"Failed to reset background IO priority" warning with v2.0.3.6

JonMikelV · May 22, 2018, 8:52pm

I’m testing 2.0.3.6 canary (Docker container from duplicati/duplicati) on unRAID 6.5.1 (Slackware based).

I recently enabled --use-background-io-priority and have since started getting the following warning with every backup:

Failed to reset background IO priority, exitcode: 0, stderr: ionice: ignoring given class data for none class

Note that I only see the warning in Live log (any level but “Error”) and in warning emails (where it’s grouped as [Warning-Duplicati.Library.Main.ProcessController-BackgroundPriorityError]).

Disabling the parameter makes the warning go away.

Has anybody else run into this?

kenkendk · May 23, 2018, 9:03am

From what I can see, operation succeeds, but prints a warning message.

The code that produces it is this:

github.com

duplicati/duplicati/blob/master/Duplicati/Library/Main/ProcessController.cs#L350




if (Duplicati.Library.Utility.Utility.IsClientOSX)

{

    // TODO: We can only give lower priority, thus not reset it ...

    data = RunProcessAndGetResult($"renice", $"{m_originalNiceLevel} -p {pid}");

    if (data.Item1 != 0)

        Logging.Log.WriteWarningMessage(LOGTAG, "BackgroundPriorityError", null, "Failed to reset background IO priority, exitcode: {0}, stderr: {1}", data.Item1, data.Item3);

}

else

{

    if (m_originalNiceLevel < 0)

        data = RunProcessAndGetResult($"ionice", $"-c {m_originalNiceClass} -p {pid}");

    else

        data = RunProcessAndGetResult($"ionice", $"-c {m_originalNiceClass} -n {m_originalNiceLevel} -p {pid}");



    if (!string.IsNullOrWhiteSpace(data.Item3))

        Logging.Log.WriteWarningMessage(LOGTAG, "BackgroundPriorityError", null, "Failed to reset background IO priority, exitcode: {0}, stderr: {1}", data.Item1, data.Item3);



}

My understanding is that the process originally had class “none” (0).
When resetting the IO priority class back to “none”, it adds the “class data” that was originally reported, but which is does nothing.

I think we can fix it by changing the line to:

if (m_originalNiceLevel < 0 || m_originalNiceClass == 0)

Only potential issue is that I do not know if all ionice versions expect the same. Potentially we would restore with a “slightly different” io-priority.

Long term, my plan is to make the backup run in an external process, which removes the problem, as we do not need to reset it, we just exit the process.

JonMikelV · May 24, 2018, 3:58pm

Thanks for digging into it!

I realized I spaced out a bit on the original post and kept saying it was an error when really it was just a warning, so it’s not actually causing the backup to fail.

With that in mind, the scope of the issue currently being 1 user affected and your future plans probably removing the whole problem I doubt it’s worth making a change that might cause other issues.

We can always revisit it if enough other people run into the issue and let us know about it (like with responses in this topic).

JonMikelV · June 18, 2018, 2:20am

Just as an FYI this still happens when running in my unRAID Docker container on 2.0.3.6 canary.

stepanma · May 20, 2019, 10:44am

I am getting the same “Failed to reset background IO priority, exitcode: 0, stderr: ionice: ignoring given class data for none class” warning on “Debian 4.19.28-2~bpo9+1 (2019-03-27) x86_64” and “Duplicati 2.0.4.5_beta_2018-11-28” (Docker container from linuxserver/duplicati).

Can this warning be suppressed somehow as getting emails on this is misleading.