Posts by Jopa

    Hi all,


    after I tried out early omv4 on my backup server but went back to omv3 last year, I gave omv4 another chance now. On same hardware as before - see my signature -, just on a different drive, so that I simply can swap back to running omv3.


    Now I have the following: With one of my filesets for testing, some 3.3GB of mp3s in 25 folders, which are blown by SMB to gb ethernet at full link speed, omv4 writes with 85-95MBps and collapses down to <1MBps inbetween, whereas omv3 writes with some 10MBps more and breaks down to about 30MBps only. (AFAIS the break-down is because the fileset is 2 main folders, with 22 subfolders in one of them.) While I tested, there were definitlvely no other accesses to omv to steel performance.


    Although the hardware, I found this out with, is my private backaup server, and for this purpose, speed is not the most important feature, it's my testing system, so I'm still interested in knowing the reason of this. And what I can do to overcome it, e.g. use an i5 rather than the i3, and/or install more RAM.


    Best regards,
    Johannes

    Well, firefox not always seems to do things quite asap...


    What I did on yesterday was:
    - set ff to "keep cookies",
    - logged into omv with language set to "german",
    - closed ff and after some seconds reopened it,
    - found omv login asking for language,
    - wrote my last replies,
    - reset ff to delete cookies at closing,
    - finished my session an shut down my pc.


    After I powered up pc today and opened ff again, I found omv login in german as wanted. Looking at the cookies, there now is the one for omv language to be kept for one year.


    Thanks for help, even if it did not work as quick as expected.


    Best regards

    Hi Folks,


    at company, where we still run OMV 2, there is no problem with WebGUI login screen. It's coming up with former selected language (german) always.
    Only at home, where I run an OMV 3 and a new OMV 4 (as a backup server), after I closed an reopend my browser (firefox) for any reason, I always must select right language again.


    Is there any screw in OMV 3/4 to trim this?


    Thanks in advance for any hint.

    Hi all,
    my problem started with nearly daily smartd alert mails like this:


    At first, when I looked after in smart extended information, I found latest error log entries to be empty. Until I invoked daily short-selftests - and waited some one week. Which changed extended information to this (different drive):


    From former empty error log entries there is just the "Warning: ATA error count 10 inconsistent with error log pointer 4" remaining. Ok, but what could be the reason for alerts still coming in? Guess it is the "Error: ABRT", which seems to be the drive's reaction on smartd's (whatever) last action when it gets terminated by autoshutdown


    To confirm that context, I disabled autoshutdown and voila - there are no smartd alerts no more for couple of days now. Only letting run the omv machine all the time is not really, what I want to do.


    So my question: How can I make sure, that autoshutdown waits for any smartd activity to be finished? Is it possible by just tuning the basic parameters (HDDIO?) or must I add a dedicated lock/unlock mechanism like here?


    Btw another question: Is a shutdown by hand, i.e. by Web-GUI or by pushing the on/off button, known safe, not to disturb drive's smart the same way, too? I just didn't watch for this yet.


    Best regards

    Thank you, I think, I'll give rsnapshot and backup plugins a try for server backup. I don't beleive, that I can convince the rest of admin team of keeping extra backup media in mind all the time ;)


    Nevertheless I'll still play around some more with urbackup - it's just a cool backup system for non-linux PCs.

    Hi,


    it may be special, but it's really painful: A couple of WD disks I used for a customer omv, in particular WD2002FYPS, obviously do not meet S.M.A.R.T. specification fully. Because they sometimes log empty error entries, for what reason ever. Which do not keep the disks from running even extended selftest successfully, but urges ovm to deliver warning mails every day.


    As the empty log entries are the only thing, I can find going "not normal" on these drives, I'd like to remove them from alert mailing. Does anyone know a way to achieve this?


    Best regards

    @macom


    Thanks, Duplicati could be interesting, but not for my actual backup project.


    @ryecoaaron


    I read the article, you linked, and I think, I understood. But it still does not seem to solve my task without extranous scripts or hasardous cron jobs to delete or move away current backup - to get a new full backup periodically, say every few month, and not once for ever. Or do I still understand something completely wrong?

    Well, I know rsync from a former suse-based server-backup setup. And it needed a carefully developed and tested script to do what I wanted...


    As I read your reply, running rsync for periodical full and differential backups in the meantime may not need as special scripts as I needed formally?

    Thanks. I'll take a look at it.


    Why I want to use UrBackup? Just because of it's charming server GUI, which lets me plan full and differential backups far more easily than a rsync script ;) Provided, server finds the clients and can connect to them...

    Hi folks,


    at company we are running an omv 2.x with two RAIDs - one for fileserving, and one for system backups of all our pcs. For the latter I did some experiments with existing drives, before I ordered the final setup. Which is working perfectly so far.


    Only, as S.M.A.R.T. log tells, the experimental drives are not forgotten and tested every time, smartd is (re)started, which is daily, always throwing log entries about drives, that are not existing and not listed under /dev/disk/ no more. They are only still listed in /etc/smartd.conf, and, because that file is auto-generated by omv, the "normal" way to edit /etc/smartd.conf will not really help.


    So my question: Where is the list of drives, omv uses to generate /etc/smartd.conf, from which I can delete the drives that are not of interest no more?

    Sorry, sparx, but the "Total uncorrected errors" will typically tell you the number of uncorrected errors only, and not necessarily point to the LBAs that are faulty. Of course, something else but 0 should catch your awareness, but you will not get further information about the failure. Same is for "Elements in the grown defect list" - more than the number of defects cannot be expected. It's basically the same as with SATA drives: They, too, don't tell you the LBAs that they reallocated, just the number of reallocated sectors and those pending for reallocation.


    And re specific tools from disk vendors to give more detailed SMART data for SAS drive - forget it. They are either not existing or kept for OEM customers only, who buy 50-100k of drives minimum. IMHO they actually do not exist at all. Why else the community, that is working smartmonitortools, which not only oms is using, should not be able to handle SAS drives better yet?

    Even if omv makes an enabled SMART status for SAS drives possible, by the time I don't think it to be as significant as the evaluation of the attributes of SATA drives.
    Just because SAS drives do not talk about their internals as much as SATA drives.
    And because SAS drives need hours to do their long selftest. As you see from omv-screenshot attached, my Seagates need 460 minutes for a long selftest. And after that time, even do not tell more about their state than usage of start and load cycles and power on time. I don't think, that read errors are actually meaningfull, as long as they get corrected by ECC fully.


    So I'm going to shift from SAS drives, even if they are really speedy - after they finished selftest -, back to SATAs for my private omv.


    Best regards,
    Johannes

    Moin,


    der große orange/gelbe Bereich in der OMV-Grafik erinnert mich an meine ersten Versuche. In meinem Fall es die laufende Systemdatenerfassung per rrd..., die meine System-SSD mit den vielen Mini-Schreibzugriffen einfach viel zu schnell kaputt geschrieben hat.


    Ich bin danach auf echte "drehende" HDDs umgeschwenkt und habe seither bestenfalls oder vielmehr schlimmstenfalls mal 10% Wait-IO. Es gibt wohl auch ein Plugin von omv-extras, das besser mit Flash-Speichern dealt, aber mich damit zu befassen, hatte ich bisher weder die Zeit noch hinreichend Spielmaterial, sprich SSDs, die eventuell auch noch sterben durften.


    Grüße

    Hi,


    I'm searching for a way to wake up my multi TB SAS-drive based OMV on demand by the client, which is a much more power-saving MacMini running longer than my OMV server.


    Upto now I always need and use an extra command or batch to issue WoL, but I'd like to omit it. As I'm a developer (h/w and s/w) and like the silence, I setup OMV's autoshutdown plugin to rather quick shutdown. Only when I'm busy with my project, I oftenly forget to reactivate OMV before next DRC or build, and then annoying error messages pop up...


    So my question as above. As all I found by google is about WoL on starting a client from shutdown/hibernate/suspend, but not about issueing WoL from a running client, I put this question here.


    Any ideas are welcome.


    Regards,
    the Jopa