Description of problem:
With the deployment of Cluster Logging 4.5 there are now 6 new cronjobs which deletes/rolls over log files in ElasticSearch. It has been seen that both the delete and rollover cronjobs fail occasionally (all log types i.e. app, infra, audit). These jobs seem to have the same error message for their respective purpose, any time they fail, AKA for rollover cronjobs the errors are:

{
  "error" : {
    "root_cause" : [
      {
        "type" : "security_exception",
        "reason" : "Unexpected exception indices:admin/rollover"
      }
    ],
    "type" : "security_exception",
    "reason" : "Unexpected exception indices:admin/rollover"
  },
  "status" : 500
}

for delete jobs the logs appear like:

Traceback (most recent call last):
  File "<string>", line 3, in <module>
TypeError: 'int' object has no attribute '__getitem__'


Version-Release number of selected component (if applicable): 
OCP 4.5
ElasticSearch 4.5
Cluster Logging: 4.5


How reproducible:
Very. At least 4 others have seen this issue.


Steps to Reproduce:
1. Install ElasticSearch 4.5 from 4.4 according to the cluster-logging upgrade guide
2. Install ClusterLogging 4.5 from 4.4according to the cluster-logging upgrade guide

Actual results:

Job failures with error messages stated above.

Expected results:

No failures