-
Notifications
You must be signed in to change notification settings - Fork 43
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
Issues with mounts in v2.1.0 #36
Comments
When I navigate via CLI, I can access When I navigate via Plex to add a library, the folder |
The exact same thing happened to me yesterday. A reboot seemed to fix things, will watch out if it happens again. |
Hmmm a few of us have run it on Ubuntu 18 for several months. It's solid as a rock, I wouldn't release an unstable product (at least not willingly). How often do you experience a drop of the mount? Is it consistent with certain actions? If it's truly unstable for you you can always go back to the old Rclone only solution, but I'm hoping this can be sorted :) |
I know, I've been using 18.04.0 since it came out and with Gooby 2.0 it worked like a charm. Yesterday I installed 18.04.1 and Gooby 2.1 to get the new mergerfs up and running. But not so lucky I guess. It seems to drop after each System reset. Then I have to reboot 1 or more times to get the mount online. So yes, it's consistent. I'm not aware of where a log would be store, but if you point into the direction I'll be happy to share it. |
From what you're telling me, it's just MergerFS that is giving you trouble right? As in: /mnt/rclone is still showing your files, but /mnt/google is empty? When that happens, please type
What does it report back to you? (you can escape that with Ctrl-Z) |
I just performed a System Clean and same problem. See the result below. I also noted that when I entered Midnight Commander to visually check the mount, it complains about the permission. Could this be linked? ● mnt-google.mount - MergerFS overlay writable storage Nov 12 19:22:39 Ubuntu-1804-bionic-64-minimal systemd[1]: Mounting MergerFS overlay writable storage... |
Here is what MC is reporting. Again. It only happens after System Clean: plexuser@Ubuntu-1804-bionic-64-minimal:~$ sudo mc |
Hmmm interesting... I researched the message and it dates back to 2012! You said it only happens after a system clean - but not when you reboot right? Or how do you get the mount to come back at all? Can you try this please:
Check if both /mnt/rclone and /mnt/google are empty before you proceed... (they should be) then
First check if your mounts are there (both /mnt/rclone and /mnt/google, there should also be a /mnt/uploads/Downloads present) Then run another system clean - did the problem come back? |
Yes, only after system clean
However, I still get the error with MC as mentioned above after step 2 |
Ok if everything is working properly I wouldn't worry about that particular error.... the real question is of course: will it keep on working?! Please let me know 😸 |
Wait... I were too trigger happy.
A reboot fixed this though. Seems like I should be careful about System clean :-) |
I'll close now and cross my fingers :-) |
LOL well a system clean doesn't do too many exciting things... basically it
It really doesn't need to be run on a VERY regular basis, just when you experience some issues really. That said, it SHOULD work properly and not CAUSE any issues 😆 |
Hmmm, something is not right. I'll reset the server and reinstall ubuntu / gooby and let you know if the issues persist afterwards |
Oof, sorry you're having such a hard time with it! |
@deedeefink I just did a system clean and I ran into the same issue as yourself! I made a change in the script, please update Gooby and see if that sorts it for you - once again apologies it's causing you grief! |
Mind is suddenly behaving weird too... It's not you. I'll look into this tomorrow! Edit: I think I know what it is, it's not quitting the mount properly so after a reboot it won't have the proper permissions... I promise this will be sorted :) |
Pheww. I was going crazy... |
I guess we're not that lucky :-(
Could this still be a permissions problem or what kind of witchcraft is going on? Here are the error codes I get with the services AFTER reboot: gooby-find.service - Gooby Pre-Cacher Nov 13 14:36:20 Ubuntu-1804-bionic-64-minimal sh[1823]: /usr/bin/find: ‘./proc/5004/task/5004/fd’: Permission denied gooby-rclone.service - Mount Google Drive (rclone) Nov 13 14:34:50 Ubuntu-1804-bionic-64-minimal systemd[1]: Starting Mount Google Drive (rclone)... mnt-google.mount - MergerFS overlay writable storage Nov 13 14:24:14 Ubuntu-1804-bionic-64-minimal systemd[1]: Mounting MergerFS overlay writable storage... |
This is really making me think that 18.04.1 has changed some things. I kept having issues like this, I reverted back to 18.04.0 and no issues anymore. It also seems like they changed the repository as well since you can no longer access some common tools like fail2ban from the repository any longer in 18.04.1. |
Could very well be. In this case @TechPerplexed should experience the same issues as I believe we're both using Hetzner who only offers 16.04.5 and 18.04.1 as default Ubuntu images. But I'll consider to downgrade if we cannot get around this |
I'm sure we can get around this, no worries... I'm pretty much sure I know where the problem lies, so all that needs to be done now is implement the correct solution! |
Guys, it's definitely something to do with whatever change happened between v18.04.0 and 18.04.1 - in fact I can reproduce the issue 100% now. That also explains why it ran flawlessly on our own systems for months before going live. The positive news is that I am pretty sure I have a solution... which I'm going to implement in a test branch and will probably ask you to test before going live with it. If you want the technical details: Currently Gooby creates a single service that calls upon Rclone and MergerFS and boots them up one by one in sequence through the script. This worked like a charm in the past, but it's broken on the latest Ubuntu. It's not even intermittent any longer, it just crashes. The solution is to go back to two separate scripts, which is the conventional way of running scripts anyway and seems to work fine. I hope to have more news tomorrow and a place for you to test my theories :) |
Woohoo!! I knew something happened! Lol...what about missing packages in 18.04.1? |
Lol, I have no idea, but frankly it's driving me nuts... I can make it work now on one machine (Hetzner) but not on my regular dedi... but eh I'm not going to give up that easily! I love using MergerFS because it prevents so many bans and other problems. New day tomorrow with some fresh and innovative thoughts - hopefully 😆 |
Oh one more note: I actually tested the reboot while the Crontab that runs on reboot was disabled and the mount came back each and every time (it didn't with the other service scripts). I haven't tested it WITH cron yet... but will do so tomorrow. In the meantime let me know your findings. I'm a little encouraged now that it does come up after a reboot by itself. So if the script breaks that, at least now we FINALLY have a basis to work with! |
3 cleanups and 4 reboots performed and both /mnt/google & /mnt/clone come back each time. |
After 2x system clean and 2x reboot both Rclone and MergerFS mounts are coming back online... Wuhuuuu.... but (sorry).....theres more...... After my first system clean non of the containers worked and I could not access any apps - But mounts were up though :-). I rebooted and checked mounts. All okay. But still Docker were not working. I performed a new system clean but it never completed as it got stuck at "Bringing services back online". It gave me time to notice the following error:
I wanted to double check that I could reproduce the above. When I tired, I encountered no problems with docker_network, but the mounts were empty...arrrggghhhh....I tried to system clean and switching back and forth between the "stable" and test version for the N number of times and:
@nesbcn, @bdschuster and @HitsvilleUK can you please check. I'm loosing track of all the things I tried and need a second opinion |
Interesting @deedeefink |
I tried a system clean again. It completed and the mounts are working, but the docker_network is not responding and I cannot access my apps. Also after a reboot. I then make another system clean and get this error again:
And "Bringing services back online" gets stuck @TechPerplexed / @ALL I'll fly off to the beach now , so I might not be online for testing that much the coming days. |
@deedeefink first of all, have a wonderful vacation (or holiday, depending on where you live 🤣 ) I'm glad the mount problem seems to be resolved... tentatively of course. As for the message "docker_default not found", this is normal and should not have any other effect (it's just to make sure everything is cleaned out properly). It is also normal for the containers not to always show the "Ok" message (that always makes me nervous too). HOWEVER your system should come back of course... unfortunately I can't reproduce it myself. For the record there have been absolutely no changes in the network settings or the rest of the script - I have only been "messing" with the mount settings. I suggest that if the mount is fine (in the new test setting), but if you continue to have problems with the docker_network not responding, that you create a new issue so we can start troubleshooting that separately... If you guys all confirm that the new services setting solves the mount problem I'll push that to stable - and then we can start afresh with new problems (YIKES I hope not...) Today I am going to take a day off too, because three days of non stop troubleshooting sort of made me neglect all my regular activities, heh. Tomorrow I'll be back with renewed vim and vigor! |
Yeah I should have said....in case it's relevant to deedeefink. Enjoy your day off Sophia and have a great holiday deedeefink. |
Ok my friendlies, this is what I'm going to do... since it seems that is new script has solved the mount problems, I'll push it to the "master" repository tomorrow. If you wish you can update Gooby from the menu (option E - A) and go with that (tomorrow that is, not just yet). However, we're not done yet with the test script.... with the help of my good friend kelinger we have made more improvements to the mount check, basically that it won't start any of the containers until it verifies that both the Rclone AND the MergerFS mount are in place and running. After I push this test repository to the live environment I'll update the test place with the improvements he has made - which will then be pushed soon afterwards hopefully after your approval :) But all that is for tomorrow, today was strictly a script free day for me 😴 |
Done! The version you guys tested is now live - and I have updated the test environment with the built in mount checks (to make sure they are really up before proceeding). Hopefully this can go live by the weekend 👍 |
lol...i will install and check everything out shorty and report back...get some rest. |
Hi from the beach 👋 EDIT: This also applies to the Master version (I'll do a clean ubuntu install).
I wouldn't necessarily trust my setup as I previously experienced issues you guys didn't. I will soon make a fresh install and test it again. But would be good to know if it's just me or a general problem though. |
So did a clean install of ubuntu and installed master version of Gooby and I still got this error when I ran a system clean
|
Well at least with the new built in checks it will show you that there is a problem somewhere... rather than ignoring it altogether and proceeding without the mount. Which I realize is only a very small comfort... That said, I have no idea why yours would still be so stubborn! In fact we now reverted almost all the way back to the old tried and proven GooPlex method of first mounting Rclone (absolutely no changes from how it ran before, except it's properly mounted as your user instead of root), then MergerFS (which should not be able to cause Rclone to crash). As for the messages "mountpoint: /mnt/rclone: No such file or directory" etc - that is what is supposed to happen (though obviously not in a never ending loop). The way the script runs now, first it shuts down the containers, then the mergerfs service, then rclone. Those seem to work fine for you (it loops until done). Ok, some troubleshooting actions: Can you please verify that the line Also please verify the permissions are correct (they should be... but...) When the script is stuck in its loop, is it possible for you to open another session and verify that when you type
Something else for you to try: rebooting with and without the cron enabled. To disable the cron, type When you uncomment the line some unexpected things may happen with the rest of your containers (because the script makes sure things happen in the correct order), but I'd like to focus on getting your mount stable first.... so when you do a "vanilla" reboot and or "cron assisted" reboot, do you have your mounts? Oh and have a wonderful time at the beach 🌞 , I'm jealousssssssssssss (looking at the first snow here) ❄️ |
My system is also stuck at this:
this is after updating gooby and system clean. Any suggestion? |
How long does it stay stuck in that loop? Forever or a certain amount of time? Also lots of suggestions in the post above you :) Edit: can you update Gooby and try again please? |
I tried everything in the post, still looping forever:
|
Can you try once more to set the permissions? Problem is I can't reproduce this any longer... testing on two systems and it just seems to work fine for me. So try this: sudo systemctl stop mergerfs Then make sure Gooby is updated and try the cleanup again from the menu... |
Well I didn't try the permissions, I updated Gooby again, rebooted then reinstalled rclone (Beta version this time), then proceeded to system clean. It worked! |
Also, would it be possible to launch rclone mount with |
I tested what you suggested in the above comment #36 (comment)
Here is my env.:
Note that I define the root folder in my rclone settings. Don't think it's relevant, but my TV folder is actual TVshows When i reboot with / without disabled crontab mounts are not working. When I test it, the result is:
|
Well that was too quick... Mounts seem to be working, now Plex somehow doesn't find the files when transcoding ... I'm looking into it atm |
Followed this, updated Gooby, reinstalled rclone and ran system clean and it worked :-) I did not add my media just yet, so can't verify @nesbcn Plex problem |
Absolutely! You can edit your rclonefs.service to your liking - it won't be overwritten unless you uninstall/reinstall :)
Did you restart the containers though? After all we stopped the services without stopping the containers first... just to make sure the permissions were correct. Do you still have the problem after a "proper" system clean?
Oh I hope so.... fingers crossed... toes crossed... eyes crossed........ |
I'm not having any issues so far with the new version. :-) |
Pfew!! Now if deedeefink and others can confirm it's working for them too, we may be able to finally close this nasty chapter 🗡 |
All seems to be working again. Thank you :-) |
Yep. I just had real life cause to use a system clean up. |
after several system cleanups, everything is still working. |
I made a fresh install of Ubuntu 18.04.1 and Gooby 2.1.0 and I'm experiencing stability issues with the Rclone/MergerFS mount. Basically it drops when I do a System Clean and I would have to reboot to get it online again. However, reboot does not always help and it seems overall that the mount is very unstable and not persistent.
Not sure how I can provide more info for debugging, but please let me know .
Rclone setup is:
[Gdrive]
type = drive
scope = drive
root_folder_id = 1hiddenn
token ={"access_token":"hidden","token_type":"Bearer","refresh_token":"1/hidden","expiry":"2018-11-12T12:18:37.767056944+01:00"}
The text was updated successfully, but these errors were encountered: