Yes I have an index group not an index pool.
The index group is backed up in the same pool that the data but after all the group.

And in the options group I selected "No index save"

It's why I just have the backup of the bootstrap. No index.

But I just discovered I forgot to select the index group in the client ressource. But I think it's not he problem.

Just an another thing, sometimes the backup is running well and sometimes it goes in loop.

Do you have Index pool? If not how is the current pool restricted?

What I usually do is group with all clients and all other groups are no index save. I have also active group with backup server client, but with client having skip schedule associated. Then from crontab twice per day we do index&bootstrap save. Other backups do not generate index and bootstrap save so we never saw something like your problem.

If this sometimes runs and sometimes not I can only assume this is NW bug.

Christophe,

are these two servers the only two storagenodes connected to this library? We had the same kind of problem, where a tape was mounted on the wrong storagenode and this was probably caused by a 64bit windows storagenode, which was only supported in NW732Jumbo after we had Fujitsu push EMC to validate this config.
You may want to check hardware compat guide for this.

One other this, you say that when backing up the database, you get a bootstrap backup of the server. This should only happen if the backup server is not configured in an enabled savegroup. To prevent the bootstrap from not being backed up, networker does a bootstrap backup at the end of every savegroup. This will lock the media database numerous times during the night and could cause a lot of havoc.
Try creating a savegroup just for the backup server and run that backup once a day, or more often if you wish. This will leave you media database open all through your backup window.

xander