grep1-4  Exercise 1: ERROR with One Line Before and Two After
=============================================================

(Run from an empty folder. The first block makes the two practice files from the chapter; your own files can skip it.)

cat > app.log <<'EOF'
09:00:01 INFO  service starting
09:00:02 INFO  loading config
09:00:03 WARN  config key 'timeout' missing, using default
09:00:04 INFO  connecting to database
09:00:09 ERROR database connection refused
09:00:10 INFO  retrying in 5s
09:00:15 ERROR database connection refused
09:00:16 ERROR giving up after 2 attempts
09:00:17 INFO  switching to read-only mode
09:00:20 INFO  listening on port 8080
09:05:00 INFO  health check ok
09:10:00 WARN  slow response: 2300 ms
09:10:30 INFO  health check ok
09:15:00 ERROR disk usage 95%, ERROR threshold 90%
09:15:01 INFO  alert sent
09:20:00 INFO  health check ok
EOF

cat > access.log <<'EOF'
10.0.0.5 GET /index.html 200
10.0.0.7 GET /about.html 200
10.0.0.5 POST /login 302
10.0.0.9 GET /missing 404
10.0.0.5 GET /index.html 200
10.0.0.7 GET /index.html 200
10.0.0.5 GET /dashboard 200
10.0.0.9 GET /missing 404
10.0.0.7 GET /logout 302
EOF

# --- the exercise ---
grep -n -B1 -A2 --group-separator='-----' ERROR app.log

Output:

4-09:00:04 INFO  connecting to database
5:09:00:09 ERROR database connection refused
6-09:00:10 INFO  retrying in 5s
7:09:00:15 ERROR database connection refused
8:09:00:16 ERROR giving up after 2 attempts
9-09:00:17 INFO  switching to read-only mode
10-09:00:20 INFO  listening on port 8080
-----
13-09:10:30 INFO  health check ok
14:09:15:00 ERROR disk usage 95%, ERROR threshold 90%
15-09:15:01 INFO  alert sent
16-09:20:00 INFO  health check ok

WHY THIS WORKS AS AN ANSWER
---------------------------
-B1 and -A2 add the context, -n numbers the lines, --group-separator replaces the default -- line.
A colon after the number marks a matching line (5, 7, 8 and 14); a dash marks a context line. The ERROR on line 5 pulls in line 4 before it and lines 6 and 7 after; but line 7 is itself a match, and so is 8, so their own context overlaps and everything from line 4 to line 10 becomes ONE group. The ERROR on line 14 is far enough away to start a second group (13 to 16). That is why there are two groups rather than four: one group per run of overlapping context, not one per match.
Line 14 is printed once although it holds ERROR twice, because grep works in lines.