An increasing memory footprint is perfectly normal as FTL holds all queries in memory to generate the statistics from. However, it is very memory efficient and will drop those queries that are older than 24 hours. After a full 24 hours of normal browsing activity, the memory footprint should stay stable and does not increase notably - if at all - afterwards.
We do this to use the least amount of memory as possible. The alternative would be to block a huge chunk of memory from the beginning which seems to be a much unfriendlier behavior.
There is no known bug with memory leaks but it might be related to you running the no-logging setup which is (probably) not used very often so it might very well be that you spotted a so far unknown bug. I will try to reproduce this over the next few days.
Okay, good to know, so I don't have to look for a bug isolated to the no-logging mode. The memory usage of pihole-FTL is fairly linearly correlated with the number of queries shown on the dashboard.
Do you observe this number growing over time as well?
Level 4 - disabled statistics (v4.1+)¶
Disables all statistics processing. Even the query counters will not be available. Additionally, you can disable logging to the file /var/log/pihole.log using sudo pihole logging off.
Note that - due to the disabled query processing - regex blocking is not available on level 4.
No. I use a few Zero W's with logging on. They both resolve DNS quickly.
You can disable /var/log/pihole.log with no ill effects on your dashboard (that data will come from the long term database). The downside is that you won't have a live log to tail, so if you want that you would need to enable the /var/log/pihole.log.
Can you run htop every few hours and see which process has the increasing memory load? Just for comparison, htop for a Zero W running the master branch, the default blocklists, NULL blocking is shown below: