Husky

Could you try connecting the NAS and your PC directly together with a single cable and running a test that way? Then we can determine whether the switch or one of the machines is at fault. You could also boot a live operating system from a USB drive on the NAS as well as your PC (such as Ubuntu 22.04 LTS) and run an iperf on both, again, to rule out any issues with drivers and the NICs.

The RAM rule is not quite accurate and is greatly overblown. You can run ZFS on a system with 128 MB of RAM if you want to and it will work. However, the more RAM you have, the more RAM ZFS can use as cache which can speed things up. For example, for a home user, 4 GB of RAM would be enough for a few TBs of storage. Remember that this RAM is not reserved, the OS will still be able to use this memory for other software as needed as it is simply cache memory and can be overwritten at any time. You can get ZFS to communicate to the disks directly by simply making sure that the disks are connected to a standard "dumb" SATA/SAS controller or HBA, and NOT a RAID controller. In addition, make sure that the SATA/SAS controller or HBA is in regular SATA AHCI (or SAS SCSI) mode (NOT RAID, Intel Matrix, ESRT2, vROC, etc...) and that's it. The SATA ports built-in to motherboards work perfectly fine for this as long as RAID mode is disabled and AHCI enabled in the UEFI/BIOS. I prefer TrueNAS Core simply because it's more mature than TrueNAS SCALE, which is a recent development. Apparently SCALE is stable and ready for use but I'm cautious especially when it comes to server stuff. I prefer FreeBSD Jails personally because I just feel more comfortable with managing them due to how simple and straightforward they are. Docker is more popular though, and most modern server apps have a Docker container ready-to-go which is often times not the case with FreeBSD Jails. TrueNAS makes this easier though since they have "plugins" which is a fancy name for "jails that are already setup and ready to go in a few clicks". I would honestly just take a few OSes for a test drive in a VM and then you can make your decision. Some of the recommendations by other forum members are also good options such as unRAID or Proxmox. As for permissions when dealing with Windows machines, it's really not too bad to get right once you've got your head wrapped around how UNIX permissions and POSIX ACLs work. An easy way would be to create a group (example: media) and then add everyone that needs access to that group (your own user, daemon users such as plex, etc...) and then chown all files to be owned by that group, chmod something like 2770 (rwxrws---), and set ACLs to force all files to be owned by that same group. There are also settings in Samba (SMB file sharing server) to force all files to have certain permissions and owners which are quite handy. TrueNAS has a Web UI for you to do these operations with so it shouldn't be too difficult.

Mangling and load balancing are different things. Mangling is simply a way of marking packets so that you can do some special processing with them later (it's a general thing that allows you to do lots of cool tricks). Load balancing is trying to balance the load evenly across multiple links (either for redundancy or extra throughput or speed). Some types of load balancing setups make use of mangling to make them work, some others do not use mangling. It depends on load balancing setup. In general - if you have two links and want to load balance a single stream of packets (such as a single TCP connection), it's not going to work very well. It works better for multiple streams (or multiple TCP connections). If you for example download a file using HTTP in your web browser, it will only utilize one of your links. But if someone else on your network starts another download, then their connection should be balanced to the other link. Your router will be able to tell the two streams apart and put each one on a separate link. Then you will be pulling down 500 Mbps on one link and they will be pulling down 200 Mbps on the other link for a nice 700 Mbps total. Instead of that other user sharing your 500 Mbps link, they were automatically balanced to the 200Mbps link because the router knew that the 500 Mbps link was busy and the 200 Mbps link was free. This is works very well in a business environment where there are multiple users all sharing the load-balanced internet connections, the load is able to be balanced quite evenly across them to increase total throughput (and provide redundancy in case one internet link goes down) instead of everyone hogging one single link and congesting it to a crawl. So yes, you can use multiple links at the same time - just not for a single stream. It works well for when you have multiple downloads/uploads happening at once (torrents can take advantage of this!), then it's able to split them across your links and you get full speed which is awesome. I think you should follow whichever example fits your needs best from that documentation and enjoy your network.

If you want an easy-to-use WebUI then TrueNAS Core (or TrueNAS SCALE if you want native Docker support instead of using FreeBSD Jails) is a good choice for operating system. You could also opt for a standard FreeBSD install, or a Linux distribution that works well as a server, such as Debian, RHEL (or its derivatives), etc... and do everything from the command line if you are comfortable with that. Ubuntu Server has ZFS built-in unlike other Linux distributions which can be an advantage in some cases. I would avoid hardware RAID unless you have a specific use case which requires it. Most applications nowadays benefit from the simplicity, flexibility and portability of software RAID. ZFS is a really good choice - it combines what would usually have been several different storage layers managed by separate tools into one unified stack, making things simple, reliable and fast. Ensure that the controller you connect your hard drives to is a regular SATA/SAS controller or HBA (not a RAID controller) to ensure that ZFS can communicate to the disks directly. Onboard motherboard SATA controllers are fine. I have run Ubuntu Server, TrueNAS Core, and FreeBSD on my servers at various times. All have worked great, some better than others for certain tasks. You can always take each option for a spin inside a VM and pretend it's your server so that you can decide which option works best for you. You can literally create a "mini" virtual version of your server with several small virtual disks and set up all your services such as Plex, UniFi controllers, etc... in that environment and whichever option you find makes most sense will become obvious after taking them all for a spin.

Have a look at MikroTik's official documentation on load-balancing and mangling: https://help.mikrotik.com/docs/display/ROS/Firewall+Marking (you may be interested in examples 2, 3, or 4) Due to you having a CRS (Switch), don't expect very good performance with firewall or mangling operations as the CPU is optimized more for switching applications than firewall applications. But it will still work. You may have to switch to RouterOS on your CRS instead of SwOS, which is very simple. Also beware that "combining" two WAN links into one isn't as simple or straightforward as it may sound - you cannot send packets randomly out both at the same time for a single packet stream as the destination will be very confused as it doesn't expect out-of-order packets coming from two difference sources. So usually the connections or packet streams are all kept together, and separate streams are balanced across the two links. Because of this - you won't usually be able to simply get 500+200 = 700 Mbps download speed because you "combined" them - unless you are downloading multiple streams of data at the same time, then it can work and get you that higher speed.

If you are using the monitor with your PC, you should set your PC to output a full color range signal and leave the monitor in "High" so that you get the full proper 0-255 level range instead of the limited 16-235 range for TV. If your PC and monitor settings for black level are mismatched, you may get crushed blacks and overblown whites, or dark whites and light grey blacks. AMD Radeon Software has this setting in the Display section, set it to Full RGB (444). For Nvidia, the Nvidia Control Panel has it somewhere under the display section. Intel also has it somewhere. Make sure it's on Full Range or "PC". And set the monitor to "High". This will ensure the correct color range.

ext4 can take longer than other filesystems to format as it writes a lot of accounting, superblock and inode table information. However, you should use the native Linux tools to format the disk instead of EaseUS, and you can additionally specify a lazy table initialization which will format the drive faster, so it will allow you to use the drive while it finishes initializing in the background. It would be best to copy everything off the drive to something else and then format the entire drive. Example (replace X with the actual disk number, find it by running lsblk, caution, the following examples will destroy all data on the disk): No partitions (format the entire disk directly): "mkfs.ext4 /dev/sdX -E lazy_itable_init" (formats the disk with the ext4 filesystem lazily) (Note, without a partition table, you will not be able to create additional partitions on this disk later, so if you need partitions, use the below method rather) Single partition using GPT/GUID partition table: "fdisk /dev/sdX" (starts fdisk against disk X) g (creates new GPT partition table, destroying the previous one) n (enter and enter again to accept defaults, creates a new partition spanning the entire disk) w (writes the changes to disk) "mkfs.ext4 /dev/sdX1 -E lazy_itable_init" (formats the new partition on disk X with the ext4 filesystem lazily) Please do not reboot the Pi or unmount the disk for at least 2 or 3 days afterwards to ensure that the filesystem has finished initializing.

Maybe you could search the entire filesystem for files containing the string "/etc/backups/mysql" or "/etc/backups/mysql/mysql-backup.log" using something like find, grep, or ack (ack-grep in Debian-based distros, including Ubuntu)? Maybe you will find the script that way since the script has to contain the path of the backups? Here is a link to some examples of commands you can use that could achieve this: https://stackoverflow.com/questions/16956810/how-do-i-find-all-files-containing-specific-text-on-linux

Yeah that sounds pretty bad - it is probably dead. One last thing you can try to do to confirm is to boot a live USB of Linux. I'd recommend Fedora Linux as it has up-to-date packages and works brilliantly on most hardware. Simply write the ISO to a USB and boot from it - you don't have to install it or anything, just use it live and see if you get a display output, and if you can use the system without it crashing.

You can achieve this with a newer method rather than using /etc/resolv.conf Simply create files in /etc/resolver/ named for example: "example" (sudo nano /etc/resolver/example) with the contents "search example.com". Then it will use those search domains in that folder for ANY connection now, even if that search domain isn't in System Preferences. You can add as many as you want into that folder and it will search them all order until it is able to resolve the DNS name.

Yeah I suppose it depends from person to person. Some people absolutely despise acceleration and want nothing to do with it in games (like me), while others actually find it helpful and even play better with it. I suppose it just depends on what makes the most sense for you, and the only way to find out is to play with both and test to figure out which way you prefer.

Sometimes, yes, if the game does not use "hardware cursor". Most well-written games either use hardware cursor or have an option to use it or not, but some older games or those without that functionality may stack acceleration on top of acceleration. This is why it is recommended to disable Windows's mouse acceleration (uncheck "Enhance pointer precision") and set the speed to exactly 6 clicks in the mouse control panel applet. If you need to adjust your mouse speed, you should then rather use the DPI settings on your mouse instead of Windows's settings. This should give you the best compatibility with old and modern games and ensure that your mouse isn't being messed with.

You can replace it with a drive from any of the major brands. Western Digital (WD), Seagate, Hitachi, Toshiba, etc... they are all great drives. I have had a mix of both WD and Seagate throughout the years and they both worked great. What you read online is mostly heavily biased as some people just get unlucky. I've seen plenty of hate for both WD and Seagate all around the web, when in reality they both make great drives. Just go and pick a drive that is the best value for you. Pro-tip: some newer cheaper drives are using SMR (Shingled magnetic recording) technology, which results in much slower write speeds. You can check the datasheet of the drive you want to purchase to make sure it says either "CMR", "PMR" or "Conventional recording" if you are going to need decent speeds. Otherwise, if it's for storing media and write speed isn't too important to you then you don't have to worry.

That drive has bad sectors and is therefore probably dying. You should back up your data as soon as possible and replace the drive. The S.M.A.R.T data says that the drive has encountered sectors that it struggled to read, which it then managed to relocate to a special area on the drive that is reserved for situations like this so that those sectors can simply be relocated instead of being lost forever. If the reallocated sector count keeps rising and rising, then it is likely that the drive is on it's way out. If the count stays stable and does not increase, then the drive might still be usable but I would highly recommend replacing it sooner than later.