Skip to main content
datumctl compute logs reads the output of a workload’s Instances. Pass a workload name, not an Instance name. By default, it merges the newest 200 lines from all of the workload’s Instances over the last 24 hours and prefixes each line with its Instance.
Install the Compute plugin with datumctl plugin install compute. If datumctl compute logs is an unknown command, run datumctl plugin upgrade compute and check datumctl compute logs --help.

​
Read and follow Instance output

Replace WORKLOAD_NAME with the name of your workload:
Use --since to read a longer or shorter window. A time window shows all matching lines unless you also set --tail:
To narrow the output, use the location, Instance, or container name. --instance accepts a full name or the short name shown in line prefixes. --container applies to standard container Instances; unikernel logs aren’t split by container.
--search matches text case-sensitively. For machine-readable output, use -o json, which prints one JSON object per line:
The command can also read logs left by replaced Instances or a deleted workload. To inspect an Instance’s preceding generation, use --instance=INSTANCE_NAME --previous and widen the time window with --since if needed.

​
Read ALB access logs

For a workload published through an ALB, use --alb to read requests to its public URL:
ALB access logs belong to the published URL, so --alb can’t be combined with --instance, --location, --container, or --search. If the workload has no published URL, there are no ALB access logs to read.

​
Open logs in the Cloud Portal

This opens the workload’s Logs tab. Add one --instance=INSTANCE_NAME to open that Instance’s Logs tab. For an overview of Compute metrics and logs, see Observability.
Last modified on October 8, 2026