New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
On boot long FTL startup and web graphs blank until FTL restarted #4459
Comments
There is a GPS source feeding chrony that is kept hot and I have experimented with a service (https://selivan.github.io/2020/12/23/ntp-sync-time-before-starting-any-services.html) that attempts to force an NTP sync before other services, but the behavior is the same. |
You have some oddities in your debug log. First, the contents of your Pi-hole configuration file. Many of these lines should not be in this file - they are in file /etc/pihole/setupVars.conf.
Your query database is showing no traffic in the past 24 hours, which is inconsistent with your privacy and database settings.
Your gravity database has problems, likely related to your third party script:
|
Here is a debug log after pihole restartdns is run. https://tricorder.pi-hole.net/aF6NNPUQ/ It looks like the pihole-FTL.conf not only has a lot of overlap with setupVars.conf, but it also has a lot of out-of-date settings. I've culled pihole-FTL.conf to only have AAAA_QUERY_ANALYSIS and PRIVACYLEVEL, as those are not in setupVars.conf. The third-party script is pihole-updatelists. https://github.com/jacklul/pihole-updatelists The time warning is the issue where the system time at boot is 2/14/2019 until chrony initializes. The writing collision is likely from pihole-updatelists also trying to update gravity. I now have turned off that behavior so it only updates the lists. I'd prefer pihole to manage the gravity database. The behavior is the same after these changes. Debug log after these changes and a reboot: https://tricorder.pi-hole.net/hlWBhXm9/ The gravity database collisions are still there so I uninstalled pihole-updatelists. The behavior is the same but it appears that the collision has stopped. Debug log after this change and a reboot: https://tricorder.pi-hole.net/jHwMNpC3/ Update: Running a pihole repair and reboot after all of this did not change the behavior. |
I did a fresh uinstall / reinstall of pihole, replaced the pihole-FTL.db with the old one, restored settings via teleporter, and still encountered the same issue. I am still using unbound, pihole-updatelists, and https. https://tricorder.pi-hole.net/sXz3kyxB/ When I start a fresh pihole-FTL.db the behavior goes away. My best guess is that the large database takes too long to initialize and some things time out. https://tricorder.pi-hole.net/c24EJbYp/ For reference, the pihole-FTL.db I've been building up for the past few years is 4.2 GB. Edit: My mistake, even with a fresh reinstall of pihole and no pihole-updatelists or https the behavior is still present on reboot until pihole-FTL is restarted. With a fresh db it is faster to get to the point where a pihole-FTL restart fixes the issue. |
This was solved by increasing the PHP memory limit in /etc/php/7.3/cgi/php.ini. |
This issue has been mentioned on Pi-hole Userspace. There might be relevant details there: https://discourse.pi-hole.net/t/found-database-entries-in-the-future/55333/2 |
Versions
Platform
Expected behavior
On boot FTL should start up within a minute and the web interface should show log graphs.
Actual behavior / bug
FTL takes a few minutes to start and afterwards the web interface does not populate the bar graphs until pihole restartdns is run. pihole repair does not fix the issue.
Steps to reproduce
Reboot. This is a build from about 18 months ago and it's had this behavior for most of its life. Maybe it has something to do with NTP taking some time to warm up and the default timestamp being too far in the past. With the use of unbound and unbound-resolvconf I think there may be a bootstrapping issue where the NTP time can't be pulled until the DNS resolver is working and the DNS resolver might not work if the system time isn't correct. There is also a local GPS/PPS module that chrony uses, but that might not be read in quickly enough.
Debug Token
Screenshots
The text was updated successfully, but these errors were encountered: