JPHobby

Just ran into this myself. Naturally openclaw was able to spot the error way before I did. But instead of this sh -c "mkdir -p /root/.openclaw; [ -f /root/.openclaw/openclaw.json ] || echo '{gateway:{\"mode\":\"local\",\"bind\":\"lan\",\"controlUi\":{\"allowInsecureAuth\":true},\"auth\":{\"mode\":\"token\"}}}' > /root/.openclaw/openclaw.json; exec node dist/index.js gateway --bind lan"it did recommend a slight variation: sh -c "mkdir -p /root/.openclaw; [ -s /root/.openclaw/openclaw.json ] || echo '{\"gateway\":{\"mode\":\"local\",\"bind\":\"lan\",\"controlUi\":{\"allowInsecureAuth\":true},\"auth\":{\"mode\":\"token\"}}}' > /root/.openclaw/openclaw.json; exec node dist/index.js gateway --bind lan"

Hi, my photo library has grown over 1TB and manipulating it or using things like immich for machine learning is becoming more and more tedious having it on rotating disks. I am thinking about one of three solution: 1) Buy one 2TB+ SSD and add it to my unraid pool (which only has rotating disks but two parities) and making sure that the Share is using that disk only and the others are not using the disk. 2) Buy one, make an additional ZFS pool and use it as cache with mover action cache -> array and mover tuner setting "synchronize primary to secondary" making sure that the mover will eventually copy all changed files to the pool after a few hours. 3) Boy two and create a new ZFS mirror pool for it... I know (3) would probably be best but a bit overkill. (1) would also be optimal for data protection but writes would be slower (which probably is not so big a deal). What would you guys do/recommend? The library is once a night backuped to a secondary system via kopia as well. Regards JP

Cool. Just upgraded to FolderView2. Thanks for the link. I missed that. So I will have to announce a "downtime" to the family and "home office" sometime this week and perform this procedure which according to above messages and "previous apps" should be easy enough to do. Dear @trurl: I am very grateful!

Thanks. That is actually a very good suggestion. Sorry I did not come up with that myself. Thanks! Ok. I am beginning to understand and am very thankful for your patience. Let's exclude docker compose for the moment. When I install a container via apps or in the docker tab via add container there will be a template. And after deleting docker.img all of those templates will show under "previous apps" and I can simply select all and hit "install all selected" and it will download the containers and reapply my settings according to the templates? In that case I would not have to do all the manual "crap" (with the exception of the manual docker network create). Is this correct? Because then --> This would be a livesaver. The docker compose stuff should be in /boot/config/plugins/compose.manager/projects and I assume a simple docker compose up for every project should do the trick!

Thanks. But is it safe to assume that once I deleted docker.img the 58 apps disappear from the "installed Apps" tab and I then have 92 apps in "Previous Apps"? If that is true I will first need to make a list of all apps that I currently have in use. There is no automatism that will magically recreate the 58 apps I have running right now. Correct? Ok. I had that coming... But there is a bit more to the question. First this would mean I now have to go through all dockers and "find" them. Some of them I have installed a year ago and I do not remember if they came from apps or me manually adding them. Is there a way to find that out or is all of that a manual process? Feels a bit tedious. And what about docker-compose (via the plugin): Where are those files stored and will they pertain the image-recreation? Since compose manager plugin remains installed (I presume) the configurations should all persist and I only have to do a compose up on those? For the containers (both installed via app and via add container) there currently are tons of settings. Path mappings, Environment variables etc. How do I get those back (esp. for the ones that I created via add container). There is a big difference between "after creation simply go to previous apps and install the apps again" and First make a list of all apps that are currently installed Then find all containers you manually added For all those containers note their position in the docker folder plugin For all those containers make note of all individual settings (folder mappings, Environment variables, runtime arguments) Then recreate the docker image. Create your networks Go to Previous Apps, find the previously installed apps and reinstall them. Manually create the containers you installed besides the apps. Go to each installed container and manually recreate all settings. etc. I might be misinterpreting something and in the end all of that is very simple because 99% of the settings are retained etc. If not, this looks like a very tedious procedure. One then could say: Stupid why did you not simply make the image much larger. True. Matter of experience. Or why did you not use overlay2 in the first place. Because it was not available at the time. I am not trying to flame. I just want to make you understand why this is worrying me quite a bit. It took months to get everything up and running the way it is. I do not want to jeopardize it. And yes: AppBackup is running. 🙂 But again: What if my img is getting corrupt? Or the server breaks down? I would essentially have to go through above steps (assuming they are correct)... That is far from a simple "restore".

Hi, since my docker.img was about to run full I wanted to extend it and maybe switch to overlay2 while doing that. I am aware of the three magic links: https://docs.unraid.net/unraid-os/manual/docker-management/#re-create-the-docker-image-file https://docs.unraid.net/unraid-os/manual/docker-management/#re-installing-docker-applications https://docs.unraid.net/unraid-os/manual/docker-management/#docker-custom-networks I wanted to clarify some things however. Ok so I disable docker, delete the img file, switch to overlay2 and reenable docker. Then no container will be there? So I have to reinstall them. I am supposed to do that via the apps and go to "previous apps". When I do that NOW I see 34 "previous apps". When I switch to "installed apps" I see 58. This does not seem correct. I am using some containers from docker hub directly which I did not install via apps. Moreover I have four docker-compose "clusters". That might show the differences. But it raises the question: How do I get THOSE back? Fine I can reinstall all "previous apps" but what about the rest? Will the docker folders come back after this or do I have to make that grouping myself as well again? About the custom networks: I assume I need to do "docker network ls" and make not of all non br* and host networks and after the process first create them with docker network create? I am aware that to the pros here these might be silly or RTFM questions. I searched quite a bit and am not sure I got things right especially regarding the non-apps. And the system is too important to "just give it a try", hence my request to you professionals. Regards JP

It was set to 599GB which might correspond to the warning message I have noticed. Silly me. Forgot about that setting. I adjusted it to roughly 80% of disk size!

I am using Move files off Primary (cache) based on age? --> YES with 15 days. The cache is under the threshold. I was of the opinion that even though the usage is not over the threshold the mover plugin should move files older than 15 days to secondary storage. Is this not true or is this a bug? Regards JP

Hi, I just noted "cache disk full" in syslog even though my cache disk still has >500 GB free. Any idea? I have not seen any other warning signs in the GUI e.g. Regards JP

Will the rebalance option honor the split level? I have some directories in my share on the array that for whatever reason are not split across the directories like they should be. unbalance helps but I would have to go to each subdirectory and fix it manually. I was hoping that rebalance not only rebalances between primary and secondary but also would "fix" wrong splits within secondary... Any chance?

Sync does not help. New lines pop up but lines in the middle are missing etc. remounting with cache=none and actimeo=0 does help.... Unless there is an option I will then obviously have to mount this manually....

For log files being constantly read and pushed to graylog this means a sync -f every 10 seconds which would be unnecessary then. I see you point and will see what I can do cron-wise...

Hi @dlandon, just ran in to a small challenge. I am using UD to mount a SMB drive from an IoT device with tons of logfiles. With a filebeats container I tail all these logfiles for changes and push them to a graylog. Works nicely with one exception. Every once in a while everything seems is stuck. While on other clients I can mount the SMB path and see updates on the files, when I go to /mnt/remotes etc. the files seem to be stuck on an old version. I assume this is due to cache=strict and a stupid SMB implementation on the server side. had to tell though. I manually mounted this with cache=none and the problems went away. I realize that "strict" is the better option in most use cases. Just not in mine. Would it be possible to make this option configurable in what way or the other please? This would greatly help me!

Hi, I want to enable some audio devices access to my music collection via DLNA. However the DLNA server and clients are on different subnets/vlans. I get the firewall to forward the multicast requests to port 1900 for discovery and it reaches the Unraid server. But since the DLNA server is on a bridge interface assigned to a VLAN Unraid does not forward the multicast packets due to IPVLAN being used in docker. I do not want so switch the default installation to MACVLAN. Is there a way to either have one container run in a MACVLAN instead of the default IPVLAN and if so, what would the disadvantage be? The other way I can think of is running smcroute in a docker, have that docker be network=host and be part of a custom network only consisting of the DLNA server container and the smcroute container. According to some posts in the net that might work if setup correctly. Does anyone have experience with this or another suggestion? Assuming the smcroute way is something worth trying, how can I attach the DLNA Server and the smcroute not only to the bridge (with a VLAN) or host but have it join the custom network as well. I believe there as an advanced setting for that. Is "&& docker network connect" as a post argument the right way to do this? Moreover how can I persist the custom network setup across reboots (if necessary)? And one additional question: For historical reasons my Unraid has two interfaces. eth0 being the one I do not use anymore with a network attached and IP1 and eth4/5 as a bond with VLANs and IP2. Unraid seems to believe that the "primary IP" is IP1. So if I run dockers with network=host the GUI tells me IP1:port as access. The container listens on all interfaces so that is not the problem but maybe the wrong IP information also gets inside the container and if the container service is announcing "its IP address" it would announce the wrong IP1. Is there a way for me to change this without having to get rid of eth0 altogether (which is something I contemplate as well). Many thanks for all suggestions!

Same problem here. Thanks for your post! Have you made any progress?

My syslog ran full. This is the culprit: nginx: 2024/07/08 04:40:27 [error] 26593#26593: *9435495 open() "/usr/local/emhttp/api/v3/system/status" failed (2: No such file or directory), client: 127.0.0.1, server: , request: "GET //api/v3/system/status?apikey= HTTP/1.1", host: "127.0.0.1" ... repeating over and over. After deleting syslog (in order to get something working) it does not reappear so far. Anyone any idea what in Unraid, plugins oder docker could cause this? Yes I googled it... 🙂

I was suspecting this answer. Sorry that I did not clarify. I am aware that it is not strictly necessary. To my understanding, does preclear not have the benefit that the drive added is immediately available with full parity protection? Anyway: Even if it is not strictly necessary: Does the disc still contain the correct preclear signature and how can I check? Or will unraid start zeroing it again once i add it to the array?

Hi, weeks ago I put 2*10TB discs in the system and did a successful preclear. I never added them so far. Now after a few reboots the unassigned drives plugin still shows the discs but the "preclear finished successfully" vanished. Is the signature gone? Do I have to redo the preclear if I want to add them ? How can I check? Regards JP

Hi, just migrated my Nextcloud instance from a FreeBSD system to standard dockers running on unraid. DB, Redis, nextcloud, elasticsearch all running and I am quite pleased with the speed…. However: The official Nextcloud Container misses things like tesseract which I was used to having. Question: what is the docker way of doing this? I keep reading „well just create your own image and either install tesseract in it or do docker compose voodo“. Excuse me for sounding confused. I am meaning no offenes but try to learn. Can someone try to teach an old pre docker dinosaur how to create “my own image” based on the official one but having e.g. tesseract on board? And while we are at it, if I could create my own elasticsearch container with the ingest plugin installed my setup would be back to normal again but based on dockers. This would be awesome. Any help, link, direction would be greatly appreciated.

6 data disks plus 2 parity Currently 90-100 MB/sec constant writes (which is why I want reconstructive write to stay on). 🙂 I currently set "spin down" to never for the time being and keep filling up to drives. Will try to reproduce in 1-2 weeks and if I can reproduce it will send diagnostics here.

I do but: That plugin also claims "reconstructive write" active. I set the number of disks to maximum (so it should never kick in and reduce it). And in order to make sure it does not fiddle with it I disabled it. The setting under Disk settings does still show md_write_method = reconstruct write and the plugin is disabled. Still believe me: After a while disks are spun down and I do see reads only from the spun up disks. Once I manually spin all disks up things are back to normal. VERY confusing.

Hi, I am in the middle of transferring TBs of data to my new Unraid NAS. I have reconstrive write set and it seems to work. But every once in a while I discover spun down discs and slower write performance. It was my impression that when reconstructive write is enabled every write should spin up all disks. Any idea? Regards JP

Hi, I looked around but found inconclusive information (or at least it appeared so for me). I have a 2TB NVMe cache pool (zfs mirror) and want to make sure, that the cache functionality for the array does not fill it up to more than 70%. After that unraid should start writing to the array directly. Is the setting "minimum free space" at the pool level correct than? Moreover at the share level I set "minimum free space" to a bit larger than the maximum file size I would expect for the share (depending on the share type I set different values). Is this correct? Moreover for the first initial data fill I assume disabling cache is advised in general?

Hi Jonathan, Supermicro 36bay enclosure with two Expanders. Each connected to a port on the internal LSI 16i. After I stopped the writes it seems to speed up to 270MB/sec which is the maximum I would have expected. Will let it finish now and start the writes then. Moreover my two NVMes for the pool arrive tomorrow. So new day new luck.

Hi, just fired up my Unraid with 20 TB Exos disks. Started Parity. Two things I noticed: The parity disks (which are in another enclosure) started off with tons (thousands) of UDMA CRC errors. Currently things are quiet. Might be cabling issue. I am unsure yet. The parity buildup is INCREDIBLY slow (10 MB/sec) and will take like a month. I am in the background syncing my old NAS with rsync streams to the Unraid server. So my assumption is that due to the random writes the parity is so slow. Is this feasible? Or am I looking at something else and need to investigate? Regards JP