Jump to content

boxer74

Members
 View Profile  See their activity

  • Posts

    158

  • Joined

  • Last visited

 Content Type 

Posts posted by boxer74

    Help with ZFS pool setup

    in General Support

    Posted

    Hi All, planning out the pools in my new server. I've got 8 disks for the unraid main XFS array, which has dual parity. I also have a pair of nvme drives which will be a zfs pool for docker and vms. I will also likely have a pair of Samsung 2.5" SATA SSD's as a zfs mirror for cache. I have a homelab 15 so lots of room for drives. I want to create a zfs storage pool for my critical shares, such as immich photos, nextcloud data, etc. I have 7x 4TB Ironwolf hard drives available. Some ideas: 1) 2x 2x-drive mirror vdevs 2) 4 disk raidz2 3) 6 disk raidz2 4) 7 disk raidz3. I have about 4TB of data to store on this pool currently. It will grow slowly as I add new family photos. Due to leaving some space for zfs headroom, options 1 and 2 may not give enough longterm capacity for growth, but at least with option 1, it is easy to expand. Since I already own 7 disks, is option 4 outrageous? Let me know your thoughts.

    Soon™️ 6.12 Series

    in Lounge

    Posted

    I think it would be a bit much to expect ZFS feature parity with a project like TrueNAS that has always been ZFS focused for years. I think allowing ZFS in pools is a nice complement to unRAID. Snapshot/replication management would be nice, not only for ZFS but BTRFS as well, but I think the existing LUKS encryption implementation is a suitable compromise for now.

    Soon™️ 6.12 Series

    in Lounge

    Posted

    2 minutes ago, FlyingTexan said:

    Can someone in layman’s terms tell ‘em the advantage of switching from BTRFs?

     

    ZFS has been around much longer and is battle tested in production with many businesses. I would trust my data longterm with ZFS more than BTRFS. Other than this, they are pretty similar in terms of feature set.

    Soon™️ 6.12 Series

    in Lounge

    Posted

    Exciting to hear some official details on this.

     

    I will keep my media on a traditional unraid array but migrate important stuff to a zfs pool (likely a mirror of two 4TB hard drives). The important stuff would be documents, family photos, etc.

     

    I will likely also migrate docker and vms to a zfs pool of two nvme disks.

     

    Loving the potential here.

    • Like 1

    [Support] Josh5 - Steam (Headless)

    in Docker Containers

    Posted

    Thanks for this container. What a slick setup. While waiting for my new GTX 1650 to arrive (yes, I know - not a high-end card but still better than what's in the Steam Deck and should be fine for streaming), I was about to setup a Windows VM but then saw this. I tested Hades using a Quadro P400 that I have and it actually worked reasonably well. Obviously that card is no good for larger games. I also ordered a Retroid Pocket 3+ to use as a streaming handheld. I find the Deck too large, loud (fan) and with poor battery life, I don't feel the urge to use it. I am more often using my Nintendo Switch but the lack of AAA titles is tough to take at times. I think this could be the ideal setup. It's also nice having a nearly bottomless pit of Unraid storage to hold all the games.

    [Plugin] IPMI for unRAID 6.1+

    in Plugin Support

    Posted

    On 1/5/2023 at 5:19 PM, adminmat said:

     

    So I got the sensors to show and the fan controls working again. Now I can't explain why but I have to leave Enable Network / LocalhostConnections set to No. And leave the IP address username/pw blank. If I enable this is will no longer show my sensors or operate the fan controls. (Maybe I'm misunderstanding what these first 5 setting do)
     

    1006775158_ipmisensors.png.fef63b5f5b653843fd7a0cdc2c61edaf.png

     

    Then, no matter the settings if you click on Readings nothing happens. That tab doesn't open. 

     

    readings.png.9dd1d386dd9b860f45374bc34e96f293.png

     

    You have to go to the Tools tab and at the bottom click on IPMI.

     

    2122216493_toolsipmi.png.081eae8e81d9859245fcfb038c6ce497.png

     

    Then it shows all my sensors, fans etc. 

     

    1540284943_sensorstools.thumb.png.b34cd0fe9934636d9069478c4f0b64fb.png

     

    Hope this helps. Now I just need to figure out how to set the thresholds lower. I have to keep my fans (mostly Noctua) at about 700 RPM or Supermicro starts to spool up the fans into panic mode. 

     

    I'm finding the same, but my situation is unique. I'm trying to run unraid as a VM in proxmox. I'd like to use the IPMI plugin to control fan speeds on the host based on the HD temps which are passed through on an LSI HBA via PCI passthrough. Can't get any sensors to show up.

    VLAN support

    in General Support

    Posted

    For this to work, I had to add --dns=X.X.X.X to the plex container.

     

    What I've now found is that my 'arr containers cannot connect to plex without using an external address. They are on different subnets but allowed by my firewall to communicate. I think this might be due to docker networking. What is the ideal way to set this up. I assume most people just have a single subnet and so don't need to worry about this.

    VLAN support

    in General Support

    Posted

    I believe I've enabled settings correctly to add a vlan, configure it in docker settings but when I try to assign a custom ip in that VLAN for my plex container, it has no DNS. I also see an error message during boot that eth0.3 can't be found. Any advice?

     

     

    boot.png

    docker.png

    network.png

    Security Concern - Constant Rsync/VSFTP/SMBD Connection Attempts

    in General Support

    Posted

    Every few hours, I see the following messages in my syslog:

      

    Nov 24 09:36:59 ur1 rsync[6282]: connect from 192.168.2.1 (192.168.2.1)
Nov 24 09:36:59 ur1 vsftpd[6281]: connect from 192.168.2.1 (192.168.2.1)
Nov 24 09:36:59 ur1 rsyncd[6282]: forward name lookup for DreamMachine.localdomain failed: Name or service not known
Nov 24 09:36:59 ur1 rsyncd[6282]: connect from UNKNOWN (192.168.2.1)
Nov 24 09:37:10 ur1 smbd[6284]: [2021/11/24 09:37:10.442874,  0] ../../source3/smbd/process.c:341(read_packet_remainder)
Nov 24 09:37:10 ur1 smbd[6284]:   read_fd_with_timeout failed for client 192.168.2.1 read error = NT_STATUS_END_OF_FILE.
Nov 24 09:39:22 ur1 vsftpd[7804]: connect from 192.168.6.1 (192.168.6.1)
Nov 24 09:39:22 ur1 rsync[7805]: connect from 192.168.6.1 (192.168.6.1)
Nov 24 09:39:23 ur1 rsyncd[7805]: forward name lookup for DreamMachine.localdomain failed: Name or service not known
Nov 24 09:39:23 ur1 rsyncd[7805]: connect from UNKNOWN (192.168.6.1)
Nov 24 09:39:33 ur1 smbd[7807]: [2021/11/24 09:39:33.981382,  0] ../../source3/smbd/process.c:341(read_packet_remainder)
Nov 24 09:39:33 ur1 smbd[7807]:   read_fd_with_timeout failed for client 192.168.6.1 read error = NT_STATUS_END_OF_FILE.

     

    192.168.2.1 is my LAN gateway IP.

    192.168.6.1 is a VLAN gateway IP for the VLAN on my UniFi network that all my docker containers are isolated on. I have firewall rules that prevent communication from the docker VLAN to my LAN. I have WireGuard running on Unraid and setup a static route as well as allowed host communication with docker containers using custom networks as recommended in setup instructions.

     

    Any ideas what is causing these constant connection attempts?

    BTRFS checksum error - help please

    in General Support

    Posted

    2 minutes ago, JorgeB said:

    I would wait some time and run another scrub, if more corruption is found there's likely an underlying hardware issue.

    Thanks. I'll keep an eye on things. I have a scrub scheduled for Dec 1st. Does unraid send a notification for btrfs errors? I didn't get any with this recent error.

    BTRFS checksum error - help please

    in General Support

    Posted

    My server recently locked up during update of a few docker containers. I was able to save a diagnostic. It appears that my docker.img had some corruption. I've since deleted it, created a new one and added back all my containers. I also ran a BTRFS scrub on my cache drive and it came back clean. What is concerning is that disk 1 in my array which has BTRFS filesystem reported 1 corruption.

      

    Nov 21 08:09:19 ur1 kernel: BTRFS warning (device dm-0): csum failed root 273 ino 9248 off 6311936 csum 0x1a753a94 expected csum 0x58a4bd31 mirror 1
Nov 21 08:09:19 ur1 kernel: BTRFS error (device dm-0): bdev /dev/mapper/md1 errs: wr 0, rd 0, flush 0, corrupt 1, gen 0

      

    btrfs dev stats -c /mnt/disk1
[/dev/mapper/md1].write_io_errs    0
[/dev/mapper/md1].read_io_errs     0
[/dev/mapper/md1].flush_io_errs    0
[/dev/mapper/md1].corruption_errs  1
[/dev/mapper/md1].generation_errs  0

     

    I'm currently running a BTRFS scrub on disk 1 to see if I will learn what file has issues. I can restore from a backup if necessary. I'm not sure if scrubbing will give me this information.

     

    This experience has me questioning my choice to use BTRFS.

     

    Advice would be greatly appreciated. Thanks.

    ur1-diagnostics-20211122-0833.zip

    Anyone Added an SSD to HPE Proliant Gen10+?

    in Hardware

    Posted

    I have a g10+. I'm currently reducing size of my media collection to attempt to consolidate onto 2 data disks with 1 parity disks. This will leave 1 of the 4 drive bays open for an SSD cache. My PCIe slot will be occupied by an Nvidia P400 GPU for plex and tdarr. I love the small form factor and build quality of the g10+ even though it limits drive expansion. With 16TB and 18TB drives out now, this is less of an issue.

    USB 3.2 gen 2 SSD cache pool?

    in General Support

    Posted · Edited by aberg83

    If this were used only for docker, would it be a concern? I have an HPE gen 10 plus with only 4 storage bays and the single PCIe slot will be populated with my GPU for HW transcoding. Therefore, no additional drive slots other than to use USB. Luckily it has two gen 3.2 slots on the front. I would just use a single USB SSD for docker on an unassigned device and backup weekly to the array. People use these as OS boot drives, so I'm thinking it would be okay. Thoughts?

×
×
  • Create New...