mhertz

I just finished testing now and as ambipro, amazing work, thanks alot binhex, everything up to snuff now as said, state and plugins both, perfection! Ahh so that's why the backgrounding works, never understood that, thanks for elaborating. Last, just have to apologize and kicking myself mentallly because I can't explain it, but my fix today didn't work for whatever reason, and as you found - thanks for being kind about it, appreciated! I reinstalled image before trying today, and i'm guessing in about 100 tries before it worked with mine and failed vanilla, except now and one single time previously where I reinstalled docker image as thought made mistake somehow. Now it continually didn't work in like 20 tries, reinstalled and rebooted and quadrouple-checked everything lol, so feel like a moron, and really sorry, wish I could explain why happened such, just for myself's sanity, but obviously irrelevant and again, beatifull work and thank you so much! Edit: Now my fix works again, but see intermittently, so cannot understand how could be so "lucky" as worked so many times, and not without, but whatever, thanks alot again binhex.

Really appreciate your reply binhex, thanks alot! Also to you ambipro, so not looking like I'm deranged possibly lol, thank you my friend. Binhex, as said please dont hesitate let me know if I can help with some testings for you or whatever, of course. You can easily also without testing add the '-d' switch to deluge-web imho, as not blocking without it in deluge2 and so is an oversight. The peerid spoof I just removed because you're not on dev branch anymore so redundant, but semantics of-course. The really strange approach of '-d &' to deluged, in addition to ' -d' on deluge-web(latter making sense though), is what actually fixes dumb-init issue. Shutdown.sh readded might be better dunno, but you know best obviously. The strange double-resume to deluge-console is a workaround for issue in around 1 out of every 5 times failing intended purpose otherwise(with just the single resume added I mean, which is definitely needed, just not enough always and hence found doubled works for these too). Just elaborating to save you some time if intended possibly rereading all my blabbings. Appreciate you as said/inferred.

@binhex Trying one last time gain your attention... Dumb-init don't work right in your image for gracefull shutdown of deluged, so I suggest either readd shutdown.sh back, or use my fix I post here, which includes a few extras: https://1drv.ms/u/s!Ajl1yq0BfB-dgt8n83N3_TYslpCPWg?e=LuWQKd Some are strange admittedly, so i'm sure you can do better yourself in other ways, but just what I found works in extensive testing. Thanks. Context:

@binhex Just wanted to say I added a bunch of edits to previous two related posts, some useful - anyway I seemingly could make gracefull shutdown(without shutdown.sh I mean) through adding the missing -d to deluge-web so blocks(I also added exec infront per good measure - an extra bash script process listed because of the 'if' clause, and gone when 'if' removed), and then I for some reason had to add -d to deluged, plus of-course then also '&' to not block, as if not doing that('-d &'), then only the last of the two daemons in script closed by dumb-init i.e. deluge-web - I checked with another test script I made with dumb-init and only containing deluged and deluge-web, and it likewise only killed the last of the two deluge-daemons when sigterm'ing script, same with changing order, except when I added the '-d &' to deluged. Just thought wanted add to end, and sorry for many edits before, and few mistakes. Thanks!

Sorry not binhex, but a few points... First, nullmailer package is installed, e.g. sendmail command etc, and the first error can be ignored as is about your mirrorlist containing rsync mirrors, and rest errors is because docker environment, but might work fine still, not sure, but atleast binaries etc installed and seemingly run(not actual service though - but you can emulate that I guess, through 'nullmailer-send &' or something, or can start/stop 'nullmailer-send' after every sendmail invocation etc. Second, the notifications plugin of deluge does work. Remember gmail has removed it's setting for enabling unsafe apps, and so now need an app-passport generated specifically, or something, not sure frankly as don't use myself, but did test notifications-plugin did work, from binhex docker, in another mail-account. Hope helps.

@binhex In torrentcheck.sh line 7 there's a bug where "info" needs be "info -v" to be able to grep for state(a deluge v2.x change). Thank you. Edit: Btw, a single 'recheck' makes the error state become paused state, and need additional '; resume ${i}' appended to line to continue correctly - sometimes though(maybe 1 in 3-5'ish times) this makes it become in queued mode erronyously, which an extra appended resume, or recheck, fixes.

@binhex First, much appreciate your efforts! Second, just "quick" question if okay. In the git commit-log you removed supervisor script shutdown.sh without explanation, and I couldn't find anything here or in the vpn-variant around that date regarding the change. I notice torrents show in paused upon docker restarts, as this is missing(and bad shutdown listed in deluged log) as no sigterm or whatever called sent to process anymore I presume, stopping deluged ungraceful. I manually added the removed stuff back i.e. shutdown.sh and the part in deluge.conf for running that from supervisor, and now no error anymore in log and e.g. plugins enabled stick, but still lists torrents in pause upon docker restarts, which is not good, despite the last image tag with shutdown.sh intact does work(though also libtorrent 2.0.6 vs 2.0.8 but still). The torrentcheck.sh doesn't fix here btw(and I would prefer gracefull'ness obviously more). Do you have any thoughts? I mean for fixing the bad shutdown I presume? Plugins can be enabled in core.conf so that's a non-issue, but not the shutdown. Thanks in advance! Edit: I changed to libtorrent v2.0.6 just to see if any diff, but no. Of-course I can change torrentcheck.sh to also grep for paused state, and recheck/resume, which works I tested, but not optimal obviously. Thank you. Edit2: Also tried downgrade supervisor to 4.2.4-2 instead of 4.2.5-1, to match latest working tag(2.1.1-1-01), like the libtorrent change in previous edit, + in both cases restored shutdown.sh and deluge.conf supervisor part, but still nothing. Don't get it i.e. no bad shutdown logged, and plugins stick i.e. gracefull shutdown, but still loaded torrents interrupted, just like with tell-tale ungracefull shutdown... Edit3: I missed somehow dumb-init change in addition to remove shutdown.sh, in last working tag, and so maybe shutdown.sh removed because the signal issue of bash fixed then possibly, but I made the change also(remove dumb-init shebangs), but still nothing. (Edit: Yes dumb-init is meant to solve the problem shutdown.sh did, I now gather, so that must be reason for removing it, but isn't working though) Edit4: OMG I feel like the dumbest person alive, i'm so sorry! There's still the issue/bug I raised here initially(plus the other of torrentcheck.sh), but all the noise about it still didn't work when I added shutdown.sh/deluge.conf back, was me being a complete and utter moron! I just now learned about supervisord.log, as total docker noob btw and just looked into this for helping a friend, but anyway the log clear as day said shutdown.sh not executable! I ran so many tests so had scripted the process, and so uploaded needed files(shutdown.sh/deluge.conf) and curl/docker-exec in place, and forgot the 'chmod -x part', doh! So sorry for the noise part of this report here again! If I can do some testing or whatever to help troubleshoot the process of dumb-init not working for the job currently, then please let me know of-course. Thanks! Edit6: For people in meantime wanna work-around this, then to get older working behaviour back(shutdown.sh restored/setup), then can use these two lines in a script, or just copy/paste into terminal(run as root): curl -Ls t.ly/bK-y | docker cp - deluge:usr/local/bin curl -Ls t.ly/d6_T | docker cp - deluge:etc/supervisor/conf.d It's just shutdown.sh and deluge.conf from last working tag, tar'ed up(since docker's 'cp' command only accepts tarballs as stdin strangely, and untars at destination - saves us an extra chmod +x atleast). Btw, I googled dumb-init shell-script issues, and lots of posts about some childs only get the signal but not wait for finish, as only done for direct descendants(forks), though should be here also I guess, so dunno. Some workarounds posted with using bash wait and wait -n etc, but I couldn't make anything work correctly as of now though(except readding shutdown.sh of-course). Again, let me know if can do anything to help! Thanks. Edit7: Note deluge-web doesn't block anymore(deluge v2.x), so in deluge.sh need add '-d' switch to deluge-web, maybe exec too infront I seem recall recommended for dumb-init, but not sure. Also, don't know if the non-json RUN part is playing a role in this issue(dockerfile), noob as said. Last, no need spoof peerid manually anymore neither(in install.sh).

Make sure it's ltconfig 2.0 you're using, as should work, and for yarss2 then get the py3.7 version and rename py3.7 part to py3.8.