotalWriteReplicas > 0.0
require.Truef(t, postiveReplicas, "Cluster size was empty: ingest=%d cluster=%v node=%v", ingest, size, node)
}
}
@ -32,4 +32,21 @@ func Test_AlgorithmTest_Algorithm(t *testing.T) {
if err := quick.Check(f, nil); err != nil {
t.Error(err)
}
// Sanity check for 1TB/Day
size := calculateClusterSize(NodeTypesByProvider["AWS"]["t2.xlarge"], 1e12, Basic)
require.Equalf(t, 4, size.TotalNodes, "given ingest=1PB/Day totla nodes must be big")
}
func Test_CoresNodeInvariant(t *testing.T) {
for _, queryPerformance := range []QueryPerf{Basic, Super} {
for _, ingest := range []float64{30, 300, 1000, 2000} {
for _, cloud := range NodeTypesByProvider {
for _, node := range cloud {
size := calculateClusterSize(node, ingest, queryPerformance)
require.LessOrEqualf(t, size.TotalCoresRequest, float64(size.TotalNodes*node.cores), "given ingest=%d node=%s total cores must be less than available cores", ingest, node.name)
}
}
}
}
}

4
pkg/sizing/helm.go
Unescape View File

@ -50,7 +50,7 @@ func constructHelmValues(cluster ClusterSize, nodeType NodeType) Values {
AuthEnabled: false,
},
Read: Read{
Replicas: cluster.totalReadReplicas,
Replicas: cluster.TotalReadReplicas,
Resources: Resources{
Requests: struct {
CPU float64 `json:"cpu"`
@ -69,7 +69,7 @@ func constructHelmValues(cluster ClusterSize, nodeType NodeType) Values {
},
},
Write: Write{
Replicas: cluster.totalWriteReplicas,
Replicas: cluster.TotalWriteReplicas,
Resources: Resources{
Requests: struct {
CPU float64 `json:"cpu"`

29
pkg/sizing/http.go
Unescape View File

@ -33,12 +33,12 @@ func decodeMesage(req *http.Request, msg *Message) error {
msg.Ingest, err = strconv.Atoi(req.FormValue("ingest"))
if err != nil {
return err
return fmt.Errorf("cannot read ingest: %w", err)
}
msg.Retention, err = strconv.Atoi(req.FormValue("retention"))
if err != nil {
return err
return fmt.Errorf("cannot read retention: %w", err)
}
msg.QueryPerformance = QueryPerf(strings.ToLower(req.FormValue("queryperf")))
@ -73,7 +73,7 @@ func (h *Handler) GenerateHelmValues(w http.ResponseWriter, req *http.Request) {
enc := yaml.NewEncoder(w)
err = enc.Encode(helm)
if err != nil {
level.Error(h.logger).Log("msg", "could not encode Helm Char values", "error", err)
level.Error(h.logger).Log("msg", "could not encode Helm Chart values", "error", err)
}
}
@ -88,7 +88,7 @@ func (h *Handler) Nodes(w http.ResponseWriter, req *http.Request) {
w.Header().Set("Content-Type", "application/json")
err := json.NewEncoder(w).Encode(nodes)
if err != nil {
level.Error(h.logger).Log("msg", "could not encode Helm Char values", "error", err)
level.Error(h.logger).Log("msg", "could not encode node values", "error", err)
}
}
@ -96,6 +96,25 @@ func (h *Handler) respondError(w http.ResponseWriter, err error) {
w.WriteHeader(http.StatusBadRequest)
_, err = w.Write([]byte(fmt.Sprintf("error: %v", err)))
if err != nil {
level.Error(h.logger).Log("msg", "could not encode Helm Char values", "error", err)
level.Error(h.logger).Log("msg", "could not write error message", "error", err)
}
}
func (h *Handler) Cluster(w http.ResponseWriter, req *http.Request) {
var msg Message
err := decodeMesage(req, &msg)
if err != nil {
level.Error(h.logger).Log("error", err)
h.respondError(w, err)
return
}
cluster := calculateClusterSize(msg.NodeType, float64(msg.Ingest), msg.QueryPerformance)
w.Header().Set("Content-Type", "application/json")
err = json.NewEncoder(w).Encode(cluster)
if err != nil {
level.Error(h.logger).Log("msg", "could not encode cluster size", "error", err)
}
}

2
pkg/sizing/node.go
Unescape View File

@ -26,7 +26,7 @@ var StandardWrite = NodePod{
var StandardRead = NodePod{
cpuRequest: 3,
cpuLimit: 0, // Undefined
cpuLimit: 3, // Undefined TODO: Is this a bug
memoryRequest: 6,
memoryLimit: 8,
rateBytesSecond: 768 * 1024 * 1024,

Write Preview
Loading…
Cancel
Save