Hello All. We've been trying to deploy a POC on an EC2 and are encountering some issues when we try to run

lakefs setup --user-name root --access-key-id EXAMPLE_KEY --secret-access-key EXAMPLE_SECRET

the error message is:

ERROR  [2021-07-27T18:44:43Z]lakeFS/pkg/db/migration.go:44 pkg/db.(*DatabaseMigrator).Migrate Failed to migrate                             direction=up error="dial tcp 127.0.0.1:5432: connect: connection refused"`

This is surprising as some of the setup appears to successfully connect to an RDS:

INFO   [2021-07-27T18:44:30Z]lakeFS/pkg/db/connect.go:57 pkg/db.ConnectDBPool connecting to the DB                          conn_max_lifetime=5m0s driver=pgx max_idle_conns=25 max_open_conns=25 uri="<postgres://postgres>:paasword@<prefixes>.<http://us-east-1.rds.amazonaws.com/postgres|us-east-1.rds.amazonaws.com/postgres>"
INFO   [2021-07-27T18:44:30Z]lakeFS/pkg/db/connect.go:77 pkg/db.ConnectDBPool initialized DB connection                     conn_max_lifetime=5m0s driver=pgx max_idle_conns=25 max_open_conns=25 uri="<postgres://postgres>:paasword@<prefixes>.<http://us-east-1.rds.amazonaws.com/postgres|us-east-1.rds.amazonaws.com/postgres>"
DEBUG  [2021-07-27T18:44:30Z]lakeFS/cmd/lakefs/cmd/run.go:61 cmd/lakefs/cmd.glob..func9 No migration, setup required
...
...
INFO   [2021-07-27T18:44:30Z]lakeFS/pkg/auth/service.go:207 pkg/auth.NewDBAuthService initialized Auth service
DEBUG  [2021-07-27T18:44:30Z]lakeFS/pkg/stats/metadata.go:34 pkg/stats.NewMetadata failed to collect account metadata            error="ERROR: relation \"auth_installation_metadata\" does not exist (SQLSTATE 42P01)"
DEBUG  [2021-07-27T18:44:30Z]lakeFS/pkg/stats/collector.go:213 pkg/stats.(*BufferedCollector).CollectMetadata could not update metadata                     error="installation ID is missing: stats: send error" service=stats_collector
DEBUG  [2021-07-27T18:44:30Z]lakeFS/cmd/lakefs/cmd/run.go:211 cmd/lakefs/cmd.checkForeignRepos lakeFS isn't initialized, skipping mismatched adapter checks

Any help or guidance would be appreciated. We haven't connected the EC2 to an elastic IP, so the S3_Gateway may be faulty, but that doesn't seem to be the issue

Hey Jesse, sorry that you’re experiencing issues. Are you running the postgres instance on that EC2 server as well? Can you share the lakeFS

config.yaml

file that you’re using?

the postgres instance is on an RDS.

logging:
  format: text
  level: DEBUG
  output: "-"
database:
  connection_string: "<postgres://postgres>:password@<prefixes>.<http://us-east-1.rds.amazonaws.com/postgres|us-east-1.rds.amazonaws.com/postgres>"
auth:
  encrypt:
    secret_key: "EXAMPLE_SECRET"
blockstore:
  type: s3
  s3:
    region: us-east-1
    profile: default
    credentials_file: /home/ubuntu/.aws/credentials
gateways:
  s3:
     # replace this with the host you will use for the lakeFS S3-compatible endpoint:
     domain_name: s3.lakefs.<company>.io

For some reason the migration doesn’t read the database connection string. I’m opening an issue for that. As a workaround, can you try using

lakefs run

with the same config file, and do the setup through the UI or REST api?

the password has a

!

in it. could that contribute to an issue?

Also, where is the config file located? If it’s not in the default path, can you try adding the flag

--config=path/to/config.yaml

?

path would be

/home/ubuntu/config.yml

default is

$HOME/.lakefs.yaml

will change and note the outcome, then see what happens with the

lakefs run

option

great, let me know if it didn’t work

moving the config file solved the setup issue

Glad that it was solved. Debug logs seem accurate as some actions depend on lakeFS being initialised first.

👍 I would also rename this file in this guide: https://docs.lakefs.io/deploy/aws.html#on-ec2

Yap, it's confusing. We'll modify it, thanks for letting us know