mvdzwaan Posted August 15, 2012 Share Posted August 15, 2012 If you replace a 2Tb with a 3Tb one, the interface starts up and says which disk is missing and its size. When you start a rebuild but stop it, shutdown and swap the new 3tb with another one the interface starts up and says the previous 3tb one is missing and its size is 3Tb. But ofcourse you're still rebuilding the 2Tb disk with the new one. Minor one I guess, and hope, but just noticed it. Link to comment
JonathanM Posted August 15, 2012 Share Posted August 15, 2012 I don't think it's a bug, because you are replacing the aborted 3TB with another 3TB. As soon as you replace a smaller drive with a larger one, it expands to fit the larger drive. You would no longer be allowed to replace it with a 2TB, as you started the build with a 3TB, so the only option is 3TB or larger. I guess the moral of the story is don't start an operation you don't intend to finish. Link to comment
mvdzwaan Posted August 15, 2012 Author Share Posted August 15, 2012 To my understanding it's rebuild and than expand... The reason for stopping the rebuild was because of problems with the new disk. It finished pre-clearing without issues, smart reports were ok, speed was ok, but during the rebuild I experienced very slow performance compared to previous rebuilds. b.t.w. the new new disk is rebuilding at the expected speed of 50Mb/s Link to comment
JonathanM Posted August 15, 2012 Share Posted August 15, 2012 To my understanding it's rebuild and than expand... That is my understanding as well. However, the rebuild procedure is one operation from the standpoint of unraid recognizing that it is supposed to put the contents of a drive it is currently simulating from parity onto a new physical drive. It needs to know the size of the drive from the beginning, as it cannot rebuild onto a smaller drive. The most efficient way to rebuild seems to fall into 2 cases, if the drive is identical in size, simple rebuild, if it's larger, copy the data and expand the file system. It still has to account for parity, so the expanded portion must have zeros written before the file system can be expanded. I guess Tom could have written the rebuild routine so that it checked for minimum size, rebuilt, checked for larger size, zeroed the remainder, then expanded the filesystem, but he apparently didn't. The edge case of someone starting a rebuild on a larger drive, aborting, then rebuilding on the original size or something in between didn't get accounted for. Link to comment
lionelhutz Posted August 16, 2012 Share Posted August 16, 2012 The first step taken when the disk is replaced is to record the new drive data into the disk inventory. Once that is done, unRAID had over-written the old data and it's like that old drive didn't exist. It coud be done with a more complex process, but that makes it easier to have a bug. FYI, there is a way you could rebuild onto the smaller drive again at least with 4.7, but it requires some manual work. Link to comment
Recommended Posts
Archived
This topic is now archived and is closed to further replies.