2 OneDrive accounts: one OK, the other many 503s

I have two different OneDrive accounts. Both I can access just fine over the web. Both are set up in rclone correctly.

One account consistently works just fine, no retries or pacer limiting.
For the other account, roughly 19 times out of 20, the access results in a "503 Service Unavailable" error.

I know the failing account is set up with correct authentication etc. because every once in a while it works. The 503s result in the pacer inserting delays. Often even after 10 retries with longer delays, it still fails.

My rclone version is:

rclone v1.47.0-015-gaba15f11-beta
- os/arch: windows/amd64
- go version: go1.11

This is confusing to me, as both accounts are served by Microsoft.

Below are the "rclone -vv" results for one iteration for the failed case. Does anyone know why this may be happening?

Thanks, Steven

2019/06/12 22:16:03 DEBUG : rclone: Version "v1.47.0-015-gaba15f11-beta" starting with parameters ["C:\\bin\\rclone.exe" "-vv" "--dump-bodies" "lsd" "ms18new:"]
2019/06/12 22:16:03 DEBUG : Using config file from "C:\\Users\\STEVEN\\.config\\rclone\\rclone.conf"
2019/06/12 22:16:03 DEBUG : >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
2019/06/12 22:16:03 DEBUG : HTTP REQUEST (req 0xc000105a00)
2019/06/12 22:16:03 DEBUG : GET /v1.0/drives/b!iJTwKgAX1U-oqqOHA0YkH0idAsn1MKNLrtIDX_HoN2W2s1b0t4OlT492YmRAcJbQ/root HTTP/1.1
Host: graph.microsoft.com
User-Agent: rclone/v1.47.0-015-gaba15f11-beta
Authorization: XXXX
Accept-Encoding: gzip

2019/06/12 22:16:03 DEBUG : >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
2019/06/12 22:16:03 DEBUG : <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
2019/06/12 22:16:03 DEBUG : HTTP RESPONSE (req 0xc000105a00)
2019/06/12 22:16:03 DEBUG : HTTP/1.1 503 Service Unavailable
Transfer-Encoding: chunked
Cache-Control: private
Client-Request-Id: 02f9e8e1-478b-4f05-8913-41f45f5ce298
Content-Type: application/json
Date: Thu, 13 Jun 2019 02:16:05 GMT
Duration: 260.2546
Request-Id: 02f9e8e1-478b-4f05-8913-41f45f5ce298
Retry-After: 30
Strict-Transport-Security: max-age=31536000
X-Ms-Ags-Diagnostic: {"ServerInfo":{"DataCenter":"East US","Slice":"SliceC","Ring":"5","ScaleUnit":"002","RoleInstance":"AGSFE_IN_48","ADSiteName":"EUS"}}

  "error": {
    "code": "serviceNotAvailable",
    "message": "Service unavailable",
    "innerError": {
      "request-id": "02f9e8e1-478b-4f05-8913-41f45f5ce298",
      "date": "2019-06-13T02:16:05"

2019/06/12 22:16:03 DEBUG : <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
2019/06/12 22:16:03 DEBUG : pacer: low level retry 1/10 (error serviceNotAvailable: Service unavailable)
2019/06/12 22:16:03 DEBUG : pacer: Rate limited, increasing sleep to 20ms
2019/06/12 22:16:03 DEBUG : >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

Hmm, not really. It must be some config at Microsoft?

You could try slowing things down with --tpslimit 1 and see if that helps? Also setting a different --user-agent may help? There was a Microsoft page which I can't find any more, describing how to form a user agent string.

1 Like

I agree, but I wonder what would be different due to the fact that I am using 2 different login accounts? They are both served by Microsoft.

I don't know - maybe you have hit some internal limit and one account is marked as being abusive or something like that.

Did you try these?

Tried the --tpslimit 1 option, and it stopped the messages. So I think that is the key.

I will fine tune the value to get the best performance vs. rate limiting.


-- Steven

1 Like

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.