v1.9.0 returns metrics 2 times slower, than v1.8.0 with the same metrics count

[nudimannui4e]

Example on a huge cluster (300+ topics):

v1.8.0:

for i in $(seq 10); do      curl -s -o /dev/null -w "%{http_code} %{time_total}s\n" http://kafka-xxx:9308/metrics; done
200 3.458325s
200 6.239394s
200 6.085925s
200 6.228413s
200 6.031971s
200 6.315268s
200 6.147002s
200 6.156012s
200 6.680867s
200 6.083318s

v1.9.0:

for i in $(seq 10); do      curl -s -o /dev/null -w "%{http_code} %{time_total}s\n" http://kafka-xxx:9318/metrics; done
200 12.879150s
200 13.050124s
200 12.767673s
200 11.880682s
200 12.581656s
200 12.709735s
200 15.758669s
200 14.664367s
200 14.021954s

Test on Instance:

os: AlmaLinux 9
ram: 16gb
cpu: 12
-Xmx6G -Xms6G

[nudimannui4e]

update

[senax]

On v1.4.2:
# for I in {1..5}; do curl -s -o /dev/null -w "%{time_total}s\n" http://localhost:9308/metrics; sleep 10; done 16.305099s 16.310399s 18.340185s 16.318807s 16.335976s

On v1.8.0:
# for I in {1..5}; do curl -s -o /dev/null -w "%{time_total}s\n" http://localhost:9308/metrics; sleep 10; done 16.351345s 16.781624s 16.473706s 16.509356s 17.355690s

On v1.9.0 "Starting kafka_exporter (version=1.9.0, branch=HEAD, revision=8ec24078707d4d3b349e71522ebd7f179746072d)"
# for I in {1..5}; do curl -s -o /dev/null -w "%{time_total}s\n" http://localhost:9308/metrics; sleep 10; done 23.115361s 26.795423s 26.590782s 24.386172s 22.977978s

On latest "Starting kafka_exporter (version=1.9.0, branch=master, revision=953e2f1745b90101f331ad629f10be5239845f69)"
# for I in {1..5}; do curl -s -o /dev/null -w "%{time_total}s\n" http://localhost:9308/metrics; sleep 10; done 24.555345s 24.380261s 24.878367s 24.394750s 23.339914s