mdadm: Add test case for syncing RAID 6 #1080
Conversation
The amount of syncable blocks differs from the total amount of blocks. Signed-off-by: Julian Kornberger <jk+github@digineo.de>
| @@ -11,18 +11,22 @@ md0 : active raid1 sdi1[0] sdj1[1] | |||
| md4 : inactive raid1 sda3[0] sdb3[1] | |||
| 4883648 blocks [2/2] [UU] | |||
|
|
|||
| md6 : active raid1 sdb2[2] sda2[0] | |||
| md5 : active raid1 sdb2[2] sda2[0] | |||
There was a problem hiding this comment.
Please leave these unchanged, so that the diff is easier to read. Maybe name the new one md5.
I would prefer do this cleanup as a separate noop PR for the end-to-end fixture changes.
There was a problem hiding this comment.
Ok, I'll split my commit up.
|
Interesting, it looks like the "syncable" value is 3x the storage value in your example. I guess we need to expose a separate metric that contains the "syncable" value. |
|
Unfortunately the syncable blocks are only displayed during the synchronization. |
|
Yea, I think we're going to have to change the behavior of these metrics and only expose We should open an issue to discuss this. |
|
Closed in favor of #1085 |
In case of RAID 6 the amount of syncable blocks differs from the total amount of blocks.
The exported metrics can not be used to calculate the synchronization progress.
Is that the expected behaviour? Or do we need a new metric for syncable blocks?