Most times my cluster comes up withouit a hitch, but I have started yesterday to have periodic problems with the CDH repository.
I am using a variation on the aws.reference.conf and twice out of 4 times over the last two days, I have recieved the following message displayed as output and in the logs:
* java.lang.IllegalArgumentException: CDH=5.4.7-1.cdh5.4.7.p0.3 not found in list of all parcels. ...
Here is the entry in my config specifying the parcel:
Should I be specifying a different repository or should my version entry be different to be more foolproof?
Does your "cluster" definition look like this? The CDH property should be the version, not the entire parcel name
Does the Director instance access the internet via a NAT instance? I'm asking because if it fails to download repository metadata to perform version validation it will silently ignore the failure but at the end of the process it will still flag the version as not being found in the list of parcel repositories.
Do you see any errors in the client log file (~/.cloudera-director/logs) at that time?
This could also be due to availability issues with archive.cloudera.com but I'm not aware of any recent incidents.
I want to add that we received similar reports through other channels and this seems to be a transient network failure. My recommendation is to try again.