Hi. I’m getting the strange GUI notice every time a folder is updated on a source host:
2017-10-11 09:58:48: Request from st01.... for file sysstates/vm-dc2/WindowsImageBackup/vm-dc2/Backup 2017-10-10 213016/22a29103-870e-11e0-a053-806e6f6e7763.vhd in unshared folder "i4udq-q2yx2"
The same message appears in the console with WARNING prefix.
The source host (the sender) is permitted to send only.
The folder (“i4udq-q2yx2”) is shared between that two hosts.
Please, help me to find out what’s going on.
OS on both hosts - CentOS 7
Syncthing versions - v0.14.39, Linux (64 and 32 bits)
I’m getting this notice every day ('cos I’m backing up Windows every day). I clear it up (by clicking OK button) and get the notice again on the next day:
I mean if you restart syncthing, does that still happen? Is there more than one source of these files, as I suspect if there wasn’t the other device would fail to sync?
I suspect it’s some sort of race condition, but it’s not clear how it happens, and I’d expect sync not to work once you hit this as the message comes from the same location that checks permissions.
2017-10-15 09:59:42: Request from st01.skip for file sysstates/vm-dc1/WindowsImageBackup/vm-dc1/Backup 2017-10-14 223007/33a29103-870e-11e0-a053-806e6f6e6963.vhd in unshared folder "i4udq-q2yx2"
I have one idea. 1st of all, the sub-folder “Backup 2017-10-14 223007” do not exist anymore on both hosts (source and destination). What does exist - the folder “Backup 2017-10-15 223008”:
Source:
[root@st01 vm-dc1]# pwd
/backup/filer-1/sysstates/vm-dc1/WindowsImageBackup/vm-dc1
[root@st01 vm-dc1]# find . -type f -ls | grep 33a29103-870e-11e0-a053-806e6f6e6963.vhd
230546325 17074352 -rwxr-xr-x 1 bacula disk 29239443456 Oct 16 16:24 ./Backup\ 2017-10-15\ 223008/.syncthing.33a29103-870e-11e0-a053-806e6f6e6963.vhd.tmp
[root@st01 vm-dc1]# date
Mon Oct 16 16:25:14 UTC 2017
So, the parent folder (“i4udq-q2yx2”) is definitely in sync.
That file in the output is pretty huge - ~28GB.
What if the new backup (on the source host) has been started and removed the folder “Backup 2017-10-14 223007” before the file sync has been completed? What error message should be in this case?
If you are talking about on disk folders rather than syncthing folders then it should have been something about peers having this file going away (as there is nowhere to find this file), if you are talking about syncthing folders, and you are cycling them via some automated method, then this is understandable, as you remove syncthing folders for which requests might already be in flight
Looks like I was unclear - I apologies. The source (filer-1) shares the folder /zmir/backup with st01 (/backup/filer-1). The sysstates and underneath subfolders are not shared separately.
Not sure I understand what you mean by “automation”. The only couple things I have are
a daily cronjob which retrieves the Syncthing errors using curl and sends out an email alarm (GET from localhost:8384/rest/system/error).
nightly cronjob which stops and starts syncing (POST to localhost:8384/rest/system/config) to avoid synchronization while the nightly backups are running on the source host
I suspect this is the cause of it, as this pauses the folder, at which point it becomes unshared with the device, but there are still requests for it in flight if the data recently changed. Better option would be to kill syncthing at that point.