Symptoms

The service is removed from PFxM or RG in PFMP, and the server is recabled for a different deployment, which may fail with the following UI error: The port channel for interfaces NIC.Integrated.1-2-1 NIC.Slot.2-2-1 cannot be created because the interfaces are connected to the same switch % {switch_certname}. exception.log #/opt/asm-deployer/lib/asm/provider/switch/base.rb:462:in `validate_connectivity' /opt/asm-deployer/lib/asm/provider/switch/nexus5k.rb:245:in `configure_server_interface' /opt/asm-deployer/lib/asm/provider/switch/nexus5k.rb:216:in `block in provision_server_networking' org/jruby/RubyArray.java:1735:in `each' /opt/asm-deployer/lib/asm/provider/switch/nexus5k.rb:215:in `provision_server_networking' /opt/asm-deployer/lib/asm/provider/switch/nexus5k.rb:153:in `configure_server' /opt/asm-deployer/lib/asm/type/base.rb:413:in `delegate' /opt/asm-deployer/lib/asm/type/switch.rb:173:in `configure_server' /opt/asm-deployer/lib/asm/type/server.rb:3030:in `block in configure_networking!' org/jruby/RubyArray.java:1735:in `each' /opt/asm-deployer/lib/asm/type/server.rb:3023:in `configure_networking!' /opt/asm-deployer/lib/asm/type/switch.rb:90:in `block in configure_server_networking!' org/jruby/RubyArray.java:1735:in `each' /opt/asm-deployer/lib/asm/type/switch.rb:89:in `configure_server_networking!' /opt/asm-deployer/lib/asm/service/switch_collection.rb:337:in `block in configure_server_switches!' org/jruby/RubyArray.java:1735:in `each' /opt/asm-deployer/lib/asm/service/switch_collection.rb:329:in `configure_server_switches!' /opt/asm-deployer/lib/asm/service/switch_collection.rb:429:in `configure_server_networking!' /opt/asm-deployer/lib/asm/service_deployment.rb:4962:in `process_switches_via_types' /opt/asm-deployer/lib/asm/service_deployment.rb:537:in `process' /opt/asm-deployer/lib/asm.rb:220:in `block in process_deployment' Impact Service or RG deployment failure.

Cause

Removing a Service or RG does not clear cached network_topology data.Cached data might be used on the next Service or RG deployment, causing the deployment to fail.

Resolution

Confirm that the problematic server is cabled in an AB-BA manner. If the cabling is validated but shows incorrect data in the UI, clear the cache to force a recalculation during the next Service or RG deployment. Follow the commands below, depending on the version, to clear out the cached entries. The resource name format is {rackserver-servicetag} For PFxM 3.x For PFxM version 3.6.1 and lower, follow the process below: 1. Run the following command: env RUBYLIB=/opt/asm-deployer/lib pry 2. From the shell that opens, run the following Ruby commands: Replace the servicetag with the service tag of the node in all lower-case. require "asm" puppetdb = ASM::Client::Puppetdb.new(:logger => Logger.new("/dev/stdout")); nil puppetdb.merge_facts_blocking!("rackserver-servicetag".downcase, "network_topology" => "{}"); nil 3. Rerun inventory on the server The next deployment using that server will recalculate the switch connectivity from scratch. For PFxM version 3.6.2 and higher, follow the process below: 1. Run the following command to log in to the psql shell: psql -U orion asm_dev 2. The following command lists all nodes and finds the entry that matches the host needed to clear the cache for: asm_dev=> SELECT certname FROM factsets WHERE certname LIKE 'rackserver%'; For example: asm_dev=> SELECT certname FROM factsets WHERE certname LIKE 'rackserver%'; certname ------------------- rackserver-1a1b123 rackserver-1a1b124 rackserver-1a1b125 rackserver-1a1b126 rackserver-1a1b127 (5 rows) 3. Modify the example command below and replace rackserver-xxxxxx with the value that matches the correct node ID: asm_dev=> DELETE FROM factsets WHERE certname = LOWER('rackserver-xxxxxx'); 4. From the psql shell, run the following command to save the transaction into the database: asm_dev=> COMMIT; 5. Exit the psql shell: asm_dev=> \q 6. Validate there are no running jobs in PFxM and restart the asmmanager process: systemctl restart asmmanager 7. Run inventory on the server from the PFxM "Resources" tab in the UI. The next deployment using that server will recalculate the switch connectivity from scratch. For PFMP 4.x For PFMP version 4.x and higher, follow the process below: 1. Run the following command from one of the Kubernetes (K8s) management nodes, also known as 'MVM' (Management Virtual Machine), to log in to the psql shell connected to the asm_dev database: kubectl exec -it -n powerflex $(kubectl get pods -n powerflex -l='postgres-operator.crunchydata.com/role=master' | grep Running | cut -d' ' -f1) -- psql -U postgres -d asm_dev 2. The following command lists all nodes and finds the entry that matches the host needed to clear the cache for: SELECT certname FROM factsets WHERE certname LIKE 'rackserver%'; For example: asm_dev=> SELECT certname FROM factsets WHERE certname LIKE 'rackserver%'; certname ------------------- rackserver-1a1b123 rackserver-1a1b124 rackserver-1a1b125 rackserver-1a1b126 rackserver-1a1b127 (5 rows) 3. Modify the example command below and replace rackserver-xxxxxx with the value that matches the correct node ID: asm_dev=> DELETE FROM factsets WHERE certname = LOWER('rackserver-xxxxxx'); For example: asm_dev=# DELETE FROM factsets WHERE certname = LOWER('rackserver-1a1b123'); DELETE 1 4. Validate the entry is cleared: asm_dev=# SELECT certname FROM factsets WHERE certname LIKE 'rackserver-xxxxxx'; (0 rows) 5. Exit the psql shell: asm_dev=> \q 6. Restart all pods in the PowerFlex namespace: kubectl rollout restart deployment -n powerflex 7. Run inventory on the server from the PFMP "Resources" tab in the UI. Validate that the switch ports correctly reflect the topology by clicking "View Details" on the resource. The next RG deployment using the server will recalculate the switch connectivity from scratch. Impacted Versions PFxM 3.x PFMP 4.x

Support Cases