OMV Server stops working after some hours uptime

1. offizieller Beitrag

    I have recently installed OMV and am now experiencing an annoying problem. When the sever has been running for some time (typically less than 24 hours) it suddenly stops working. When this has happened I can no longer access the Web GUI (server typically throws a 403 authentification error), also SSH does not let me connect and SMB shares are visible but no longer possible to connect to. I have googled for answers to the problem without success. I read somewhere that a failing usb stick for the OS might cause problems like this, but mine is a brand new sandisk cruzer blade, which it seems as others are using without problems. Another post suggested that insufficient capacity of the OS disk could make it impossible to login through the web GUI but this is not my problem either (df -h says that only ~800 MB of 4 GB is used so far).


    As a quick fix I tried to enable the auto reboot feature but it seems as the server does not reboot when it has entered this state. The only remedy I have found is to cut the power and reboot the server. I have no idea what could be causing the problem and was hoping for some guidance how to further troubleshoot.


    Below is some short info about my installation


    Hardware:
    MB:Intel DN2700MT (integrated Atom CPU)
    RAM: 8 GB
    HDD: 2x WD 1TB (WD10EZRX) in RAID mirror
    OS Disk: Sandisk Cruzer Blade 4 GB


    Software:
    OMV Version: 0.3.0.17 (Omnius)
    Kernel: Linux 2.6.36-5-amd64
    Plugins: no additional plugins installed
    Services running: SMB/CIFS, SSH, NFS


    Any help how to troubkeshoot or fix this problem would be highly appreciated. I would really like to avoid to reinstall OMV since I am worried for the data on my raid disks and the OMV configuration data.


    Many thanks.
    /Jonas

    • Offizieller Beitrag

    Using the CLI is still possible? If the system is completely frozen the watchdog should reboot it after a given timeout. Hmmm. Is the filesystem set to read-only? Some useful informations in syslog?

    Hi!


    Generally I would not recommend to run a NAS from a Flash-stick unless it is specially optimized for that.


    OVM is running on a standard debian distribution, and this IS going to kill the stick over time, but for sure not within 24 Hours.


    My first guess would be that something was filling up your OMV-partition - like some logfile, or that someone tried to write data to a device which should have been mounted but was not, and now all the data is copied in the mountpoint-folder.


    This leads to very strange behaviour of linux.
    This is easy to check: You need to do a df when the system is in that state - from the CLI (=monitor and keyboard connected to your NAS)


    Regards


    CDS

    Thanks for quick responses.


    I do not own a VGA-monitor but when I need CLI I connect it to the TV via HDMI. If I plug the TV in when OMV has frozen the screen is just completely black so I can't really tell if i have any way to control of OMV or not...


    I did however get some more information when the server stopped working today (this time after less than 2 hours!). At this point I was logged in via ssh as root when suddenly the following lines appeared:

    Code
    Message from syslogd@openmediavault at Aug  5 21:11:42 ...
 kernel:[ 6394.843670] journal commit I/O error


    After this happened I tried to access OMV via SMB, Web GUI and a new SSH connection but did not succeed. The old SSH connection did stay alive, but no commands that I was trying to execute after this were succesful:


    Code
    jay@openmediavault:/$ ls
-bash: ls: command not found
jay@openmediavault:/$ df -h
-bash: df: command not found
jay@openmediavault:/$ su
-bash: /bin/su: No such file or directory


    Could the errors be caused by the file system filling up or could it be that the file system is read-only? Unfortunately I am new to Linux and not experienced enough to find out if the filesystem is read-only (sorry for this). Can you perhaps guide me how to do this?


    I also have another piece of info which might be interesting. Two days ago I realised that I did not have any DNS server specified in the OMV settings. Because of the missing DNS server entry I had not seen any available updates in the web gui for some time as the apt-get url's could not be resolved. There were quite a few packages in there which I upgraded hoping that they would solve the problem, but as you already know it didn't. Nevertheless I now believe I have all OMV and Debian software up to date. (I don't know if I missed to enter a DNS server at the initial setup but updating through the web gui and the monit e-mails have been working before and had stopped a while ago without me noticing - Strange, right?)


    I have just restarted OMV to have a look in the syslog, any tips what to look for?


    My idea is also to manually monitor the available space on the OMV USB stick periodically through df -h. Even if i can not do it when the server has stopped working I might be able to see something happening before then. Is this a good approach or are there possible better ways, using cron or something similar perhaps?


    Many thanks for your assistance, I really appreciate that your taking the time.

    Zitat von "cds"


    Generally I would not recommend to run a NAS from a Flash-stick unless it is specially optimized for that.


    I did not realise it could be such a big risk using a USB stick as OS drive. The 5 year warranty specified by the manufacturer might not apply for this case but I definitely thought they would last for a year or so...


    Quite possibly I will switch to another setup later on, but at this point this is my only option since the Intel motherboard only has two SATA connections which I use for the RAID array. When the mSATA SSD's have dropped a little more in price I might consider that option...

    I can not see that anything looks out of the ordinary in the syslog right before and after the server crashed:



    I believe 21:11 was the time of the crash and 21:53 was the time when the server was restarted.

    Unfortunately I have not found a solution to my problem, but in case others experience this frustrating problem you might be interested to know that this is a bug in OMV which has already been reported:


    http://bugtracker.openmediavault.org/view.php?id=526


    The bug seems to affect people running OMV from USB sticks as well as external USB HDD's which are using USB 2.0. The suspected cause of the problem seems to be faulty debian USB 2.0 drivers, but I don't know if this is confirmed or not.


    If anyone knows when (/if) a fix to this bug can be expected I would be very interested to hear that.

    I have the same problem (actually have a very similar board to you, the Intel DN2500). I only installed OMV a few days ago and noticed it happened last night, so I pushed the power button, at which point the screen came up complaining about an I/O error. I wonder if the USB sockets are being put to sleep after a few hours or something.


    I hope there is a fix for this as I much prefer it to FreeNas etc as I have managed to get Shairport working (allows you to plug speakers into it and then stream music from Apple devices), which I couldn't in FreeNas.

    I gave up looking for a solution to this after a while and bought a msata SSD Instead, unfortunately it turned out both SATA channels could not be used when the msata drive was connected so I also had to buy an additional PCI controller card and the NAS therefore became a bit more expensive than I was expecting. However the SSD system drive really speeds up the response time of web GUI, which is now quite OK. I have been running the server with the new configuration for a few weeks without major problems.


    I hope you find a solution and if you do please share it with the rest of us.

Jetzt mitmachen!

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

Benutzerkonto erstellen Anmelden