Version 4.1 #370
Version 4.1 #370
Conversation
Signed-off-by: David Hutchison <dhutchison86+github@googlemail.com>
Signed-off-by: Adam Hill <adam@diginc.us>
Audit error fix
Signed-off-by: Peter Dave Hello <hsu@peterdavehello.org>
Implemented missing support for DNSMASQ_LISTENING environment variable
Implemented missing support for DNSMASQ_LISTENING environment variable Signed-off-by: Adam Hill <adam@diginc.us>
…evelopment Signed-off-by: Adam Hill <adam@diginc.us>
Refactor Dockerfile for smaller and less layer image
…evelopment Signed-off-by: Adam Hill <adam@diginc.us>
Signed-off-by: Adam Hill <adam@diginc.us>
Signed-off-by: Adam Hill <adam@diginc.us>
Signed-off-by: Adam Hill <adam@diginc.us>
Synology sudo audit error fix, disable audit in pam modules
Signed-off-by: Adam Hill <adam@diginc.us>
| @@ -34,7 +29,7 @@ EXPOSE 443 | |||
| ENV S6_LOGGING 0 | |||
| ENV S6_KEEP_ENV 1 | |||
| ENV S6_BEHAVIOUR_IF_STAGE2_FAILS 2 | |||
| ENV FTL_CMD no-daemon | |||
| ENV FTL_CMD debug | |||
There was a problem hiding this comment.
@pi-hole/api-approvers FYI no-daemon wasn't staying in the foreground for me in my hasty update last night, so I changed it to debug as a workaround to get things running (s6-service requires a foregrounded process)
There was a problem hiding this comment.
Thanks, I found the issue: pi-hole/FTL#430
There was a problem hiding this comment.
I just tried to run it as pihole-FTL no-daemon and - as @Mcat12 commented in his PR - observed that Ctrl+C, etc. are not caught anymore, however, pihole-FTL indeed stays in foreground and works as expected. We should continue this here: pi-hole/FTL#432
| @@ -1,22 +1,25 @@ | |||
| #!/bin/bash -ex | |||
| #!/bin/bash -ex / | |||
There was a problem hiding this comment.
wait where did this / come from.......
| echo y | bash -x pihole checkout core $CORE_TAG | ||
| echo y | bash -x pihole checkout web $CORE_TAG | ||
| echo y | bash -x pihole checkout ftl ${CORE_TAG/v/} |
There was a problem hiding this comment.
New way to checkout dev copies of FTL added as I was pre-release testing. Noticed no 'v' was in this tag where as the others weren't so I'm search/replacing it in this bash variable.
| @@ -10,6 +10,10 @@ fi | |||
| # used to start dnsmasq here for gravity to use...now that conflicts port 53 | |||
|
|
|||
| $bashCmd /start.sh | |||
| # Gotta go fast, no time for gravity | |||
There was a problem hiding this comment.
Maybe not required, added before I added --dns 1.1.1.1 which actually solved my gravity issues (known issue that never manifested in tests before)
| if expected_stdout == 'IPv4': | ||
| assert 'IPv6' not in function.stdout | ||
| # weird slow write/sync problem; no sleep == old state of file, sleep == updated/setup state of file | ||
| time.sleep(1) |
There was a problem hiding this comment.
setup_ipv4_ipv6 does what it is supposed to do according to every manual test I perform but automation/docker seems to read the file from the old state unless you wait a small amount of time at which point it shows the result of setup_ipv4_ipv6's execution
| @@ -4,10 +4,10 @@ envlist = py27 | |||
| [testenv] | |||
| whitelist_externals = docker | |||
| deps = -rrequirements.txt | |||
| # auto parallel max b/c race condition with docker fixture (I think?) | |||
| # 2 parallel max b/c race condition with docker fixture (I think?) | |||
There was a problem hiding this comment.
My local > 2 core machine started failing more when I set this to auto. I thought I had a parallel race condition fixed previously but there maybe more or perhaps it's related to the time.sleep() "fix"; slow to sync/write disks in docker's overlay2 FS.
|
Hopefully my commentary helps other understand more of these changes where I didn't leave enough code comments. I pushed the dev image tag with 4.1 which needs a little more testing before going to the prod tag. Besides changing FTL |
Signed-off-by: Adam Hill <adam@diginc.us>
Signed-off-by: Adam Hill <adam@diginc.us>
Signed-off-by: Adam Hill <adam@diginc.us>
…ng all local manifest cache Signed-off-by: Adam Hill <adam@diginc.us>
|
The dev tag was suffering from an issue I had encountered previously with 'development' - the fix was This should be good to go, I tried the new features on my copy and test are passing-ish (albeit flakily, I have a separate plan for that). |
Signed-off-by: Adam Hill <adam@diginc.us>
|
As I feared the arm images still give an error when ran on real arm hardware. The multiarch base images aren't good anymore :( I'll have to dig into this more and report back. What I tried last time is peeking at the history of our current v4.0 image to see if I could pull out the base layer of the old version of multiarch (they don't seem to have any old tags) but the history command just says "missing" instead of giving a hash to tag. |
|
Any chance you guys might be willing to have a chat with the team at linuxserver.io to discuss the build process for this image? I asked if any of them might have any insight on this PR and they are open to discussing it on their discord server. To any who don't know lsio is a team of pretty awesome people who build a lot of pretty popular docker images. |
Signed-off-by: Adam Hill <adam@diginc.us>
|
Found the root of the issue : 10c44c9 I had introduced a template variable to make updating s6 a single change rather than 3 and somehow it got reverted / hardcoded to amd64 again during the #358 work. Travis is being very slow to run but tests are passing very consistently locally now so I think I'll force this through and release. |
|
This pull request has been mentioned on Pi-hole Userspace. There might be relevant details there: https://discourse.pi-hole.net/t/4-1-docker-logging-to-stdout/15633/5 |
Here are code changes for the new version's release. I'm running some more automated tests locally first.
I need to check the multi-architecture version out to see if it's still suffering from a problem that appeared after it's base image changed (see #361)