The general consensus is that RAID5 rebulid times are getting quite high for large RAID 5 volumes. During rebuilds the drives can be quite stressed and rebuilds may fail if another drive has any URE's (unrecoverable read errors). It's better to think of probablity of URE's rising with the overall RAID volume size rather than number of drives which is why this issue is being considered more and more with todays large drive sizes.
The typical method to defeat this is to employ RAID 1/10/6/60.
RAID 5 is slower than RAID 1 or 10 for rebuilds due to the parity calculations. RAID 6 and 60 are also hampered by these issues but can tolerate another drive dropping during the stressful rebuild so the problem isn't quite so bad.
In your situation the rebuild time shouldn't be too bad, but software RAID controllers aren't great. If you really depend on your data I would recommend a hardware RAID controller with battery backed cache. This way even if you had a power cut as data was being written, any data committed to the RAID controllers cache would be preserved by the battery to be written to the drives when power is restored.