Unassigned shards delaying allocation

Hi guys!
I have a ES cluster with 6 nodes,
today two of these nodes was down, hosts were rebooted
After that i see a lot of uassigned shards in KOPF.

and a lot of messages in logs:

>  [2015-10-15 11:55:12,765][INFO ][cluster.routing          ] [master] delaying allocation for [0] unassigned shards, next check in [0s]

also i want to say that three nodes havenot space on disk, they are in readonly

[2015-10-15 11:53:49,270][INFO ][cluster.routing.allocation.decider] [master] high disk watermark exceeded on one or more nodes, rerouting shards
[2015-10-15 11:54:49,253][WARN ][cluster.routing.allocation.decider] [master] high disk watermark [20gb] exceeded on [N9Z7j6TbQjaSdIkBIHoSgw][data3] free: 18gb[0.8%], shards will be relocated away from this node

What can I do to resolve this problem?

The hosts that were rebooted, are shards not being allocated on them? Is the low disk space condition on the other nodes?

Now all cluster looks green, strange...
In 4 hours there were unassigned shards...
Where can I trace what was happened?