d00phy

Waffle cut are a novelty. Crinkle cut are the best. Fight me.

I know it's CoM, who is basically Apple propaganda (forever living up to their name), but I've seen these iPhone 17 mock-ups before, and I really think they look like Pixels! Seriously, I have a Pixel 9 Pro that I use for work, and the design of the iPhone 17 is the definition of uninspired. Please, Tim, hire some good industrial designers. It's been a long boring stretch since Ive left.

cross-posted from: https://lemmy.world/post/31711492

cross-posted from: https://lemmy.world/post/31710629

I'm trying to boot some VMs using a script w/ a kickstart file. I'm using the following script that I found online and modified:

#!/usr/bin/env bash
#set -x
## Define variables
MEM_SIZE="8192"      # Memory setting in MiB
VCPUS="2"             # CPU Cores count
#OS_VARIANT="rocky9"   # List with osinfo-query  os
OS_VARIANT="rhel7.9"   # List with osinfo-query  os
ISO_FILE="~/Documents/software/os/RHEL-7.9-20200917.0-Server-x86_64-dvd1.iso" # Path to ISO file

case $OS_VARIANT in
        rhel7.9)
                KS=ks7.cfg;;
        rocky9)
                KS=ks9.cfg;;
esac

echo -en "Enter vm name: "
read VM_NAME
OS_TYPE="linux"
echo -en "Enter virtual disk size : "
read DISK_SIZE

DISK=~/.local/share/libvirt/images/${VM_NAME}.qcow2

echo "Creating disk"
sudo virt-install \
     --name ${VM_NAME} \
     --memory=${MEM_SIZE} \
     --vcpus=${VCPUS} \
     --location ${ISO_FILE} \
     --network network=default \
     --disk path=${DISK},size=${DISK_SIZE} \
     --graphics=none \
     --os-variant=${OS_VARIANT} \
     --console pty,target_type=serial \
     --initrd-inject ~/virt/${KS} --extra-args "inst.ks=file:/${KS} console=tty0 console=ttyS0,115200n8"

I've obfuscated the directory paths, but they're all full paths and the script will build a VM. So basically just setting up a basic system, using the default network. Here's the config for that:

<network connections='3'>
  <name>default</name>
  <uuid>61afc7f1-9c5e-4cra-8d18-e3cf4f9358e9</uuid>
  <forward mode='nat'>
    <nat>
      <port start='1024' end='65535'/>
    </nat>
  </forward>
  <bridge name='virbr0' stp='on' delay='0'/>
  <mac address='52:54:00:7c:32:9b'/>
  <ip address='192.168.122.1' netmask='255.255.255.0'>
    <dhcp>
      <range start='192.168.122.2' end='192.168.122.254'/>
    </dhcp>
  </ip>
</network>

Looking at the XML for the VM, I see the following for the network:

<interface type='network'>
      <mac address='52:54:00:07:82:78'/>
      <source network='default' portid='800dfd67-d90a-42te-a0b7-c4c78cdae481' bridge='virbr0'/>
      <target dev='vnet7'/>
      <model type='virtio'/>
      <alias name='net0'/>
      <address type='pci' domain='0x0000' bus='0x01' slot='0x00' function='0x0'/>
    </interface>

When this VM is installing, and when it's booted, it does not have an IP. Meanwhile, if I go through virt-manager and select the default network, it gets an IP just fine. I've tried running the virt-install command w/ and w/o sudo (I run virt-manager as me - I'm in the libvirt group). Looking at the virt-manager built VM:

    <interface type='network'>
      <mac address='52:54:00:5e:f5:05'/>
      <source network='default' portid='d57dbc56-759e-40f9-856f-9623f4801a93' bridge='virbr0'/>
      <target dev='vnet8'/>
      <model type='virtio'/>
      <alias name='net0'/>
      <address type='pci' domain='0x0000' bus='0x01' slot='0x00' function='0x0'/>
    </interface>

Looking at virbr0:

$ ip link show master virbr0
11: vnet7: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master virbr0 state UNKNOWN mode DEFAULT group default qlen 1000
    link/ether fe:54:00:07:82:78 brd ff:ff:ff:ff:ff:ff
12: vnet8: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master virbr0 state UNKNOWN mode DEFAULT group default qlen 1000
    link/ether fe:54:00:5e:f5:05 brd ff:ff:ff:ff:ff:ff

Only difference I can see is one is created using virt-manager and the other using virt-install (which calls to virt-install, no?). I thought there was a way to see the actual virt-install command virt-manager was about to use when creating a VM, but I can't find it. Also can't find any logs to give me an idea why the VM isn't getting an IP. Running ethtool on the VM interface shows a link. I've wasted too much time getting this to work, and all the documentation suggests it should "just work!"

I'm trying to boot some VMs using a script w/ a kickstart file. I'm using the following script that I found online and modified:

#!/usr/bin/env bash
#set -x
## Define variables
MEM_SIZE="8192"      # Memory setting in MiB
VCPUS="2"             # CPU Cores count
#OS_VARIANT="rocky9"   # List with osinfo-query  os
OS_VARIANT="rhel7.9"   # List with osinfo-query  os
ISO_FILE="~/Documents/software/os/RHEL-7.9-20200917.0-Server-x86_64-dvd1.iso" # Path to ISO file

case $OS_VARIANT in
        rhel7.9)
                KS=ks7.cfg;;
        rocky9)
                KS=ks9.cfg;;
esac

echo -en "Enter vm name: "
read VM_NAME
OS_TYPE="linux"
echo -en "Enter virtual disk size : "
read DISK_SIZE

DISK=~/.local/share/libvirt/images/${VM_NAME}.qcow2

echo "Creating disk"
sudo virt-install \
     --name ${VM_NAME} \
     --memory=${MEM_SIZE} \
     --vcpus=${VCPUS} \
     --location ${ISO_FILE} \
     --network network=default \
     --disk path=${DISK},size=${DISK_SIZE} \
     --graphics=none \
     --os-variant=${OS_VARIANT} \
     --console pty,target_type=serial \
     --initrd-inject ~/virt/${KS} --extra-args "inst.ks=file:/${KS} console=tty0 console=ttyS0,115200n8"

I've obfuscated the directory paths, but they're all full paths and the script will build a VM. So basically just setting up a basic system, using the default network. Here's the config for that:

<network connections='3'>
  <name>default</name>
  <uuid>61afc7f1-9c5e-4cra-8d18-e3cf4f9358e9</uuid>
  <forward mode='nat'>
    <nat>
      <port start='1024' end='65535'/>
    </nat>
  </forward>
  <bridge name='virbr0' stp='on' delay='0'/>
  <mac address='52:54:00:7c:32:9b'/>
  <ip address='192.168.122.1' netmask='255.255.255.0'>
    <dhcp>
      <range start='192.168.122.2' end='192.168.122.254'/>
    </dhcp>
  </ip>
</network>

Looking at the XML for the VM, I see the following for the network:

<interface type='network'>
      <mac address='52:54:00:07:82:78'/>
      <source network='default' portid='800dfd67-d90a-42te-a0b7-c4c78cdae481' bridge='virbr0'/>
      <target dev='vnet7'/>
      <model type='virtio'/>
      <alias name='net0'/>
      <address type='pci' domain='0x0000' bus='0x01' slot='0x00' function='0x0'/>
    </interface>

When this VM is installing, and when it's booted, it does not have an IP. Meanwhile, if I go through virt-manager and select the default network, it gets an IP just fine. I've tried running the virt-install command w/ and w/o sudo (I run virt-manager as me - I'm in the libvirt group). Looking at the virt-manager built VM:

    <interface type='network'>
      <mac address='52:54:00:5e:f5:05'/>
      <source network='default' portid='d57dbc56-759e-40f9-856f-9623f4801a93' bridge='virbr0'/>
      <target dev='vnet8'/>
      <model type='virtio'/>
      <alias name='net0'/>
      <address type='pci' domain='0x0000' bus='0x01' slot='0x00' function='0x0'/>
    </interface>

Looking at virbr0:

$ ip link show master virbr0
11: vnet7: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master virbr0 state UNKNOWN mode DEFAULT group default qlen 1000
    link/ether fe:54:00:07:82:78 brd ff:ff:ff:ff:ff:ff
12: vnet8: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master virbr0 state UNKNOWN mode DEFAULT group default qlen 1000
    link/ether fe:54:00:5e:f5:05 brd ff:ff:ff:ff:ff:ff

Only difference I can see is one is created using virt-install and the other using virt-manager (which calls to virt-install, no?). I thought there was a way to see the actual virt-install command virt-manager was about to use when creating a VM, but I can't find it. Also can't find any logs to give me an idea why the VM isn't getting an IP. Running ethtool on the VM interface shows a link. I've wasted too much time getting this to work, and all the documentation suggests it should "just work!"

ETA: PEBKAC! Remember where I said I was using kickstart? Yeah, well it helps to actually include a network configuration, even if it is just DHCP! Feeling stupid, but it's working.

Current setup is PMS running on a Synology 5-bay, and another PMS running on a Shield Pro. The NAS server is primarily used for remote streaming, while the Shield serves to my home LAN (AppleTVs mainly).

I've been seeing stuttering on larger files, either using the Plex app or Infuse, and I'm fairly certain the Synology is the weak link. Network performance in the house has pretty solid, though admittedly I could stand to test it more thoroughly. I've been looking at moving my library to a standalone system. I've been looking at the Beelink ME Mini (which happens to be on sale!). What I don't know is the best way to build this out.

I don't want to have to buy all 6 SSDs (ar at least 6x4TB ones!) at once, so I'd be looking at either a stock Linux (Ubuntu or Rocky) install w/ I guess a BTRFS pool for the SSDs (I'm guessing I can use the eMMC for OS depending on how big the install is - that or use the SSD in slot 4). Alternatively, i could possibly set up TruNAS w/ the Plex pp to manage the storage.

As for populating the media, I plan to keep the Synology as the central repo of my data. I have it replicating to another NAS at my dad's house, with movies/music/tv replicating using Syncthing. I plan to also use Syncthing to populate the Beelink.

Anyway, please poke holes in this plan and/or suggest a better one. My main goals are to get the media I'm streaming off spinning disk w/ minimal power draw (didn't mention that above) in a way that I can expand storage as necessary to accommodate the media library. Nothing's purchased yet, so I'm not married to the hardware. I would ideally like to convert the library to h.265 or even AV1 if I can make it work.

ETA: For clarity: I'm not transcoding AFAIK. My Shield mounts the Synology over SMB and mostly works fine, until I try to play anything 4k - then I get stuttering. On the surface, this sounded like a network issue, but I can't find a problem w/ the LAN. My thought was to move the PMS to a single location w/ local storage, and use the Synology just as an archive.

ETA2: FWIW, I have not expanded the memory on the Synology or installed any cache drives.

I want to start by saying I recognize that everyone's needs & priorities are different.

My wife and I both have iPhones, and i have a Pixel 7 Pro I use for work (and sometimes to compare the camera to the iPhones). All of our photos are currently backed up to iCloud (Apple One Premier - 2TB storage) and via Synology Photos. The Pixel has "unlimited" storage for photo backup w/ Google, and also backs up to the Synology. In general, I would like to get off of Google, but it's 99% work stuff that I wouldn't miss if it was lost.

There's a lot that I really like about Immich, but there are also some real pain points for me. I'm not going to comment on the discrepancies between the mobile vs. web interfaces as I expect them to be addressed as the product matures.

• The rapid development is both a blessing and a curse. I love that the team are really working through the roadmap. But sometimes it feels like new features arrive somewhat half-baked. The most common example being something is released working on just the web or mobile app. But the pace also creates extra work for me in that every release requires me to look for breaking changes and make appropriate fixes. I get it, it's beta software, and heavy development often requires this.
• If it mis-identifies a face, the mechanism for correcting that is pretty clunky. I have to first, say it's a different person, and then, if I don't care about tagging that face, I have to go to People to hide it. I don't really care about faces that it completely misses because I don't consider facial recognition as a "archive-grade" feature. We have tags/keywords for that.
• The tagging is both cool and clunky. I love the nested tags and the drill-down tags interface. I hate that I can only add a new tag from the tags admin page. Would also like to see auto-tagging, or suggested tags implemented.
• Image rotation is half-addressed at best. For one, I'm not sure why it only works on the mobile interface since the web interface has direct access to ImageMagick. I mainly see image orientation issues w/ raw files. To fix this, I have to edit it on mobile, save it to my phone's library and upload the newly created JPG, which shows up as a separate file w/ metadata that doesn't align w/ the original (like creation date). It's just a mess.

I started playing with PhotoPrism a little bit, and while it addresses many of my complaints w/ Immich, it also raises some of its own pain points.

• Probably the biggest issue I have with PhotoPrism is the lack of mobile apps. There are some out there, but the recommended app is a third-party WebDav app called PhotoSync. I tried it and wasn't overly impressed. At least, not enough to pay for it. This would be a dealbreaker except that I can simply use the Synology Photo backup, and have PhotoPrism mount those directories as its library ( can also do this with Immich's "External Library" feature).
• The metadata editing is comprehensive. In this one regard it is streets ahead of Immich. Seriously, you have so much more access to the photo metadata. Unfortunately, it's hampered by the limited batch capabilities.
• Batch editing isn't really batch editing. It's just editing a smaller subset of individual files one at a time. So when go to to the next or previous file, it the next or previous one in the selected subset.
• Keywords are supports, and new ones can be created on the fly. That said, nested keywords don't appear to work.
• There's also labels. Both are auto-suggested, and both can be manually edited. Labels are also accessible from the sidebar. No nested labels, either, but it does auto sort labels into broad categories. For example, "dog" and "cat" are placed into an "animals" category. You can switch between showing/hiding the broad categories. You can also have favorite labels.
• Image orientation/rotation is done right in the photo editing dialog. One more area where PP beats Immich.

I currently haven't decided which one I will keep. I could use either with the Synology Photo app to back up my phones. PhotoPrism's lack of mobile app is really bad, but the mobile web interface is fine for navigating the library. Immich is a more wholistic solution, but it's handling of some key organizational and editing functions is pretty glaring as well. I know Immich is the overwhelming favorite of most self-hosting communities, but I found PhotoPrism to be pretty compelling in its own right - especially the metadata editing capabilities.

ETA: I see lots of people talking about Immich’s facial detection. Out of curiosity, what are your detection settings? I’ve found it to be pretty good compared to Photo Prism’s, but not exactly game changing. My settings are:

• Model: antelopeV2
• Min Score: 0.2
• Max distance: 0.5 Min recognized faces: 1

Trying to do a couple things. I have 2 jump hosts I can use to get into my cluster login node. From my laptop to the jump hosts is password. From jump hosts to login node can be key-based, so if I do it all from CLI:

[me@home ~]$ ssh user@jump1
Password:
[user@jump1 ~]$ ssh user@login1
[user@login1 ~]$ 

Same process if I use jump2.

So first thing I'm trying to do is set up my ~/.ssh/config to use the ProxyJump host and key file to get to login1. I have the following:

Host jump1
  Hostname jump1.domain
Host jump2
  Hostname jump2.domain
Host login1
  Hostname login1.cluster
  ProxyJump jump1
  #ProxyJump jump2

I'm not sure how to configure the IdentityFile entries for each jump host. The user on the jump hosts has different id_rsa keys in ~/.ssh, but both are in the authorized_keys file on login1.

Second thing I'm trying to do is join or start a tmux session. From CLI, I can run:

tmux has-session -t mysession || tmux new -s mysession && tmux a -t mysession

I've learned that to just join a running session (tmux a -t mysession), I need to include "RequestTTY yes" in my ssh config entry for login1. What I can't get working is the conditional statement that will fire up a new tmux session if it doesn't already exist.