timmylicheng commented on a change in pull request #678: HDDS-3179 Pipeline
placement based on Topology does not have fallback
URL: https://github.com/apache/hadoop-ozone/pull/678#discussion_r392677663
##########
File path:
hadoop-hdds/server-scm/src/main/java/org/apache/hadoop/hdds/scm/pipeline/PipelinePlacementPolicy.java
##########
@@ -292,10 +294,15 @@ DatanodeDetails fallBackPickNodes(
int nodesToFind = nodesRequired - results.size();
for (int x = 0; x < nodesToFind; x++) {
// Pick remaining nodes based on the existence of rack awareness.
- DatanodeDetails pick = rackAwareness
- ? chooseNodeFromNetworkTopology(
- nodeManager.getClusterNetworkTopologyMap(), anchor, exclude)
- : fallBackPickNodes(healthyNodes, exclude);
+ DatanodeDetails pick;
+ try {
+ pick = rackAwareness
+ ? chooseNodeFromNetworkTopology(
+ nodeManager.getClusterNetworkTopologyMap(), anchor, exclude)
+ : fallBackPickNodes(healthyNodes, exclude);
+ } catch (SCMException e) {
Review comment:
Debug log makes sense.
My thought was: fall back will happen every time when there is only one node
on the rack. Under all other scenarios, this fallback logic shouldn't take
effects. As a known case, I was reluctant to add logs for it regardless of its
log level. But I guess it can be valuable if fallback logic takes effect in
other cases.
----------------------------------------------------------------
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:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
