[
https://issues.apache.org/jira/browse/CLOUDSTACK-8749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14711275#comment-14711275
]
ASF GitHub Bot commented on CLOUDSTACK-8749:
--------------------------------------------
Github user jburwell commented on a diff in the pull request:
https://github.com/apache/cloudstack/pull/735#discussion_r37865908
--- Diff:
plugins/hypervisors/kvm/src/com/cloud/hypervisor/kvm/resource/BridgeVifDriver.java
---
@@ -286,7 +287,11 @@ private void deleteExitingLinkLocalRouteTable(String
linkLocalBr) {
String[] lines = parser.getLines().split("\\n");
for (String line : lines) {
String[] tokens = line.split(" ");
- if (!tokens[2].equalsIgnoreCase(linkLocalBr)) {
+ if (tokens != null && tokens.length < 2) {
+ continue;
+ }
+ final String device = tokens[2];
+ if (!Strings.isNullOrEmpty(device) &&
!device.equalsIgnoreCase(linkLocalBr)) {
--- End diff --
Reading through this block, it feels like lines 289-293 should be extracted
to a private method:
```
static parseDevice(final String line) {
final String[] tokens = line.split( " ");
if (tokens != null && tokens.length < 2) {
return "";
}
final String device = tokens[2];
return Strings.isNullOrEmpty(device) : " " ? device;
}
```
The extraction of this method would not only shorten the
deleteExitingLinkLocalLocalRoutetTable method, but clearly express the intent.
It also allows a unit test to be written to verify that parsing works as
expected for different types of input.
> KVM: link local route cleanup
> -----------------------------
>
> Key: CLOUDSTACK-8749
> URL: https://issues.apache.org/jira/browse/CLOUDSTACK-8749
> Project: CloudStack
> Issue Type: Bug
> Security Level: Public(Anyone can view this level - this is the
> default.)
> Reporter: Rohit Yadav
> Assignee: Rohit Yadav
> Fix For: 4.5.3, 4.6.0
>
>
> The KVM agent attempts to delete link local interfaces whenever found, so
> they don't conflict with the system vm's link local route on cloud0, however
> it doesn't specify which device to delete the route for.
> The aim is to fix the edge case which (unlikely) can try to delete existing
> cloud0 routes/interface.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
