[SOLVED] Install recoverable?

[RAMAN]

Hello, new to Unraid and the forum. I decided to try Unraid after my second Storage Spaces DR failure. Everything seemed to be going well with the installation. After configuring plugins and dockers, I decided to switch from a cache drive using RAID0 to a cache drive pool with RAID 1. I had Brand X SSD in Cache slot 1 and two Brand Y SSDs in Cache 2 and 3. Everything seemed fine, so I added the fourth Brand Y SSD by increasing the slots to 4 and assigned the fourth SSD. My intention was to switch to RAID10. Not sure if it was before or after that, one of my data drives began reporting write errors, then one of my parity drives(1 of 2) reported errors during the parity check. I'm at the point now where the data disks SEEM to be ok and I can see my data on data disk 1 but I lost my shares, configured dockers, and multiple plugins. This it seems was caused by losing the cache pool data, as the users are all still available and many other plugins are still working. I have yet to run a parity check again on both parity drives. Is this installation salvageable or should I reinstall at this point? I've attached my Diagnostics for review. 

tower-diagnostics-20200423-1542.zip

Edited April 24, 2020 by RAMAN
Issue Resolved

[trurl]

Your array data is likely OK, but looks like cache pool isn't, and that is likely the cause of any docker or user share issues.

 

Were you following some FAQ about converting your cache pool?

 

Never tried anything other than raid1 with 2 identical SSDs personally. @johnnie.black is the expert and wrote the FAQ procedures on cache pool. Maybe he will reply.

[RAMAN]

Thanks, @trurl. From what I read, there is no way to read the cache pool data without having the 3 original drives. The 3 original cache drives consisted of the 1 Brand X SSD(which I still have and was left unchanged) and two of the Brand Y SSDs. So I attempted to restore the cache pool by attaching the original 3 cache drives but since two of them were connected to a docking station via USB, they had the same name in the GUI, which led to me trying each combination of drives in order to have the cache pool recognized again. Forgive my ignorance with how the cache pools work, but I wasn't able to get the cache pool back and so I started a new cache pool with 4 Brand Y SSDs, and set the 1 Brand X SSD aside hoping to somehow recover the cache data.

[trurl]

Did you have backups of anything on cache, such as appdata backup with the CA Backup plugin?

 

Setting up your dockers again with the same mappings etc. isn't hard. You can use the Previous Apps feature of the Apps page for that. But without appdata the applications themselves would be starting from scratch.

 

 

[RAMAN]

Oh, I'm not worried about anything other than missing any core functionality within Unraid itself. I'm worried that I removed file dependencies from the base OS, since I don't have appdata, VMs, isos, etc listed as shares any longer. Also, my Previous Apps doesn't list anything any longer, as well. Unfortunately, I learned about CA Backup only after having searched for backup solutions after this all happened  When I browse using mc, I can see the shares in /mnt/user0, but they're not listed in Shares. Does this allude to a more serious problem or can I simply recreate the shares? I do see a docker.img file in /mnt/disk1/system/docker. Can this be restored since the docker service fails to start?

Edited April 23, 2020 by RAMAN

[trurl]

Your shares problem is almost certainly due to the cache problem, since cache is part of the user shares. /mnt/user is the user shares, which isn't there, but /mnt/user0 is the user shares excluding anything on cache.

 

Probably that docker image on disk1 is newly created and the one you were using was on cache. You should disable dockers until you get your cache fixed, and delete that docker image so when you enable dockers again it will be recreated on cache where it belongs. In any case, docker image is unimportant and easily recreated. It is just the downloaded executables of your applications. Without the appdata and the templates (see next paragraph) it is useless, and with the appdata and templates the executables will just be downloaded again.

 

Not sure about your Previous Apps. It works from the docker templates on flash, but if your user shares aren't working I'm not sure how it would behave. And in any case you don't want to mess with any of that until you fix cache. You can see those templates in /boot/config/plugins/dockerMan/templates-user.

 

There is no need to recreate any shares. The user shares are simply the aggregate of all top level folders on cache and array, and if your cache were good they would still be there.

 

Another possibility would be to remove cache completely if you just want to reassure yourself, but I don't recommend enabling docker that way since you would have to do some extra work to get all that moved to cache when you get cache going again.

[RAMAN]

Sure enough, /boot/config/plugins/dockerMan/templates-user had a list of all the extra plugins I had installed. Regardless, I disabled Docker for now to get to the root of the cache issue. So removing cache drives altogether would entail using the New Config option and switching all cache drives to Unassigned? Then would you recommend I start the array again? 

[itimpi]

1 hour ago, RAMAN said:

Sure enough, /boot/config/plugins/dockerMan/templates-user had a list of all the extra plugins I had installed. Regardless, I disabled Docker for now to get to the root of the cache issue. So removing cache drives altogether would entail using the New Config option and switching all cache drives to Unassigned? Then would you recommend I start the array again? 

You do not need to use (and should.not use) the New Config option.   You can simply Unassign the cache drives with the array stopped and then restart the array.

[trurl]

1 hour ago, RAMAN said:

/boot/config/plugins/dockerMan/templates-user had a list of all the extra plugins I had installed.

Technically these are dockers. Plugins are something else and they are installed at boot using the .plg files in /boot/config/plugins

[RAMAN]

I  Unassigned all cache drives and have started the array. The parity check has started with 19hrs remaining. 

[JorgeB]

Just for future reference, the cache problem was because you went from 4 devices to 1, you can only remove one device at a time from a redundant pool, and after that you re-assigned the old devices but without "forgetting" the existing pool, so they were erased.

[RAMAN]

Thank you, @johnnie.black, @trurl, and @itimpi! I will move carefully as I begin to add functionality and configure the server. This was a great learning experience, as it would have been much more disastrous had the server been full configured and this happened. After reboot, shares returned and all I did was recreate a few shares and reinstall dockers. Thanks @trurl for the tip on templates-user for the easy reference on reinstallation. I went ahead and converted to RAID10 without issue and will issue a parity check before adding the second parity drive. Looks like I will start to plan my backup and DR strategy before I move forward with addons. Installation recovered and issue resolved! 

[trurl]

👍