[sheepdog-users] failed to read inode header

Maxim Terletskiy terletskiy at emu.ru
Wed May 28 17:12:39 CEST 2014


Now I see next things. 3 of 4 nodes online, they are replicating as I 
see via iostat and vnstat. Fourth node can't start sheep daemon. Dying 
with this error:
May 28 18:04:42 ERROR [main] init_vdi_state(213) failed to read inode 
header 80bb794400000000 0

I found file with the same name on other node, counted md5 and they are 
the same. So what's wrong with this piece? How can I skip reading this 
inode and start sheep? Can I just delete it?

The next strange thing is that when I do "dog node list" I see my failed 
node in list(but no sheep process running on this node). Is it normal? 
If not how can I kick it manually? In other nodes logs I see many 
messages about error while connecting to this node.

Can anyone help with this? :)

28.05.2014 17:50, Maxim Terletskiy пишет:
> Hi!
> While replicating some nodes in my cluster fail. Now can't start sheep 
> on this node with nex error:
> failed to read inode header
> On other nodes I can't even do "dog node recovery", getting only errors:
> failed to connect to Connection refused
> - addres of failed node.
> VMs stuck.
> Can someone say how to deal with this? How to start node which can't 
> read inode?

More information about the sheepdog-users mailing list