Hi,
we have exactly the same problem (but we are not bonding and our mtu is set properly). After some network errors and dropped packages we have the state consistent/diskless on three LV as can be seen here: http://imageshack.us/photo/my-images/42/diskless.png/
There were no errors from the Raid-controller.
How can we come back to a consistent/consistent state?
I have also opened a ticket for this: 1031171
The system start to be in Diskless when there are mostly an I/O errors, so the system try to protect your data by going to diskless mode, as keep reading/writing while there are I/O errors that could lead to lose some data.
In such case, the best to do is to check the RAID controller and storage disks health.
Hi,
we don`t have any disk errors logged by the raid-controller but we had some network problems. My question is how I can come back to an consistent/consistent state and how I can update now. The primary system has 6 LVs, the secondary too. On each LV is a replication task(primary is source, secondary is target). On the primary system 5/6 LVs are in the state diskless/consistent. So my first question is: does that mean, that the storage for these LVs is actually accesed on the secondary system via the replication interface?
The last of the LVs is in the state consistent/consistent. So how should I update the system and come to an consistent/consitent state? Should I shut down the primary and use the secondary as main storage (as it has all LVs in a consistent state) while updating the primary and afterwards, when both are in a consistent state again, switch all back?
Or should I shutdown the secondary (on which in my opinion the storage is accesed), update it and start the replication from an (diskless) primary again?
Administrator
Posting Permissions
Reply With Quote