You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@trafficcontrol.apache.org by GitBox <gi...@apache.org> on 2020/06/11 17:11:22 UTC

[GitHub] [trafficcontrol] zrhoffman opened a new pull request #4786: Permit edge -> edge topology node parentage

zrhoffman opened a new pull request #4786:
URL: https://github.com/apache/trafficcontrol/pull/4786


   <!--
   ************ STOP!! ************
   If this Pull Request is intended to fix a security vulnerability, DO NOT submit it! Instead, contact
   the Apache Software Foundation Security Team at security@trafficcontrol.apache.org and follow the
   guidelines at https://www.apache.org/security/ regarding vulnerability disclosure.
   -->
   ## What does this PR (Pull Request) do?
   <!-- Explain the changes you made here. If this fixes an Issue, identify it by
   replacing the text in the checkbox item with the Issue number e.g.
   
   - [x] This PR fixes #9001 OR is not related to any Issue
   
   ^ This will automatically close Issue number 9001 when the Pull Request is
   merged (The '#' is important).
   
   Be sure you check the box properly, see the "The following criteria are ALL
   met by this PR" section for details.
   -->
   - [x] This PR fixes #4785<!-- You can check for an issue here: https://github.com/apache/trafficcontrol/issues -->
   
   ## Which Traffic Control components are affected by this PR?
   <!-- Please delete all components from this list that are NOT affected by this
   Pull Request. Also, feel free to add the name of a tool or script that is
   affected but not on the list.
   
   Additionally, if this Pull Request does NOT affect documentation, please
   explain why documentation is not required. -->
   
   - Traffic Ops
   
   ## What is the best way to verify this PR?
   <!-- Please include here ALL the steps necessary to test your Pull Request. If
   it includes tests (and most should), outline here the steps needed to run the
   tests. If not, lay out the manual testing procedure and please explain why
   tests are unnecessary for this Pull Request. -->
   Run the topologies API tests in CDN-in-a-Box, make sure they pass.
   
   ## If this is a bug fix, what versions of Traffic Control are affected?
   <!-- If this PR fixes a bug, please list here all of the affected versions - to
   the best of your knowledge. It's also pretty helpful to include a commit hash
   of where 'master' is at the time this PR is opened (if it affects master),
   because what 'master' means will change over time. For example, if this PR
   fixes a bug that's present in master (at commit hash '1df853c8'), in v4.0.0,
   and in the current 4.0.1 Release candidate (e.g. RC1), then this list would
   look like:
   
   - master (1df853c8)
   - 4.0.0
   - 4.0.1 (RC1)
   
   If you don't know what other versions might have this bug, AND don't know how
   to find the commit hash of 'master', then feel free to leave this section
   blank (or, preferably, delete it entirely).
    -->
   Includes a bug fix. `GET /topologies` used to return `null` for 0 topologies, now it returns an empty array.
   
   - master (2dd9a0cdf1)
   
   Otherwise, not a bug fix.
   
   ## The following criteria are ALL met by this PR
   <!-- Check the boxes to signify that the associated statement is true. To
   "check a box", replace the space inside of the square brackets with an 'x'.
   e.g.
   
   - [ x] <- Wrong
   - [x ] <- Wrong
   - [] <- Wrong
   - [*] <- Wrong
   - [x] <- Correct!
   
   -->
   
   - [x] This PR includes tests OR I have explained why tests are unnecessary
   - [x] This PR includes Go docs
   - [ ] An update to CHANGELOG.md is not necessary
   - [x] This PR includes any and all required license headers
   - [x] This PR does not include a database migration
   - [x] This PR **DOES NOT FIX A SERIOUS SECURITY VULNERABILITY** (see [the Apache Software Foundation's security guidelines](https://www.apache.org/security/) for details)
   <!--
   Licensed to the Apache Software Foundation (ASF) under one
   or more contributor license agreements.  See the NOTICE file
   distributed with this work for additional information
   regarding copyright ownership.  The ASF licenses this file
   to you under the Apache License, Version 2.0 (the
   "License"); you may not use this file except in compliance
   with the License.  You may obtain a copy of the License at
   
       http://www.apache.org/licenses/LICENSE-2.0
   
   Unless required by applicable law or agreed to in writing,
   software distributed under the License is distributed on an
   "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
   KIND, either express or implied.  See the License for the
   specific language governing permissions and limitations
   under the License.
   -->
   


----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



[GitHub] [trafficcontrol] mitchell852 commented on pull request #4786: Permit edge -> edge topology node parentage

Posted by GitBox <gi...@apache.org>.
mitchell852 commented on pull request #4786:
URL: https://github.com/apache/trafficcontrol/pull/4786#issuecomment-644281064


   > The code changes look good, unit tests pass, and the TO API tests pass.
   > 
   > But I did notice something weird with TP. When I created an EDGE_LOC1 -> EDGE_LOC2 topology, TP let me click on the `2nd parent` button to make EDGE_LOC2 a 2nd parent of EDGE_LOC1. This should be prohibited via the API since it creates a cycle in the parentage, but it looks like TP is not actually adding the 2nd parent and sending that in the request:
   > 
   > ![Screen Shot 2020-06-12 at 3 14 27 PM](https://user-images.githubusercontent.com/6724053/84546986-cca02600-acbf-11ea-9b91-f744c8ff59ac.png)
   > ![Screen Shot 2020-06-12 at 3 15 19 PM](https://user-images.githubusercontent.com/6724053/84546989-ce69e980-acbf-11ea-8f1a-e368d9285b0d.png)
   > 
   > So the API creates the topology without the cyclic parentage (since TP scrubbed out that 2nd parent), but this behavior was surprising and should be prevented in the UI.
   > 
   > This issue seems to apply to MID_LOC cycles in the 2nd parentage as well (not just EDGE_LOC).
   
   ok @rawlinp, looking into...


----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



[GitHub] [trafficcontrol] mitchell852 edited a comment on pull request #4786: Permit edge -> edge topology node parentage

Posted by GitBox <gi...@apache.org>.
mitchell852 edited a comment on pull request #4786:
URL: https://github.com/apache/trafficcontrol/pull/4786#issuecomment-644281064


   > The code changes look good, unit tests pass, and the TO API tests pass.
   > 
   > But I did notice something weird with TP. When I created an EDGE_LOC1 -> EDGE_LOC2 topology, TP let me click on the `2nd parent` button to make EDGE_LOC2 a 2nd parent of EDGE_LOC1. This should be prohibited via the API since it creates a cycle in the parentage, but it looks like TP is not actually adding the 2nd parent and sending that in the request:
   > 
   > ![Screen Shot 2020-06-12 at 3 14 27 PM](https://user-images.githubusercontent.com/6724053/84546986-cca02600-acbf-11ea-9b91-f744c8ff59ac.png)
   > ![Screen Shot 2020-06-12 at 3 15 19 PM](https://user-images.githubusercontent.com/6724053/84546989-ce69e980-acbf-11ea-8f1a-e368d9285b0d.png)
   > 
   > So the API creates the topology without the cyclic parentage (since TP scrubbed out that 2nd parent), but this behavior was surprising and should be prevented in the UI.
   > 
   > This issue seems to apply to MID_LOC cycles in the 2nd parentage as well (not just EDGE_LOC).
   
   ok @rawlinp, looking into...
   
   first of all, that first edge_loc has no primary parent so a 2nd parent should not even be allowed...


----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



[GitHub] [trafficcontrol] rawlinp merged pull request #4786: Permit edge -> edge topology node parentage

Posted by GitBox <gi...@apache.org>.
rawlinp merged pull request #4786:
URL: https://github.com/apache/trafficcontrol/pull/4786


   


----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



[GitHub] [trafficcontrol] rawlinp commented on a change in pull request #4786: Permit edge -> edge topology node parentage

Posted by GitBox <gi...@apache.org>.
rawlinp commented on a change in pull request #4786:
URL: https://github.com/apache/trafficcontrol/pull/4786#discussion_r439622755



##########
File path: traffic_ops/traffic_ops_golang/topology/validation.go
##########
@@ -53,17 +53,41 @@ func checkForSelfParents(nodes []tc.TopologyNode, index int) error {
 	return nil
 }
 
-func checkForEdgeParents(nodes []tc.TopologyNode, cachegroups []tc.CacheGroupNullable, nodeIndex int) error {
-	node := nodes[nodeIndex]
+// checkForEdgeParents returns an error if an index given in the parents array, adds a warning + returns a nil error if
+// an edge parents an edge, and returns an error if an edge parents a non-edge cachegroup.
+func (topology *TOTopology) checkForEdgeParents(cacheGroups []tc.CacheGroupNullable, nodeIndex int) error {
+	node := topology.Nodes[nodeIndex]
 	errs := make([]error, len(node.Parents))
-	for parentIndex, cachegroupIndex := range node.Parents {
-		if cachegroupIndex < 0 || cachegroupIndex >= len(cachegroups) {
-			errs = append(errs, fmt.Errorf("parent %d of cachegroup %s refers to a cachegroup at index %d, but no such cachegroup exists", parentIndex, node.Cachegroup, cachegroupIndex))
+	for parentIndex, parentCacheGroupIndex := range node.Parents {
+		if parentCacheGroupIndex < 0 || parentCacheGroupIndex >= len(topology.Nodes) {
+			errs = append(errs, fmt.Errorf("parent %d of cachegroup %s refers to a cachegroup at index %d, but no such cachegroup exists", parentIndex, node.Cachegroup, parentCacheGroupIndex))
 			break
 		}
-		cacheGroupType := cachegroups[cachegroupIndex].Type
-		if *cacheGroupType == tc.CacheGroupEdgeTypeName {
-			errs = append(errs, fmt.Errorf("cachegroup %v's type is %v; it cannot be a parent of %v", nodes[cachegroupIndex].Cachegroup, tc.CacheGroupEdgeTypeName, node.Cachegroup))
+		parentCacheGroupType := *cacheGroups[parentCacheGroupIndex].Type
+		if parentCacheGroupType != tc.CacheGroupEdgeTypeName {
+			continue
+		}
+		switch cacheGroupType := *cacheGroups[nodeIndex].Type; cacheGroupType {
+		case tc.CacheGroupEdgeTypeName:
+			parentTerm := "parent"
+			if parentIndex == 1 {
+				parentTerm = "secondary " + parentTerm
+			}
+			topology.Alerts.AddNewAlert(tc.WarnLevel, fmt.Sprintf(
+				"%s-typed cachegroup %s is a %s of %s, unexpected behavior may result",
+				parentCacheGroupType,
+				topology.Nodes[parentCacheGroupIndex].Cachegroup,
+				parentTerm,
+				node.Cachegroup))
+			break

Review comment:
       nit: explicit break in switch cases is unnecessary -- this is done automatically in Go.




----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



[GitHub] [trafficcontrol] zrhoffman commented on a change in pull request #4786: Permit edge -> edge topology node parentage

Posted by GitBox <gi...@apache.org>.
zrhoffman commented on a change in pull request #4786:
URL: https://github.com/apache/trafficcontrol/pull/4786#discussion_r440192559



##########
File path: traffic_ops/traffic_ops_golang/topology/validation.go
##########
@@ -53,17 +53,41 @@ func checkForSelfParents(nodes []tc.TopologyNode, index int) error {
 	return nil
 }
 
-func checkForEdgeParents(nodes []tc.TopologyNode, cachegroups []tc.CacheGroupNullable, nodeIndex int) error {
-	node := nodes[nodeIndex]
+// checkForEdgeParents returns an error if an index given in the parents array, adds a warning + returns a nil error if
+// an edge parents an edge, and returns an error if an edge parents a non-edge cachegroup.
+func (topology *TOTopology) checkForEdgeParents(cacheGroups []tc.CacheGroupNullable, nodeIndex int) error {
+	node := topology.Nodes[nodeIndex]
 	errs := make([]error, len(node.Parents))
-	for parentIndex, cachegroupIndex := range node.Parents {
-		if cachegroupIndex < 0 || cachegroupIndex >= len(cachegroups) {
-			errs = append(errs, fmt.Errorf("parent %d of cachegroup %s refers to a cachegroup at index %d, but no such cachegroup exists", parentIndex, node.Cachegroup, cachegroupIndex))
+	for parentIndex, parentCacheGroupIndex := range node.Parents {
+		if parentCacheGroupIndex < 0 || parentCacheGroupIndex >= len(topology.Nodes) {
+			errs = append(errs, fmt.Errorf("parent %d of cachegroup %s refers to a cachegroup at index %d, but no such cachegroup exists", parentIndex, node.Cachegroup, parentCacheGroupIndex))
 			break
 		}
-		cacheGroupType := cachegroups[cachegroupIndex].Type
-		if *cacheGroupType == tc.CacheGroupEdgeTypeName {
-			errs = append(errs, fmt.Errorf("cachegroup %v's type is %v; it cannot be a parent of %v", nodes[cachegroupIndex].Cachegroup, tc.CacheGroupEdgeTypeName, node.Cachegroup))
+		parentCacheGroupType := *cacheGroups[parentCacheGroupIndex].Type
+		if parentCacheGroupType != tc.CacheGroupEdgeTypeName {
+			continue
+		}
+		switch cacheGroupType := *cacheGroups[nodeIndex].Type; cacheGroupType {
+		case tc.CacheGroupEdgeTypeName:
+			parentTerm := "parent"
+			if parentIndex == 1 {
+				parentTerm = "secondary " + parentTerm
+			}
+			topology.Alerts.AddNewAlert(tc.WarnLevel, fmt.Sprintf(
+				"%s-typed cachegroup %s is a %s of %s, unexpected behavior may result",
+				parentCacheGroupType,
+				topology.Nodes[parentCacheGroupIndex].Cachegroup,
+				parentTerm,
+				node.Cachegroup))
+			break

Review comment:
       Removed break in a70986165




----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



[GitHub] [trafficcontrol] mitchell852 edited a comment on pull request #4786: Permit edge -> edge topology node parentage

Posted by GitBox <gi...@apache.org>.
mitchell852 edited a comment on pull request #4786:
URL: https://github.com/apache/trafficcontrol/pull/4786#issuecomment-644281064


   > The code changes look good, unit tests pass, and the TO API tests pass.
   > 
   > But I did notice something weird with TP. When I created an EDGE_LOC1 -> EDGE_LOC2 topology, TP let me click on the `2nd parent` button to make EDGE_LOC2 a 2nd parent of EDGE_LOC1. This should be prohibited via the API since it creates a cycle in the parentage, but it looks like TP is not actually adding the 2nd parent and sending that in the request:
   > 
   > ![Screen Shot 2020-06-12 at 3 14 27 PM](https://user-images.githubusercontent.com/6724053/84546986-cca02600-acbf-11ea-9b91-f744c8ff59ac.png)
   > ![Screen Shot 2020-06-12 at 3 15 19 PM](https://user-images.githubusercontent.com/6724053/84546989-ce69e980-acbf-11ea-8f1a-e368d9285b0d.png)
   > 
   > So the API creates the topology without the cyclic parentage (since TP scrubbed out that 2nd parent), but this behavior was surprising and should be prevented in the UI.
   > 
   > This issue seems to apply to MID_LOC cycles in the 2nd parentage as well (not just EDGE_LOC).
   
   ok @rawlinp, looking into...
   
   first of all since, that first edge_loc has no primary parent so a 2nd parent should not even be allowed...


----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org