4.19.01 Edge for Private Cloud release notes

Metadata file format for apigee-sso install

When installing apigee-sso and using a metadata.xml file instead of a URL, if the metadata.xml file does not include a new line (CRLF) at the end of the file, the metadata.xml file is not read entirely.

Workaround:

To work around this issue, you can do one of the following:

• Add a CRLF to the end of the metadata.xml file

  OR

• Use a URL-based location for the metadata

apigee-monit not supported on Amazon Linux 1

Workaround:

Message Processor backup not backing up the correct set of files

Workaround:

Run the backup a second time and it should back up the correct set of files.

Session tokens not invalidated

Session tokens are not invalidated on the server when the user clicks on the Apigee Edge Logout button.

Workaround:

Cassandra operations not working when JMX auth is enabled

Workaround:

HEAD requests to Node.js targets hang

HEAD requests to a Node.js target can hang, leaving connections pending.

Workaround:

To work around this issue, define a handler for HEAD requests to explicitly return an empty response.

Hostnames not resolving

After installing or upgrading Edge for Private Cloud, hostnames might not resolve to their addresses.

Workaround:

To resolve this issue, restart the Edge UI component:

/opt/apigee/apigee-service/bin/apigee-service edge-ui restart

DataAccessException in multiple data center configurations

In multiple data center configurations, if one datastore becomes unavailable, then you might see the following error:

DataAccessException: Error while accessing datastore;
Please retry later

The result is that Management Server may not start because it is trying to connect to Cassandra nodes in both dc-1 and dc-2. The DataAccessExceptions occurs if a Cassandra node is down. This might also result in API traffic disruption, where Message Processors report DataAccessExceptions while trying to retrieve KVMs.

Note that the expected state is for the Management Server not to connect to datastore components across regions.

Workaround

The workaround is to deregister the following Cassandra node types in the unavailable data center and then re-register them after the Cassandra nodes are available again:

• kms-datastore
• dc-datastore
• keyvaluemap-datastore

To deregister and reregister these Cassandra node types:

1. Get the UUIDs of the Cassandra nodes by using the following curl command:

   curl -u ADMIN_EMAIL:ADMIN_PW \
     "http://MS_IP:MS_PORT/v1/servers?region=REGION&pod=GATEWAY_POD \
     &type=CASSANDRA_NODE_TYPE"

   Where:

   • ADMIN_EMAIL and ADMIN_PW are the credentials of your Apigee account.
   • MS_IP and MS_PORT are the Management Server's IP address and port number.
   • REGION is the name of the data center in which the Management Server is located.
   • GATEWAY_POD is the pod name, which is by default "gateway". You might have renamed it to something else, though, so check your implementation.
   • CASSANDRA_NODE_TYPE is one of kms-datastore, dc-datastore, and keyvaluemap-datastore.

   For example:

   curl -u nickdanger@google.com:myP@$$w0rD
     "http://192.168.0.1:8080/v1/servers?region=dc-1&pod=gateway&type=dc-datastore"

   The response uses the following format:

   {
     "internalIP" : "POD_IP_ADDRESS",
     "isUp" : [true|false],
     "pod" : "GATEWAY_POD",
     "reachable" : [true|false],
     "region" : "dc-1",
     "tags" : {
       "property" : [ ]
     },
     "type" : [ "kms-datastore", "dc-datastore", "keyvaluemap-datastore" ],
       "uUID" : "POD_UUID"
   }

   For example:

   {
     "internalIP" : "192.168.1.11",
     "isUp" : false,
     "pod" : "gateway",
     "reachable" : false,
     "region" : "dc-1",
     "tags" : {
       "property" : [ ]
     },
     "type" : "dc-datastore",
     "uUID" : "13cee956-d3a7-4577-8f0f-1694564179e4"
   }

   Note the values of the uUID field in the response. You will use these to deregister the nodes.

2. Repeat step one for each Cassandra node type: kms-datastore, dc-datastore, and keyvaluemap-datastore. Be sure to take note of the UUIDs that are returned.
3. Deregister the nodes using the following command:

   curl -u ADMIN_EMAIL:ADMIN_PW "http://MS_IP:MS_PORT/v1/servers/UUID" -X DELETE

   Where UUID is the UUID returned in the previous command's response.

4. Repeat step 3 for each UUID you collected in steps 1 and 2.
5. Re-register the nodes using the following command:

   curl -u ADMIN_EMAIL:ADMIN_PW "http://MS_IP:MS_PORT/v1/servers -d \
     "Type=kms-datastore&Type=dc-datastore&Type=keyvaluemap-datastore& \
     Type=counter-datastore&Type=cache-datastore&InternalIP=POD_IP_ADDRESS& \
     region=REGION&pod=GATEWAY_POD" -H \
     'content-type: application/x-www-form-urlencoded' -X POST

Note that these operations register and deregister nodes from Zookeeper and do not have any impact on the Cassandra cluster. For more information about these commands, see Update datastore registrations.

MessageLogging policy including extra information in the log message

The FormatMessage element of the MessageLogging policy controls the format of the logged message. When FormatMessage=false, the logged message is not supposed to include any Apigee-generated information. However, even if you set FormatMessage=false, the log message still includes the following information:

• The priority score
• The timestamp

Workaround:

Resetting system admin password fails

Resetting the system admin password for the Management Server that contain special characters will fail.

Workaround:

Define the password in a data file that you then pass to the /users endpoint in the Management API. For more information, see Resetting system admin password.