• Posts

  • Joined

  • Last visited

  • Days Won


jortan last won the day on February 12 2021

jortan had the most liked content!


  • Gender

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

jortan's Achievements


Contributor (5/14)



  1. /mnt/user is where unraid does it's magic for shares backed by Unraid arrays / cache devices. My advice would be to leave /mnt/user alone and not attempt to manually mount (or link) filesystems here. Maybe I'm missing something but can you not go to: - Settings - VM Manager And set "Default ISO storage path" to /zfs-sas/isos ? Fairly sure this is what most Unraid/ZFS users are doing.
  2. Can confirm the current "next" build of 6.10.2-rc3 is also working if you wanted to try that. VVV oops, you're right - thanks
  3. The other thing you can do is zfs set sync=disabled poolname This forces all writes to be asynchronous. This doesn't risk data corruption, but it does risk the last 5 seconds of your data in the case of a power failure which could lead to data loss in some scenarios. If you happen to be moving data to the array at the time - The sender has been told the data is moved so the source is deleted, but it won't actually get written to the destination for 0-5 seconds. You could see significant performance benefit on a busy pool though - particularly one seeing lots of small synchronous writes.
  4. This seems about what I would expect. You're not streaming data directly and uninterrupted to your spinning rust as you would be in a RAID0-like configuration. For every write to one disk, ZFS is having to store that data and metadata redundantly on another disk. Then the first disk gets interrupted because it needs to write data/metadata to provide redundancy for another disk. You're not streaming data neatly in a row, there are seek times involved. If you want performance with spinning rust, get more spindles and ideally switch to RAID10 (mirrrored pairs)
  5. To clarify, it's relevant for any application doing synchronous writes to your ZFS pool. For these writes, the file system won't confirm to the application that the write has been completed until the data is confirmed as written to the pool. Because writes to SLOG are much faster, there can be significant improvements in write performance. SLOG is not a read cache - ZFS will never read from the SLOG except in very rare circumstances (mostly after a power failure.) Even if it could, it would be useless as the SLOG is regularly flushed and any data is already in memory (ARC) If your application is doing asynchronous writes, ZFS stores the write in memory (ARC) and reports reports to the application that the write was complete (it then flushes those writes from ARC to your pool - by default every 5 seconds I think?) The SLOG has zero benefit here. I have a feeling QEMU/VMs will do synchronous writes also, but I don't have an SLOG running to test. From memory NFS shares default to synchronous and SMB shares will default to asynchronous? >>Intel Optane P1600X 118GB That said if you have/are getting this for L2ARC anyway, you may as well slice off 10GB for a SLOG. I was running a P4800X in exactly the same way a while ago.
  6. Yes for HDD pools zstd makes more sense as you are IO bound by slow spinning rust. Needing to read and write less data to spinning rust is more likely to be beneficial than any increase in computation required for compression/decompression. For NVME you are less bound by IO, so the increase in computation required by zstd is more likely to impact performance negatively. On modern computers it probably makes very little difference. If you're rocking 10-year-old Xeons (like me) then it might. That said, zstd also gives better compression (to varying degrees, depending on the type of data) so if that's important to you it's another thing to consider.
  7. For NVME you're probably better of using lz4 by default, and zstd for datasets with very compressible data (large log/text files), or for datasets you don't read often (archives/backups.) Some insights here:
  8. You may want to look at ZFS + snapshots using sanoid + syncoid if you want to replicate those snapshots to another machine. It's a lot more manual work to configure and maintain, but a good learning experience if you're up for it. It seems native ZFS support will be coming to Unraid eventually. It's great for running dockers/vms, though with ZFS you don't have the ability to cache writes to the array within Unraid like you can with the built-in "pool" functionality (my guess is this will be possible when Unraid supports ZFS natively) There's a number of Youtube guides for implementing ZFS in Unraid also.
  9. From what I understand it has long been the case that some people report issues with docker on ZFS and some people have none. This might be due to ZFS only having problems with specific containers? I've had issues with dockers using "sendfile" syscall on ZFS previously: But it seems likely this is fixed now: Could this have caused some of the other docker + ZFS issues seen in the past? I've had issues with docker + ZFS previously (both using docker.img and using a direct file path on ZFS). I've never used ZFS zvols. I don't have the bandwidth right now to try migrating this back to ZFS. I will try to revisit this when 6.10 is released.
  10. Typically this is ~/.vimrc (aka /root/.vimrc) but this path isn't located on persistent storage so it won't survive reboots. Run this once but also add it to /boot/config/go echo "set tabstop=4" >> ~/.vimrc
  11. zfs import is what you wanted here, not zfs create I suggest before you do anything else, you zfs export the pool (or just disconnect the drive) prevent any further writing and consider your options (but I'm not sure if there are any)
  12. More deduped data also means more compute resources to compare each write to hashes of all the existing data (to see if it can be deduplicated) I don't know for sure, but I suspect not - in the same way that adding a normal vdev does not cause ZFS to redistribute your data or metadata. It should by design, but it also might just break (situation may have improved in the last 11 months?)
  13. To clarify earlier comments about ZFS memory usage - the ARC doesn't show how much memory ZFS needs to function, the ARC will dynamically consume memory that the system doesn't otherwise need. This is why you can't assume how much memory ZFS "needs" for dedupe/DDT based on ARC size before/after turning on dedupe. It is expected that the ARC would be the same size in both scenarios. To demonstrate that the ARC is dynamic and doesn't actually show how much memory ZFS "needs", you can artificially reduce the amount of memory available to ZFS by consuming memory in a ram disk mount -t tmpfs -o size=96G tmpfs /mnt/ram/ As you copy files to the ram disk, and as available memory approaches 0%, the ZFS ARC will release memory back to the system, dynamically reducing its size: \ ZFS will continue functioning without issue (but with less data cached) until the DDT starts getting pushed out of the ARC because it no longers fits. This can happen if you: - keep adding data (DDT size increases), and/or - reduce the amount of memory available to ZFS At this point, performance of the ZFS filesystem will reduce, likely by orders of magnitude.
  14. Neither me nor the documentation I'm referencing are saying dedupe performance with a special vdev is bad (at least not over dedupe performance without a special vdev) Without the special vdev your normal pool devices will very busy writing your data, but also all the hashes/references for the deduplication table (DDT). For spinning rust disks, this is a lot of additional random I/O and hurts performance significantly. Unless designed very poorly, the special vdev will increase performance because it spreads the DDT writes to separate, fast storage devices. The DDT is not a cache, and neither is the special vdev --- The DDT needs to be stored within your pool (in vdev or special vdev) and constantly updated for every block of data that you write to the pool. Every write involves more writes to the DDT (either new hashes or references to existing hashes) When ZFS writes new entries to DDT (or needs to read them from the pool/special vdev) it's cached in memory (ARC) and will push out other information that would otherwise be stored in the ARC. If your DDT becomes large enough to exceed the amount of memory that ZFS is allocating for ARC, that's where you will run in to significant performance issues. That's not counting the fact that there is other data that ZFS wants to keep in the ARC (regularly accessed data and ZFS metadata) for performance reasons. Keeping the hashes/references to what has already been written in memory is fundamental to how de-duplicating file systems work. However, those hashes/references are fundamental structures of the file system, which is why they can't only exist in memory and must also be written to the filesystem.
  15. By default openzfs on Linux will "consume" up to half of system memory for the ARC, subject to other memory requirements of the system. This memory is allocated to the ARC regardless of whether dedupe is enabled or not and the amount allocated can't be used to measure how much memory is being used by dedupe. The special vdev provides faster read/write access for permanent storage of the metadata and DDT, but the DDT still needs to fit in ARC memory to avoid significant performance issues. The amount of memory required to keep DDT cached in ARC ... zdb doesn't function by default in Unraid due to lack of persistant storage location for its database. You can enable zdb database in memory using these commands: Further recommendations for estimating DDT size in a ZFS pool (and subsequently, the memory required for performant dedupe in ZFS): Hope this helps.