tjb_altf4

Members
  • Posts

    1373
  • Joined

  • Last visited

  • Days Won

    1

tjb_altf4 last won the day on November 4 2018

tjb_altf4 had the most liked content!

Retained

  • Member Title
    not great, not terrible

Converted

  • Gender
    Undisclosed

Recent Profile Visitors

6816 profile views

tjb_altf4's Achievements

Community Regular

Community Regular (8/14)

377

Reputation

9

Community Answers

  1. Yes you need to use macvlan if you want unique mac address per container, otherwise they will share the nic's mac (ipvlan setting). The caveat being macvlan historically has caused some instability in some systems, although that seems to be largely fixed in 6.12+
  2. Asked for the same thing (and a little more) 3 years ago, and still think its a good idea
  3. I approach this by configuring a DHCP reservation in the router, and leaving the OS in DHCP mode. Note that you can get the MAC address needed for the DHCP reservation from the VM template.
  4. Just don't get an infiniband only card, QSFP+ is just a connector/cable and can be used for either protocol. This one does both modes, you may to change modes to use when setting up. This one is ethernet only. Note the final character of the model (T or S) means the bracket height (Tall or Short)
  5. There are SKUs of the mellonox cards that are infiniband only, but don't think there are any in the MCX353A line up. You can confirm support in the product manual https://network.nvidia.com/pdf/user_manuals/ConnectX-3_VPI_Single_and_Dual_QSFP_Port_Adapter_Card_User_Manual.pdf MCX353A working in Jerod's video below (after some troubleshooting)
  6. Upgraded from 7.3.83 to 7.4.162 today, couldn't get WireGuard working under the old version, but new version worked straight away. For anyone else using this version, do the QR code scans on your device? My pixel4a could not scan, even tried adjusting the contrast ratio in a photo editor.
  7. Just went through this myself, solution is to change permissions from read-only for group/other to R/W:
  8. I don't mind being challenged on the setup, this was all whipped together over a few days, even though I wanted to do it for ages The M.2 is a 250GB NVME that came with the unit from auction, its a freebie I wouldn't of bought for this but it'll get used for appdata etc for convenience + separation of roles, the pool name "cache" is the default first pool name and I've haven't sought fit to change it yet. I was able to setup what I wanted out of Unraid in about 5min, but I'm not afraid to nuke and pave this to get the format right. The targets for this server will be backups of vms, appdata, user docs, photos, music etc... basically all the important non-video media from my main servers. A mirrored 4TB will be plenty for now, and its really just a +1 safeguard given existing data is already mirrored and/or parity protected on other respective servers. Once first backup is done, I'm not expecting the backup deltas to use a lot of bandwidth... and if it does I won't notice anyway. Main array is a spare thumbdrive, next release should remove the requirement for it to exist. In release 6.12 you can customise the visible panels, I've simply hidden the array + parity panels (see below).
  9. Final ingredient was small + thin sata cables, which came today! Took the time while waiting to reprint drive carrier and riser/spacer in ASA black filament Happily running on 6.12.4 at the moment Looking forward to testing out subvolumes and snapshots plugin as part of this backup server journey
  10. A project server that I've been meaning to do for some time is a small backup server, able to be located seperately from my main rack. I came across the M710q at local auctions and eventually picked one up for under AUD$100. Out of the box it is well specced, with one m.2 slot and one 2.5" drive... but I really wanted 2x 2.5" so I could run a raid1 pool for backups. Enter a slew of additional purchases, modifying a sata splitter, and a couple hours on the 3D printer to prototype a solution:
  11. Can confirm this was impacting me on 6.11.5, and the Connect update resolved the issue.
  12. sounds like a browser cache issue, try clearing it out.
  13. Spotted this parsing error at the end of start up for some older drives in a (non-critical) pool, doesn't seem to have impacted server operation.