Hey everybody, glad to see I'm not the only one facing this problem. I have two GDrive mounted, each with it's own clientid on CentOS 7. One of them keeps failing and the other is working perfectly.
As others have explained, the issue only happens after a while. I notice that the command that makes it crash for me is: df -h it will basically hang the session. This is the log file it generated:
2019/10/04 10:15:26 DEBUG : Google drive root 'USB1': Checking for changes on remote
2019/10/04 10:20:56 INFO : Google drive root 'USB1': Change notify listener failure: Get hxxps://www.googleapis.com/drive/v3/changes?alt=json&fields=nextPageToken%2CnewStartPageToken%2Cchanges%28fileId%2Cfile%28name%2Cparents%2CmimeType%29%29&includeItemsFromAllDrives=true&pageSize=1000&pageToken=1847898&prettyPrint=false&supportsAllDrives=true: stream error: stream ID 1921; INTERNAL_ERROR
2019/10/04 10:26:26 INFO : Google drive root 'USB1': Failed to get StartPageToken: Get hxxps://www.googleapis.com/drive/v3/changes/startPageToken?alt=json&prettyPrint=false&supportsAllDrives=true: stream error: stream ID 1923; INTERNAL_ERROR
2019/10/04 10:31:56 INFO : Google drive root 'USB1': Failed to get StartPageToken: Get hxxps://www.googleapis.com/drive/v3/changes/startPageToken?alt=json&prettyPrint=false&supportsAllDrives=true: stream error: stream ID 1925; INTERNAL_ERROR
and this is the mount script I use for both (changed the log to debug to see if anything extra showed up):
I upgraded from an older version (1.48.0). Mainly wanted to get access the the rclone gui, but I still should have stayed on the old version. The beta seems to be working.
@ncw So far, so good. When i was on 1.49.4 and kicked off a sonarr scan, it immediately started freezing. Now on the beta, scans are back to normal speed and the system appears to be working well.
i am on 1.49.4 too and have had trouble with my google drive too in the last days. I have read something about http2 in this thread which i have no idea about, but i found it in my logs https://pastebin.com/NYjkq3t7
low level retry keeps increasing (2/10, 3/10) and sleep time too)
The mount is browsable but i cannot do anything with the media. Does not matter if i try to copy a file from the remote or plex wants to read a file, it doesn't work. But the directories are browsable and the debug log will update all of this The log also writes the request of these files, yet it cannot be opened.
API limit has not been reached. There are actually no API requests logged at google after this incident happenes.
Nick, this morning I re-mounted with the latest beta (178), and it seemed to work for a while, but I just got home to find the same errors again. The mount was started at 9:48 am, and the first error occurred at 6:36 pm. See here:
Can you try with the latest beta but with HTTP disabled? So set this environment variable before you run rclone
export GODEBUG=http2client=0
All the error reports I've seen have HTTP2 stuff in them so that may help narrow the problem down if it works.
All the errors in this thread look like this which are internal errors from the http library
Failed to get StartPageToken: Get https://www.googleapis.com/drive/v3/changes/startPageToken?alt=json&prettyPrint=false&supportsAllDrives=true: stream error: stream ID 1785; INTERNAL_ERROR
Digging through lots of go issues, I think this may be caused by a timeout, which kind of makes sense as the changes api is long living.
I don't know whether these errors are normal, but rclone isn't retrying them...
This might be something to do with the context being cancelled also...
