Comment 21 for bug 1369524

Ivan,

> Current CentOS behavior is to remove lvm when device/lvm became not used.

I don't think lvm could possibly operate that way. Waiting until the LV in question gets released
might take an indefinitely long time.

> Parallel run of this two commands is simpler way to run/emulate 'remove' after write to volume with zero delay.

Well, such a simulation is misleading, also, just because the dd process has exited doesn't mean
the LV it was writing to gets immediately released (there might be some data in the block cache,
and it has no way to find out that those zeros are actually useless and can be discarded). Perhaps
you patch (conv=fsync) will improve things a bit.