AFP on OMV3 crashing all system

1. offizieller Beitrag

    Hi,


    I'm trying to setup my Time Machine with OMV 3.0.64 and afp plugin.


    Time machine is detecting my drive on OMV but when i start the first backup, some Mb are transferred then the system crashes and reboots.


    Any idea of what i could do for that ?


    Thanx a lot !

    • Offizieller Beitrag

    Do you mean the output during the boot? Never needed to do that. Always looked at the VM's boot.log. Maybe this will help - https://kb.vmware.com/selfserv…playKC&externalId=1030667

    omv 7.0.5-1 sandworm | 64 bit | 6.8 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.13 | compose 7.1.4 | k8s 7.1.0-3 | cputemp 7.0.1 | mergerfs 7.0.4


    omv-extras.org plugins source code and issue tracker - github - changelogs


    Please try ctrl-shift-R and read this before posting a question.

    Please put your OMV system details in your signature.
    Please don't PM for support... Too many PMs!

    • Offizieller Beitrag

    Oh, I thought you meant a text file based log. If you login into the ESXi box with the web client http://esxi_ip_address/ui, you can click on the VM and get a realtime console. If you have vCenter, you can login to vCenter and do the same thing. ESXi 6.5 only uses the web client instead of the pre-6.0 versions that needed the fat client.

    omv 7.0.5-1 sandworm | 64 bit | 6.8 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.13 | compose 7.1.4 | k8s 7.1.0-3 | cputemp 7.0.1 | mergerfs 7.0.4


    omv-extras.org plugins source code and issue tracker - github - changelogs


    Please try ctrl-shift-R and read this before posting a question.

    Please put your OMV system details in your signature.
    Please don't PM for support... Too many PMs!

    Ok so i tried to check the console when crash is happening to see what's going on and... nothing. The system reboots without saying anything.


    I can reproduce the crash just by selecting the AFP disk in time machine settings...

    Ok i solved my problem.


    I think OMV3 images are not fully compatible with ESXi 6.5. ESXi says it doesn't recognize OS on the image. At boot i have an error message saying something about cpufreq failed to initialize policy (sorry i don't remember the exact message).


    So i did a clean install by installing Debian jessie first then OMV3 on top of it. Then installed AFP and everything is working fine now ;)


    Thanx for your help anyway !

    • Offizieller Beitrag
    Zitat von jaydee99

    I think OMV3 images are not fully compatible with ESXi 6.5. ESXi says it doesn't recognize OS on the image. At boot i have an error message saying something about cpufreq failed to initialize policy (sorry i don't remember the exact message).

    I run almost all of my OMV images on ESXi 6.5 and don't have any issues. If you have your image set to Debian (which OMV is), it will say it isn't Debian because it is saying it is openmediavault. You need to change it to "Other 3.x and later Linux 64 bit". Even when I had it set Debian 8 64 bit, I didn't have any issues though.

    omv 7.0.5-1 sandworm | 64 bit | 6.8 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.13 | compose 7.1.4 | k8s 7.1.0-3 | cputemp 7.0.1 | mergerfs 7.0.4


    omv-extras.org plugins source code and issue tracker - github - changelogs


    Please try ctrl-shift-R and read this before posting a question.

    Please put your OMV system details in your signature.
    Please don't PM for support... Too many PMs!

    I have the same problem.


    ESXi 6.5 VM with RAM 4Gb
    OMV 3 (3.0.83)
    Netatalk 3.1.9-1
    MAC OS X El Capitan


    When Finder.app list content on mounted AFP share VM is crash.

    Hi,


    I'm having the same problem:


    ESX 6.5 HPE 650.9.6.5.27 (MicroServer Gen8, P420 Raid 10)
    openmediavault 3.0.84 (Erasmus)
    openmediavault-netatalk 3.2.10
    OS X Sierra 10.12.6


    The shares are working over CIFS without any problem. Copied 14GB minutes ago.
    But if I use the same shares via AFP and try to list a directory or acces files, the whole VM reboots. The share's root directory is shown, though.


    The syslog has no sign for anything that went wrong, as you can see here at 22:51:59:


    Any other hints for hunting down this problem?


    All other VMs on this ESXi (mainly Ununtu) are working without any flaws. I've already changed the OMV VM to "Other Linux > 3.x 64bit". On other machines (bare metal and VMs) I have an Ubuntu 16.04 with netatalk 2.2.5-1 installed which works without any problem.


    Cheers,
    Patrik



    //edit:


    found this in the daemon.log "^@^@^@^@^":

    • Offizieller Beitrag

    I'll see if I can rebuild the package and post here during weekend. Otherwise you can do try yourself. Google Adrian knoth netatalk repo in github. He has a debianized source repo of netatalk usually in sync with latest.
    Also as @tkaiser mentioned is not a good idea to use the same share in between samba and netatalk as it can lead to data corruption.

    Zitat von subzero79

    Also as @tkaiser mentioned is not a good idea to use the same share in between samba and netatalk as it can lead to data corruption.

    That's simply asking for trouble unless you have a specific (and way more recent) Samba version than included in OMV which needs also some modules stacked in the right order: https://bugs.freenas.org/issues/5904


    Otherwise all sorts of encoding and metadata trouble can occur when using the same share with both Netatalk and Samba. Using SMB with different OS X / macOS clients can also be challenging since Apple changed the way metadata and resource forks are represented (IIRC with 10.7). Before they used a single AppleDouble file lying next to the data file with a name prefix '._' containing this stuff, later they switched to so called 'alternate data streams'. In other words: If you access a SMB share with OS X 10.6 and another machine with a more recent OS X version this is also asking for troubles since metadate / resource forks are stored in different ways.


    Rule of thumb: Avoid SMB on OMV when using macOS clients.


    @subzero79: on a related note: Currently the openmediavault-netatalk plugin doesn't make use of the 'vol dbpath' option. All CNID databases are stored below /var/lib/netatalk/CNID/ by default. If the rootfs is on a flash media (USB thumb drive or SD card or eMMC) then this both increases wear out and slows things down. Setting this


    Code
    xmlstarlet ed -L -u "/config/services/afp/extraoptions" -v "vol dbpath = /var/log/netatalk/CNID/%v/"

    would probably work together with flashmemory-plugin but it would've to be set as part of netatalk installation and also it's no clean solution since /var/log/ is for logs by default (and users might move it to another filesystem by intention, eg. one with btrfs and highest compression).


    Two variants come to my mind:

    • Extending openmediavault-netatalk plugin in a way it makes use of the 'vol dbpath' option on a per volume basis (letting the user choose a volume path but then create two subdirs 'data' and 'cnid' and setting path / vol dbpath accordingly -- this will make users angry who want to share already existing volumes with Netatalk too which is IMO a freature and not a bug but that's obviously just me
    • @ryecoaaron adds simply the path to flashmemory-plugin by default (as I do it currently with the OMV ARM images since random IO performance is magnitudes faster when CNID databases are handled by folder2ram)

    What do you guys think?

Jetzt mitmachen!

Sie haben noch kein Benutzerkonto auf unserer Seite? Registrieren Sie sich kostenlos und nehmen Sie an unserer Community teil!

Benutzerkonto erstellen Anmelden