PiHole on docker won't start anymore with the latest version #794
Comments
|
Unfortunately it happens here too. Definately something is wrong with latest release. Running on Synology Docker in DSM 6.2 |
|
Can confirm for amd64 and arm64 platforms. Same loop of |
|
"Me too" "Same Here" and 👍🏻 comments don't help at all so please just use the icons for the original post. |
|
Can confirm. I researched how to load an old image in Docker and couldn't find any help. As @jgeusebroek mentioned, you could use Sorry if this isn't helpful for the devs |
|
Can confirm, same here. |
|
pihole/pihole:v5.6 is the previous build. FYI. |
|
|
Could this be caused by pi-hole/FTL@49ba60e? |
|
This workaround worked for me (assuming your docker container name is |
|
@PromoFaux The workaround by @gergnz suggests that |
|
I believe that as of that commit: pi-hole/FTL@49ba60e Also we're using docker-pi-hole/s6/debian-root/etc/cont-init.d/20-start.sh Lines 23 to 24 in 75328c6 TL;DR we never did "clean" shutdowns. But since FLT v5.7 it's become necessary to do so. |
|
@DL6ER I've just checked that the image doesn't have |
|
Thanks. I'm not at all in how this container works, so I was just assuming what may be going wrong. Thanks for the research! Whatever the solution will be, |
|
To confirm, the image doesn't contain lockfiles. docker run --rm -ti --entrypoint="" pihole/pihole:v5.7 ls -la /dev/shmBut if we look at this dir at the end of the init script it's apparently unclean. # Dockerfile
FROM pihole/pihole:v5.7
RUN echo "ls -la /dev/shm" >> /etc/cont-init.d/20-start.shdocker build -t pihole-locking .
docker run --rm -ti pihole-locking exit 0 |
|
I should add: Cleaning up behind you is somewhat good practive and FTL tries to open the shared memory objects and only fails with This is actually a good thing, just the logic in the docker container has to be improved. I'm not familiar at all, but maybe the logic can be changed from
To only start when we actually want it. |
|
OK, what appears to work (as a quickfix here) is adding |
This works for me |
|
I've been trying out a non-kill-9 approach, but it appears to be flaky. # Kill dnsmasq because s6 won't like it if it's running when s6 services start
FTLPID=$(pgrep pihole-FTL)
kill ${FTLPID} || true
while kill -0 ${FTLPID}; do
echo "FTL ${FTLPID} still running..."
sleep 1
done
while [ -e "/dev/shm/FTL-lock" ]; do
echo "Lock file still exists..."
sleep 1
done
pihole -vUsually this works as expected. But occasionally it will not clean up. So possibly a normal terminate signal still leaks the files? |
|
Also adding Edit, in fact that's more or less what happens on a bare metal instance https://github.com/pi-hole/pi-hole/blob/master/advanced/Templates/pihole-FTL.service#L25-L41 |
Shouldn't https://github.com/pi-hole/FTL/blob/2999e2b57c62b4455187ee9b77840d49df0a8e2e/src/main.c#L127 And
|
|
This issue has been mentioned on Pi-hole Userspace. There might be relevant details there: https://discourse.pi-hole.net/t/docker-update-to-v5-7-causes-ftl-to-crash-at-launch/44464/3 |
Nope @DL6ER I'm pretty sure it does, but seems unrelated to the Full log |
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
|
I can confirm that #797 fixes this issue. |
|
Thanks @thedanbob, I've just pulled Everyone here should now be able to re-pull |
|
My watchtower just updated PiHole to the latest version and I can confirm this issue is now resolved 😄 |
|
This issue has been mentioned on Pi-hole Userspace. There might be relevant details there: https://discourse.pi-hole.net/t/docker-update-to-v5-7-causes-ftl-to-crash-at-launch/44464/4 |
I can confirm that, manual update of one (the problematic one) worked, automatic update with WatchTower of the other one worked as well (after resuming WatchTower from pause). |
|
This issue has been mentioned on Pi-hole Userspace. There might be relevant details there: https://discourse.pi-hole.net/t/pi-hole-wont-start-after-docker-update/44454/8 |
|
Thanks for fixing the problem. Works now in my Docker with latest version! |
|
Hi Guys, I have the same issue with the lastes release. Can't find how to solve it. Pleas help. Lg |
|
I am new to pihole and was installing it on an odroid with docker-compose, I also had this issue with pihole-FTL [services.d] done. My problem was that I had enabled the log volume So if anyone made the same mistake, then start the with docker-compose and it works fine :) I also moved all mounts to '/opt/pihole'... |
I've actually removed this from the example file - because it's probably an unnecessary mount, and people often miss the part to create it first |
Versions
My watchtower just automatically updated my PiHole running on a raspberry pi:
However after the PiHole containers shows up as Unhealthy. Here's the container log:
Platform
Expected behavior
PiHole should start correctly
Actual behavior / bug
PiHole FTL service won't start
Steps to reproduce
Steps to reproduce the behavior:
Edit
I've also checked the FTL log found by executing
cat /var/log/pihole-FTL.loginside the docker container. This is the output:The text was updated successfully, but these errors were encountered: