I am currently working in a Windows 2003 server 32bits environment.
I have 4 nodes that have synced correctly, they are all up to date and syncing the changes in the 4 servers (one is only for copy that is not shared in the network).
I added a new node but I have problems constantly; I am still awaiting for an initial load, before adding files from the office where I added the node.
It hung in random periods, 10-20-40 minuts, in data size, 500MiB, 2GiB, 4,5GiB, its not regular.
I checked the config, its the same that all the other nodes. I added a ping to check the WAN connection. Also, I activated the debuging, but the log seems the same.
How can I try to detect the problem ??
I will change all the servers, for new ones (ubuntu 16 - 64bits), but I need them running fine now, not with the new ones.
I can have both nodes at the same time ? they are compatible (32 and 64) ?
Javascript has nothing todo with this.
It’s not clear to me what hangs, if its the transfers or the scanning. Regardless, check system resources on all boxes, I suspect you’ll find that either disk, cpu or memory is maxed out, hence the issue.
I have about 450.000 files and 500GiB in the sync folder (4 nodes up to date, 1 downloading).
I haven’t seen any strange event in any server, all the servers uses 2TB new drives.
I have all the servers with a 600s scanning interval.
The node where I am experiencing problems is a Xeon 3040 @ 1,86hz, with 4 GB ram. And yes, syncthing it loading all CPU. (I am preparing a new server to replace it)
Sounds like a deadlock issue we are trying to hunt dow . There is an issue on github (search for deadlock) which provides an instrumented build and vague explanations how to debug.
I suspect I’m meeting a similar issue since a couple of days, as I’m adding a new sync server. Populating the new server just hang once in a while, and I have to restart the process.
I have it on debian jessie x64 and raspbian jessie. Syncthing is 14.9 on all machines, including “client” machines.
Do you need some other log or info?
I disabled the slowest node.
I have found that the other nodes syncking/updating spend few seconds.
But the slowest one takes some time scaning. Could this stop or break the transfer ?
I am controling the progress, I will comment.
Please try the build from Folders out of sync, even after rebuilding indexes with the environent variables I’ve listed set, and look out for panics. If you can pinpoint which exact release broke it for you, that would also be useful.
I am currently working with 0.14.10 its running better than 0.14.09.
Due to lack of time I am still awaiting to change the servers from windows 32 bits to linux 64 bits.
Before I wish to avoid bottlenecks with processor, memory and disk bandwith.