[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [PATCH 07/14] Move identifyMultipaths from DeviceTree to devicelibs.



> >+    # this is the list of devices we want to keep from the original
> >+    # device list, but we want to maintain its original order.
> >+    singlepath_disks = filter(lambda d: d in devices, singlepath_disks)
> >+    #multipaths = filter(lambda d: d in devices, multipaths)
> >+    partition_devices = filter(lambda d: d in devices, partition_devices)

Note that I have no idea why multipaths = is commented out there.  I
need to recheck that.

> This might be me not parsing the python code above correctly, but
> I don't think this does what you want it do, it does not put the devices
> inside singlepath_disks / partition_devices in devices order.
> 
> Also note that this is not necessary. udev not always give us devices in
> a nice logical order, so sometimes we scan devices before we have scanned
> there parents (I've fixed several BIOS RAID related bugs caused by this),
> so the devicetree.py scanning code is setup to handle parents not being scanned
> yet and in that case force scanning the parent first.

I believe the idea here isn't that devices are in any logical order,
just that they are in the same order on output from identifyMultipaths
as they were on input.  Anyway I just modified the code enough to change
the return type and move it to a different file.  Peter will have to
speak to why the ordering is important, but I believe it's all due to
multipath.

- Chris


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]