thanks for response, I created this job so I really know that it couldnt be cause...

Next point

- I checked that was full RAM - it was full about 3days before I think

is just my opinion, that during backup was some IO operation and somebody or program executed poor query and it had to use DISK instead RAM, and there can be problem, that DISK was not response>> lot of IO and disk couldnt handle ...what do you think ?

But still I dont know why sql service was restarted,, it is possible that it was not responded during this situation and it was automatically restart? Any experience with this?