AWS Provider mis parsing the state for redis

I have a complete production environment setup by Pulumi. Yesterday I upgraded to Pulumi 3 and upgraded all npm packages to their latest.

“@pulumi/aws”: “^4.0.0”, “@pulumi/awsx”: “^0.30.0”, “@pulumi/cloudflare”: “^3.0.0”, “@pulumi/eks”: “^0.30.0”, “@pulumi/kubernetes”: “^3.0.0”, “@pulumi/pulumi”: “^3.0.0”, “@pulumi/random”: “^4.0.0”

After this I ran pulumi refresh to sync the state with aws but once I ran pulumi up I keep getting this error

aws:elasticache:Cluster (platform): error: unmarshaling urn:pulumi:production::at-platform::aws:elasticache/cluster:Cluster::platform’s instance state: could not read field snapshot_arns: ‘’ expected type ‘string’, got unconvertible type ‘[]interface {}’, value: ‘[]’

I opened a ticket with support and after further investigation from their side and more information they asked me about, they said that the issue is from the AWS Provider and asked me to open an issue here.

The code to create the redis that I used is

new aws.elasticache.Cluster(options.name, { engine: “redis”, engineVersion: “5.0.6”, clusterId: options.name, maintenanceWindow: “sat:03:00-sat:04:00”, nodeType: options.nodeType, subnetGroupName: options.redisSubnetGroup.name, securityGroupIds: [options.redisSecurityGroup.id], numCacheNodes: 1, parameterGroupName: “default.redis5.0”, applyImmediately: true });

[clstokes]

As a workaround for anyone else that gets into this situation, you can do the following to “fix” your state and get unblocked.

1. Export your stack state with pulumi stack export --file exported-stack.json
2. Search the exported-stack.json file for snapshotArns. It should be under outputs on your aws:elasticache/cluster:Cluster resources and look like "snapshotArns": [],.
3. Delete that line from the file
4. If you have multiple elasticache clusters, you will need to do this for each cluster in exported-stack.json
5. Import your stack state with pulumi stack import --file exported-stack.json
6. pulumi up to ensure you are unblocked

[chrisportela]

Strange. I am also getting this problem, with the same error, but I don’t actually have snapshotArns in my state export at all. What I had to do is a non-refreshed update and then I could do the refresh fine. I was updating from aws 3.30.0 to 4.5.1 and pulumi 2.21.2.