ausca

Hi, I want to remove my cache, I have a bunch of shares that are set to prefer cache, including my VMs. how do I do this safely? I can set the shares to not use cache, but I figured this would only take effect when the mover works? If files are in use (like the VM disks) my understanding is the mover won’t move them? im guessing the right approach is to set disks to no cache, stop the VMs and invoke the mover manually? But can anyone else confirm? also how do I know for sure there’s nothing left on the cache?

That's fair. You can't fix all the container & app problems. I think that the log retention is a viable workaround for them. @Squid not sure where your screenshot is taken from though, but my UnRAID instance looks different to yours, even though I am running 6.9.2. I am missing the "Docker LOG number of files" and "Docker LOG maximum file size" despite setting the view to Advanced View.

Interesting! I've only been an UnRAID user since 6.9 and yet my logs don't appear to be getting rotated as described. I had one log JSON @ 9GB (my personal docker container registry that seems to log everything including GET requests to syslog). The problem is that some of the bad log design you're talking about are not at control of the (consumer) of the docker container, but at the control of the author of the container or containerised application. More on that assumed "mission" I referred to in my original post, I don't know if this "mission" extends to masking poor logging within containers/apps in-order to make this as simple as possible for the person consuming the container image. Especially considering that I would say the average UnRAID user might not know how to fix the logging issue. There are plenty of posts here on people trying to identify which container is even the cause of the problem. My feedback comes from a position of love and not hate. Docker is a complex subject and I appreciate the efforts in "consumerising" it in UnRAID. For me I think I may go back to using docker in a VM for now.

This post might anger some people, But considering the amount of confusion that this issue is causing, I wonder if there needs to be a better solution to this problem that is not instructing people to "run this command from their shell" or "add this advanced configuration to each of your docker files to truncate the logs". I'm in the same boat, I've identified myself that it's logging causing the problem & also a very large btrfs folder! I feel like the intention of docker support in UNRAID is to make management of docker a little bit more simpler than just running docker in a VM, but the management overhead of keeping the volume disk size low seems contradictory to that mission - can we just not make this simpler?

a lot of people have been waiting some time for NFS4 I think they’d understand/still appreciate if you took an iterative approach. Supporting NFS4 at all in 6.10 would be cool..

Do you know if the tachometer on a PWM fan is compatible with a voltage controlled fan?

Will 6.10 be the release we finally get NFS4? Awesome news @gcolds

Such awesome news. I wanted to test too, but when I tried to install the plugin to get me to rc2 on a test box, I got the error "plugin: not installing older version". Weird?

Because the Icy Box controls that fan plug using a thermostat which only switches on at high disk temperatures. I could theoretically replace the stock fan with a non-PWM three pin fan and use the Icy Box controller, but I want the fans always spinning at low RPM to keep disk temperature low. The Icy Box fan control design seems to be built around the concept of cooling drives down, rather than keeping them steady at a certain temperature. This is partly the reason why I replaced the fan with one controlled using the onboard PWM plugs - I can now control the temperature using ipmi on the motherboard.

Which has given me the idea to deliberately obstruct the fan from spinning to see if the warning lights come on..

Yeh this is what i don't know - it's a three pin fan, so the yellow wire is used for a square wave signal tachometer. The boot "process" for the drive is that it spins the fans up to what appears max RPM, and then turns them off if the heat is lower than the pin threshold (set on the back of the box). The more that I think about this, the more I suspect it's spinning it up and reading the tachometer response to see if the fan is actually spinning - because that would detect things like an obstruction causing the fan to not spin. Research shows me I could probably create a timer circuit to emulate the square wave.. starting to become quite complex!

It's just extremely positive to hear this isn't going on death ears and being ignored. Would be amazing if we could make nfs4 happen in 6.10. I'll see if i can do some tests today. I'm using NFS for docker swarms on raspberry pi's using the in-built mount options.

@DerCarsten Yeh the saying "Jack of all trades, master of none" really does come in to play here. There does appear to be an identity crisis happening. Has there been any offical reasoning behind no NFS4? E.g, has someone from Unraid come out and say "We tried, but are blocked by x" or is there just radio silence? I'm not sure which annoys me most, a lack of NFS4 or no acknowledgement.

My 80mm noctuas have come today and I’ve modded the icybox to support it. unfortunately the noctua is slightly thicker than the stock fans with the icybox, so I had to take the icy box apart and attack the fans so the screws come from the inside to the outside, rather than outside in. the pwm fan is now always on at low rpm and I can control it using ipmi from the motherboard based on the disk temperatures. The disks are now nice and cool! one annoying thing though, the icybox seems to do a fan check when it turns on, and if it detects a fan failure, the front lights will flash on one second interval between blue and red and it doesn’t appear you can disable this. The drives will still turn on though thankfully. You can fix this by attaching the stock fan at boot and then removing it, and because of that I suspect that I will be able to permanently fix it with the use of a resistor between the pins.. now I just need someone with electrical experience to help me decide what resistance I need.

Thread has been going on for quite sometime but there's no updates. Why is NFS4 which has been around for over 20 years so hard to implement?