Removed it from my lists for all tests.
I've typed this during the pihole -g
wait periods, so forgive me, if some comments are no longer necessary or valid after this AMAZING result.
- new systems specs: raspbian Release date: 2019-09-26 with sudo apt-get update && sudo apt-get -y upgrade, unbound 1.9.6 + redis, pihole 4.2.3
- running all test on the production pi, so no different results, due to different network location.
pihole 4.3.2 (pihole -g) approximately 2.5 minutes
start: Thu 23 Jan 09:40:40 CET 2020
finished: Thu 23 Jan 09:42:03 CET 2020
Number of unique domains trapped in the Event Horizon: 2.150.519
new SD card (etcher), pihole with v4.3.2 (fully configured with existing whitelist, regex and adlists.list - 75 entries)
echo "release/v5.0" | sudo tee /etc/pihole/ftlbranch
pihole checkout core release/v5.0
pihole checkout web release/v5.0
pihole beta5 (pihole -g / database v9) approximately 1 hour 46 minutes
start: Thu 23 Jan 11:10:41 CET 2020
time before the first list starts processing (Flushing gravity table): Thu 23 Jan 11:19:44 CET 2020 (9 minutes)
finished: Thu 23 Jan 12:56:16 CET 2020
Number of gravity domains: 3303054 (2.154.469 unique domains)
database size (gravity.db): 214.760KB
applied "Add gravity database 9->10 update script #3089"
pihole beta5 (pihole -g / database v10) approximately 1 hour 50 minutes
start: Thu 23 Jan 13:00:37 CET 2020
time before the first list starts processing (Flushing gravity table): Thu 23 Jan 13:10:56 CET 2020 (10 minutes)
finished: Thu 23 Jan 14:50:59 CET 2020
Number of gravity domains: 3303342 (2.154.728 unique domains)
database size (gravity.db): 214.868KB
new SD card (etcher), pihole with v4.3.2 (fully configured with existing whitelist, regex and adlists.list - 75 entries)
echo "release/v5.0" | sudo tee /etc/pihole/ftlbranch
pihole checkout core tweak/gravity_performance
pihole checkout web release/v5.0
pihole tweak/gravity_performance (pihole -g / database v9) approximately 2 minutes
start: Thu 23 Jan 15:04:22 CET 2020
finished: Thu 23 Jan 15:06:20 CET 2020
Number of gravity domains: 530379 (530.379 unique domains)
database size (gravity.db): 72.176KB
applied "Add gravity database 9->10 update script #3089"
pihole tweak/gravity_performance (pihole -g / database v10) approximately 2.1 minutes
start: Thu 23 Jan 15:09:02 CET 2020
finished: Thu 23 Jan 15:11:12 CET 2020
Number of gravity domains: 530379 (530.379 unique domains)
database size (gravity.db): 72.176KB
@DL6ER: THIS IS AMAZING...
however, there appears to be something wrong, look at the number of unique domains (difference between beta5 and gravity_performance - a small difference can be explained by changing lists, gravity_performance produces only 25% of the beta5 result).
never the less, GREAT IMPROVEMENT, find the last bug, and we are one step closer to an ideal world.
I checked the firewall logs, no abnormal messages (blocks or drops) from pfsense firewall or SURICATA.
Also checked the pihole log, no abnormal entries found.
Checked the system logs, no abnormal warnings or errors found.
I noticed during list processing, a file /etc/pihole/gravity.db-journal is created, size changes during processing. This file can be quite large, example 88.885KB
Flushing the database in beta5 takes a very long time, my earlier suggestion to build gravity in a temp table eliminates the need to flush it...
again a lot to ask but anyway, while your at it...
could you change:
[i] Target: https://hostfiles.frogeye.fr/multiparty-trackers-hosts.txt
[✓] Status: Retrieval successful
[✓] Adding adlist with ID 11 to database table
into
[i] time - Target: https://hostfiles.frogeye.fr/multiparty-trackers-hosts.txt
[✓] time - Status: Retrieval successful
[✓] time - Adding adlist with ID 11 to database table
[✓] time - Done adding adlist with ID 11 to database table
possible other suggestion, to increase performance:
- download all the lists, prior to processing
- sort the downloads by file size, smallest to biggest and store the result in a TEMP database
- process the lists from the TEMP database, this will ensure the database contains lesser entries during the processing of the first lists (the end result will be the same), as opposed to accidently starting with the biggest list -> impact on duplicate check
- processing done -> remove the TEMP database