Quick answer

some-command 2>&1 | tee "$(date +%Y%m%d-%H%M%S).log"

This generates a timestamp when the command starts, uses it in the filename, displays output in the terminal and writes the same combined standard output/error stream to the file.

What does 2>&1 mean?

Shell processes normally use file descriptor 1 for standard output and 2 for standard error. 2>&1 redirects stderr to wherever stdout is currently going. That matters with tee: a normal pipe carries stdout, so merging stderr first lets both normal output and errors reach the terminal and the log file.

some-command 2>&1 | tee run.log

Redirection order matters because the shell processes redirections left-to-right.

Readable timestamp formats

date +%Y%m%d-%H%M%S
# 20260905-143012

date +%Y-%m-%d_%H-%M-%S
# 2026-09-05_14-30-12

Avoid characters such as : when filenames may be moved between operating systems or tools with different filename rules.

Use a variable in a script

#!/usr/bin/env bash

stamp=$(date +%Y%m%d-%H%M%S)
log="diagnostic-${stamp}.log"

some-command 2>&1 | tee "$log"

Quoting the filename variable is a good habit even when the current format contains no spaces.

Append instead of overwrite

some-command 2>&1 | tee -a application.log

Use a timestamped filename when each run should be a separate artifact; use tee -a when multiple runs intentionally belong in one file.

Know what you are capturing

tee receives whatever reaches its standard input. Adding 2>&1 before the pipe merges standard error into standard output so errors are captured too. For long-running or production diagnostics, also consider log rotation, permissions and whether the output may contain secrets.