That looks like a reasonable fix. Would you like to try opening a PR to fix that?
On Fri, 19 Apr 2019 at 00:38, ahsankhanm...@gmail.com < ahsankhanm...@gmail.com> wrote: > In the code > https://github.com/jclouds/jclouds/blob/master/apis/ec2/src/main/java/org/jclouds/ec2/compute/EC2ComputeService.java#L151 > > If createNdoesInGroup fails to create a tag > (addTagsAndNamesToInstancesInRegion fails), it currently returns > AWSResponseException. The caller has no way to cleanup the the failed node > as the exception doesn't have the node id field. It should return > RunNodesException which has fields containing failed nodes. The caller can > then remove such node and retry. To repro the issue try to create a node > with name larger than 300 character, it will create nodes with empty name > and possibly multiple of them if caller retries. > > @Override > public Set<? extends NodeMetadata> createNodesInGroup(String group, int > count, final Template template) > throws RunNodesException { > Set<? extends NodeMetadata> nodes = super.createNodesInGroup(group, > count, template); > String region = > AWSUtils.getRegionFromLocationOrNull(template.getLocation()); > > if (client.getTagApiForRegion(region).isPresent()) { > Map<String, String> common = > metadataAndTagsAsValuesOfEmptyString(template.getOptions()); > if (generateInstanceNames || !common.isEmpty() || > !template.getOptions().getNodeNames().isEmpty()) { > return addTagsAndNamesToInstancesInRegion(common, > template.getOptions().getNodeNames(), > nodes, region, group); > } > } > return nodes; > } > >
