I had 3 major problems One node couldn't complete indexing. I suspect the problem was related to it being the only node running a 32bit executable as the machine was far more powerful than some others in the network. The web gui is extremely slow probably due to the indexes being stored in flat files rather than some sort of database. This also makes indexing large numbers of files very slow. I have somewhere in the order of 500,000 files of all sorts of sizes. A fully indexed node was using ~ 1Gb of resident memory. Rather than keeping a stats table the gui scans the entire index to get the current status. During indexing the index is constantly changing so using the gui was nearly impossible during indexing. The real killer was the lack of deletion protection. I had read about it and turned on the staggered backup keeping files for 30 days after deletion. 2 days after I did this one node dropped a disk whilst scanning. The node announced that the files were deleted and wiped a 1.6TB share from the network. I tried to go to the staggered backup but all the files are renamed in there. There is a 'sed' command to reverse the renaming so I ran that and restored the files on the one node where I was using staggered backup. On another node I restored from backup and on the node that dropped a disk, I managed to get the disk back. The backup node and the restored disk node both contained the same data, the staggered backup node did not and was missing ~ 100GB of data..... I will take another look at syncthing when it has had some more testing, meanwhile I will use it for some less important data. Restoring 1.6TB across 14 nodes is not as fast as deleting it