Beiträge von ManniX

    ryecoaaron


    Bleeding edge hardware is a pain to configure and maintain but sometimes is rewarding :)


    I could get the idle power consumption (AC) down to 42-45 Watts.

    And I'm still missing all the goodies from the amd-pstate driver.


    Bet there's no Intel configuration that can beat this cost effective combo on power consumption/performances/upgradability.



    Since I've also both an USB HDD and a USB 4xHDD external enclosures I had to scout for something that could put the HDDs in standby.

    Ended up reading that hd-idle was able to.

    Indeed it works like a charm for USB disks.


    Then I realized my 4 x Toshiba 18TB drives were able to sleep set with Standby in OMV UI but not the other 4 (2 x Toshiba and 2 x Seagate), all SATA internal.

    At this point I could not get below 54W idle with some very sporadic dips down to 44W.


    I've also found that querying the temperature with hddtemp would wake up briefly the MG09A disks.


    Did a try enabling hd-idle also for the SATA drives and indeed it's much better.


    Not only the standby power consumption went down by 10 Watts but they all go in standby and the MG09A drives are not wakeup by the hddtemp query.

    At the same time the temperatures via the lm_sensors drivetemp module can be read without issue while they are sleeping.


    For the temperatures is a godsend, gained 3-4 °C on the MG09A and almost 10 °C on the others (they finally sleep)



    Before:



    Is there any reason why OMV doesn't use hd-idle?

    Incorrect. Look at /etc/netplan/10-openmediavault-default.yaml and you will see that netplan is using networkd for its backend. netplan doesn't actually configure networking itself. It can work with network-manager as well. OMV just chooses to use networkd.


    Yes, I noticed it later.


    What is the output of: dpkg -l | grep dhcp

    ii dhcpcd5 7.1.0-2+b1 amd64 DHCPv4, IPv6RA and DHCPv6 client with IPv4LL support

    ii isc-dhcp-client 4.4.1-2.3 amd64 DHCP client for automatically obtaining an IP address

    ii isc-dhcp-common 4.4.1-2.3 amd64 common manpages relevant to all of the isc-dhcp packages

    As a workaround I've disabled the dhcpcd daemon.

    Couldn't find what is controlling it.


    Unfortunately I've also had a look at the speed of the network backups and found out they were all much slower.


    The Marvell NIC is not working properly with Bullseye.

    Bandwidth with iperf from 2.35 Gbps is down to 1.8 Gbps.


    I'm going to backup this installation and go back to Debian 10 / OMV 5.

    Not sure if I have another issue or it's really Bullseye.

    From what I'm reading online seems Debian 11 should have the systemd-networkd services masked.

    Which is not the case for me.

    Could be it's conflicting with netplan? or netplan is using it?


    Output of networkctl:


    IDX LINK TYPE OPERATIONAL SETUP

    1 lo loopback carrier unmanaged

    2 enp5s0 ether routable configured

    3 enp4s0 ether routable configured

    4 docker0 bridge routable unmanaged

    6 veth8e8199f ether routable unmanaged

    8 veth1ad689d ether routable unmanaged

    10 veth174d2b7 ether routable unmanaged

    12 veth5f4289b ether routable unmanaged

    14 veth902ace7 ether routable unmanaged

    16 veth093e921 ether routable unmanaged


    10 links listed.


    systemctl status systemd-networkd.socket

    ● systemd-networkd.socket - Network Service Netlink Socket

    Loaded: loaded (/lib/systemd/system/systemd-networkd.socket; disabled; vendor preset: enabled)

    Active: active (running) since Mon 2022-08-15 08:56:38 CEST; 1h 6min ago

    Triggers: ● systemd-networkd.service

    Docs: man:systemd-networkd.service(8)

    man:rtnetlink(7)

    Listen: route 1361 (Netlink)

    CGroup: /system.slice/systemd-networkd.socket


    status systemd-networkd.service

    ● systemd-networkd.service - Network Service

    Loaded: loaded (/lib/systemd/system/systemd-networkd.service; enabled-runtime; vendor preset: enabled)

    Active: active (running) since Mon 2022-08-15 08:56:40 CEST; 1h 1min ago

    TriggeredBy: ● systemd-networkd.socket

    Docs: man:systemd-networkd.service(8)

    Main PID: 9890 (systemd-network)

    Status: "Processing requests..."

    Tasks: 1 (limit: 18363)

    Memory: 4.1M

    CPU: 626ms

    CGroup: /system.slice/systemd-networkd.service

    └─9890 /lib/systemd/systemd-networkd


    Aug 15 08:56:47 solidpc systemd-networkd[9890]: veth1ad689d: Gained carrier

    Aug 15 08:56:47 solidpc systemd-networkd[9890]: veth1321f05: Link DOWN

    Aug 15 08:56:47 solidpc systemd-networkd[9890]: veth1321f05: Lost carrier

    Aug 15 08:56:47 solidpc systemd-networkd[9890]: veth902ace7: Lost carrier

    Aug 15 08:56:47 solidpc systemd-networkd[9890]: veth902ace7: Gained carrier

    Aug 15 08:56:47 solidpc systemd-networkd[9890]: veth753ad81: Link DOWN

    Aug 15 08:56:47 solidpc systemd-networkd[9890]: veth753ad81: Lost carrier

    Aug 15 08:56:47 solidpc systemd-networkd[9890]: veth093e921: Lost carrier

    Aug 15 08:56:47 solidpc systemd-networkd[9890]: veth093e921: Gained carrier

    Aug 15 08:56:51 solidpc systemd-networkd[9890]: enp4s0: Gained carrier


    systemctl status systemd-networkd-wait-online.service

    ● systemd-networkd-wait-online.service - Wait for Network to be Configured

    Loaded: loaded (/lib/systemd/system/systemd-networkd-wait-online.service; enabled-runtime; vendor preset: disabled)

    Active: active (exited) since Mon 2022-08-15 08:56:45 CEST; 1h 8min ago

    Docs: man:systemd-networkd-wait-online.service(8)

    Process: 9903 ExecStart=/lib/systemd/systemd-networkd-wait-online (code=exited, status=0/SUCCESS)

    Main PID: 9903 (code=exited, status=0/SUCCESS)

    CPU: 27ms


    Aug 15 08:56:40 solidpc systemd[1]: Starting Wait for Network to be Configured...

    Aug 15 08:56:45 solidpc systemd[1]: Finished Wait for Network to be Configured.

    I would guess /etc/network/interfaces is configured? I would remove the file. You could even remove the ifupdown package. Make sure the only netplan files in /etc/netplan/ are the ones created by OMV.


    Removed interfaces and was re-created by OMV, it's empty only referencing interfaces.d which does not exist


    In /etc/netplan only the OMV files:

    -rw-r--r-- 1 root root 43 Aug 15 05:39 10-openmediavault-default.yaml

    -rw-r--r-- 1 root root 328 Aug 15 05:39 20-openmediavault-enp4s0.yaml

    -rw-r--r-- 1 root root 186 Aug 15 05:39 20-openmediavault-enp5s0.yaml


    Which are looking fine, as configured in the UI


    I have removed ifupdown package as well


    But still no luck


    Is it normal that the networking service is masked?


    ● networking.service

    Loaded: masked (Reason: Unit networking.service is masked.)

    Active: inactive (dead)

    Since you had a desktop environment installed, there probably another source of network config like network-manager. I would:

    Indeed there was. It also removed the Proxmox 5.15 for whaterver reason.


    Sadly the only outcome is that it enabled ipv6 at reboot.

    Re-applying the network interfaces config from the WebUI removed it.

    Disabling ipv4 on the 2nd NIC left the dhcp address active.


    Another suggestion for the offending package to remove?

    I would call that bleeding edge hardware in the server world.

    Indeed.

    Not a big issue, it's manageable at the end.


    Seems to work really well so far!


    Didn't get again the error 504 so it was probably the kernel.


    I have one last issue that I'm struggling to fix...


    After the upgrade the NICs got a couple of IPs in dhcp.

    And it doesn't seem I can't remove them...


    In OMV GUI they are set with 2 static IPs.

    And also ifconfig shows only the 2 static IPS;


    enp4s0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500

    inet 192.168.178.2 netmask 255.255.255.0 broadcast 192.168.178.255

    ether 24:5e:be:65:4f:b4 txqueuelen 1000 (Ethernet)

    RX packets 58653 bytes 33821235 (32.2 MiB)

    RX errors 0 dropped 0 overruns 0 frame 0

    TX packets 9400 bytes 949525 (927.2 KiB)

    TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0


    enp5s0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500

    inet 192.168.178.3 netmask 255.255.255.0 broadcast 192.168.178.255

    ether a8:a1:59:58:1f:05 txqueuelen 1000 (Ethernet)

    RX packets 109262 bytes 19987541 (19.0 MiB)

    RX errors 0 dropped 0 overruns 0 frame 0

    TX packets 242624 bytes 314669182 (300.0 MiB)

    TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0

    device memory 0xfc600000-fc61ffff


    I thought they were removed but then the UrBackup client connected to the server using the ip 192.168.178.44


    And indeed with ip address:


    2: enp5s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000

    link/ether a8:a1:59:58:1f:05 brd ff:ff:ff:ff:ff:ff

    inet 192.168.178.3/24 brd 192.168.178.255 scope global enp5s0

    valid_lft forever preferred_lft forever

    inet 192.168.178.44/24 brd 192.168.178.255 scope global secondary noprefixroute enp5s0

    valid_lft forever preferred_lft forever

    3: enp4s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000

    link/ether 24:5e:be:65:4f:b4 brd ff:ff:ff:ff:ff:ff

    inet 192.168.178.2/24 brd 192.168.178.255 scope global enp4s0

    valid_lft forever preferred_lft forever

    inet 192.168.178.54/24 brd 192.168.178.255 scope global secondary noprefixroute enp4s0

    valid_lft forever preferred_lft forever


    Not sure how I can get out of this :P

    I'm guessing that is caused by your disk errors. Otherwise, what kind of media is the OS installed on?

    I think I got it also while running the 5.11 Kernel but now I'm not so so sure...

    Let me see if it happens again.


    The system is running on another disk, M.2 PCIe.

    That was my new media array, 50TB Raid5.

    But there's no real disk error... I had another 6TB Raid5 on those ports and I had the same errors with the same port with another Disk and other SATA cables...

    It's a weird effect of some obscure bug in the kernel...

    The vanilla 5.10 and Proxmox 5.11 are running just perfect.

    I had to boot a couple dozen times with Proxmox 4.x kernels to find one that would work...

    yep. To fix: RE: OMV 5 to OMV 6 upgrade issue

    Fantastic...


    The most resilient upgrade script I've ever seen :P

    I never thought it could end well.


    I get sometimes a 504 error:



    But then retrying the same action works fine.

    Should I be worried?

    Don't see any messages in daemon or syslog


    Unfortunately the Proxmox 5.15 kernel is unstable for me :(


    Like many earlier it has some issue with the AMD SATA controller:



    Which results in a single port out of 8 having issues...


    And also has some issue with the Marvell NIC driver:

    Code
    [   10.596675] UBSAN: array-index-out-of-bounds in drivers/net/ethernet/aquantia/atlantic/aq_nic.c:1262:48
    [   10.596691] index 8 is out of range for type 'aq_vec_s *[8]'

    omv-firstaid did not help

    the fix5to6upgrade script did help


    what did the magic at the end was


    sudo omv-salt stage run deploy


    now I'm the WebUI and seems to work pretty well...


    I'm installing the Proxmox 5.15 Kernel right now!

    (the update did not update the PGP for the bullseye rep, I had to do it manually)


    The only stain is omv-aptclean:


    8. Updating plugin index...

    /usr/sbin/omv-aptclean: 50: /usr/sbin/omv-mkaptidx: not found


    Is this ab-normal?


    What is the output of: dpkg -l | grep openme

    I managed to continue the upgrade with fix broken dependencies.

    I've also removed all desktop environments, purging xfce also removed something from openmediavault.


    Now it's done but I get 502 Bad Gateway when I try to login.


    Running the fix script ends with:



    I'm trying to understand what is still missing.

    I had to unmask with systemctl the openmediavault-* services.

    Desktop environments can be uninstalled...

    Is there an optimal way to do it?

    I tried apt remove task-desktop but I feel is going to remove too much stuff. especially python modules...


    The openmediavault package is removed. You might be able to reinstall it but you have quite the mess. sudo apt-get install openmediavault Running the fix script won't hurt either.

    I'll try, otherwise I'll restore from the backup.


    With the fix script you mean omv-firstaid?

    It looks like OMV was uninstalled. What is the output of: dpkg -l | grep openme


    Should you remove something else? yes. Do you have to? no but it still recommended to not run a desktop environment. Having the desktop environment installed just adds hundreds if not thousands more packages that have to be upgrade that could cause problems. The display manager checks are just the easiest way to tell if a desktop is installed. Removing them gets by the check but doesn't fix the problem on why we tell people not to have a desktop environment installed.

    Sorry, the desktop environment came earlier than OMV on this machine :P

    Personally didn't have any issue with 4.x and 5.x but I guess with 6.x can be different.

    Usually that happens when OMV is removed before the plugin. sudo rm /var/lib/dpkg/info/openmediavault-backup.postrm will fix the error. But it looks like the system had a desktop environment installed which probably broke the upgrade.

    I removed the desktop manager packages that the scripts checks for, should I remove something else as well?


    Is there a way to restart the upgrade from there?

    The script is gone now together with the OMV 5 install.

    There were some other warnings and errors before:



    netplan maybe?

    Thanks, I'm blind...


    And of course at the end I forgot to check the plugins ;(

    Too hot, not enough sleep....


    Before I make more disasters here's how it ended up like this:


    Code
    Removing openmediavault-backup (5.2.4) ...
    /var/lib/dpkg/info/openmediavault-backup.postrm: 6: .: Can't open /usr/share/openmediavault/scripts/helper-functions
    dpkg: error processing package openmediavault-backup (--remove): installed openmediavault-backup package post-removal script subprocess returned error exit status 127
    Removing borgbackup (1.1.9-2+deb10u1) ...
    Removing python3-msgpack (0.5.6-1+b1) ...
    Removing python3-llfuse (1.3.6+dfsg-1) ...
    Errors were encountered while processing: openmediavault-backup
    W: --force-yes is deprecated, use one of the options starting with --allow instead.
    E: Sub-process /usr/bin/dpkg returned an error code (1)


    I guess that's not how it was supposed to end.


    Is there something I can do to fix it or should I restore from the backup and start again?