I picked OneDrive because out of all the other Cloud Storage because it had a clean interface, generous space for the money and suposedly treated files and files. Also the OneDrive API is suprisngly good, even if the documentation is a little dated. I have been quite happy using OneDrive for a while until I noticed some issues with syncing. Files I had not changed for a while, were being resynced to OneDrive.
I use rclone to sync from local hard drive to OneDrive cloud. All was working well. Then I started to notice some JPG photos that I had not changed were syncing again.
Turning on rclone debugging showed what was going on:
2021/05/08 12:37:12 DEBUG : 20181225-060457-SW1-000.jpg: Sizes differ (src 8630795 vs dst 7400740) 2021/05/08 12:37:12 DEBUG : 20181225-060457-SW1-000.jpg: Starting multipart upload 2021/05/08 12:37:13 DEBUG : 20181225-060457-SW1-000.jpg: Uploading segment 0/8630795 size 8630795 2021/05/08 12:37:24 DEBUG : 20181225-060457-SW1-000.jpg: SHA-1 = 7221b4e84fad93c05df6fbb796dc6201cb23f536 OK 2021/05/08 12:37:24 INFO : 20181225-060457-SW1-000.jpg: Copied (replaced existing)
To my shock the version on OneDrive was NOT the same as the version I had previously uploaded! The sizes Sizes, original = 8630795 vs OneDrive version: 7400740. Thats over 1GB difference. WTF?!
Checking version historyon OneDrive:
Every time I uploaded the file, OneDrive was changing the file a minute later. So next time I did a sync, the files were checked/verified as different and my original local file uploaded again. Only to be changed by oneDrive again.
This is NOT what I exepect. OneDrive should maintain the integrity of the files I upload.
I made sure the following options were disabled in case these change files.
![]() |
![]() |
Albums off | Tagging off |
I downloaded the file from OneDrive to compare against the original.
7400740 bytes = 20181225-060457-SW1-000-orig-one.jpg 8630795 bytes = 20181225-060457-SW1-000-orig.jpg
Holy Smoke, thats 1GB differnce!
sha256sum 20181225-060457-SW1-000-orig.jpg efac4ec9e56b38cc3e13fee423d1e6a01c79a8cbeb2a9d76fc6dd1636ca7028d 20181225-060457-SW1-000-orig.jpg sha256sum 20181225-060457-SW1-000-one.jpg f6d1fd2666532c0c23de65296ce5fca135be25d77bf99bcc01cc9583fa87737e 20181225-060457-SW1-000-one.jpg
These files are different.
Visually checking the photos look the same. Double checking with ImageMagick the actual image was the same, but there is much more to this.
![]() |
![]() |
Original Photo | OneDrive Version |
Original Version:
file 20181225-060457-SW1-000-orig.jpg 20181225-060457-SW1-000-orig.jpg: JPEG image data, Exif standard: [TIFF image data, little-endian, direntries=15, manufacturer=Panasonic, model=DMC-LX15, orientation=upper-left, xresolution=214, yresolution=222, resolutionunit=2, software=Ver.1.0 , datetime=2018:12:25 06:04:57], baseline, precision 8, 5472x3080, components 3
OneDrive Version:
file 20181225-060457-SW1-000-one.jpg 20181225-060457-SW1-000-one.jpg: JPEG image data, JFIF standard 1.01, resolution (DPI), density 180x180, segment length 16, Exif Standard: [TIFF image data, big-endian, direntries=18, description=Ayutthaya, the Ancient Capital of Thailand, manufacturer=Panasonic, model=DMC-LX15, orientation=upper-left, xresolution=294, yresolution=302, resolutionunit=2, software=Ver.1.0 , datetime=2018:12:25 06:04:57], baseline, precision 8, 5472x3080, components 3
Hmmm, the image container has been messed with. Lets check the Thumbnail and Metadata...
I extracted the thumbnail with exiftool.
exiftool -a -b -W thumb-orig-%f_%t%-c.%s -preview:all 20181225-060457-SW1-000-orig.jpg 1 output files created 1232353 bytes = thumb-orig-20181225-060457-SW1-000-orig_PreviewImage.jpg
Now for the OneDrive version:
exiftool -a -b -W thumb-orig-%f_%t%-c.%s -preview:all 20181225-060457-SW1-000-one.jpg 0 output files created JPEG EOI Warning = [minor] Error reading PreviewImage from file
Looks like the EXIF thumbnail has been REMOVED on the OneDrive version. That probably explains why the OneDrive file size is 1GB smaller. But this means OneDrive has DELETED information from my files.
Comparing the metadata for thumbnails:
Original Photo | OneDrive Version |
exiftool -All 20181225-060457-SW1-000-orig.jpg | egrep -i "preview|thumb" MP Image Type : Large Thumbnail (full HD equivalent) Preview Image : (Binary data 1232353 bytes) |
exiftool -All 20181225-060457-SW1-000-one.jpg | egrep -i "preview|thumb" MP Image Type : Large Thumbnail (full HD equivalent) Preview Image : (Binary data 1232353 bytes, use -b option to extract) |
Hmmm, metadata reckons the thumbnail is there, but it has been removed from the actual file.
Checking the MetaData on the OneDrive version of the file some new warnings have been introduced:
exiftool -validate -warning -error -a 20181225-060457-SW1-000-one.jpg Warning : [minor] Possibly incorrect maker notes offsets (fix by -320?) Warning : [minor] Non-standard ExifIFD tag 0xea1d OffsetSchema Warning : [minor] Fixed incorrect URI for xmlns:MicrosoftPhoto Warning : Missing required JPEG ExifIFD tag 0xa001 ColorSpace
Infact if you look at the MetaData using exiftool it seems to be a bit of a mess:
exiftool -v3 20181225-060457-SW1-000-one.jpg | | | 94) FaceRecInfo (SubDirectory) --> | | | - Tag 0x0061 (148 bytes, undef[148]): | | | 6f22: 20 00 41 00 6e 00 63 00 69 00 65 00 6e 00 74 00 [ .A.n.c.i.e.n.t.] | | | 6f32: 20 00 43 00 61 00 70 00 69 00 74 00 61 00 6c 00 [ .C.a.p.i.t.a.l.] | | | 6f42: 20 00 6f 00 66 00 20 00 54 00 68 00 61 00 69 00 [ .o.f. .T.h.a.i.] | | | 6f52: 6c 00 61 00 6e 00 64 00 00 00 53 00 74 00 65 00 [l.a.n.d...S.t.e.] | | | 6f62: 76 00 65 00 3b 00 54 00 68 00 61 00 69 00 6c 00 [v.e.;.T.h.a.i.l.] | | | [snip 68 bytes] | | | + [BinaryData directory, 148 bytes] | | | | FacesRecognized = 32 | | | | - Tag 0x0000 (2 bytes, int16u[1]): | | | | 6f22: 20 00 [ .] | | | | RecognizedFace1Name = n | | | | - Tag 0x0004 (20 bytes, string[20]): | | | | 6f26: 6e 00 63 00 69 00 65 00 6e 00 74 00 20 00 43 00 [n.c.i.e.n.t. .C.] | | | | 6f36: 61 00 70 00 [a.p.] | | | | RecognizedFace1Position = 105 116 97 108 | | | | - Tag 0x0018 (8 bytes, int16u[4]): | | | | 6f3a: 69 00 74 00 61 00 6c 00 [i.t.a.l.] | | | | RecognizedFace1Age = | | | | - Tag 0x0020 (20 bytes, string[20]): | | | | 6f42: 20 00 6f 00 66 00 20 00 54 00 68 00 61 00 69 00 [ .o.f. .T.h.a.i.] | | | | 6f52: 6c 00 61 00 JPEG EOI Warning = [minor] Error reading PreviewImage from file
What initially looks to be sinister Microsoft Face Recognition, is infact the XMP:Description "Ancient Captial of Thailand" being pushed to the wrong offset.
Click to see the full differences between the Original and OneDrive metadata using (exiftool -htmlDump).
Comparing the XMP and IPTC Metadata, we see that OneDrive has introduced some new MicrosoftPhoto xml fields in XMP metadata:
Original:
exiftool -a -s -G1 -XMP:All -IPTC:All 20181225-060457-SW1-000-orig.jpg [XMP-x] XMPToolkit : Image::ExifTool 12.05 [XMP-iptcExt] LocationShownSublocation : Pattaya [XMP-dc] Creator : Steve (photo@stargw.net) [XMP-dc] Description : Ayutthaya, the Ancient Capital of Thailand [XMP-dc] Rights : Steve (photo@stargw.net) [XMP-dc] Subject : Steve [XMP-exif] DateTimeOriginal : 2018:12:25 06:04:57+00:00 [IPTC] Caption-Abstract : Ayutthaya, the Ancient Capital of Thailand [IPTC] Keywords : Steve [IPTC] By-line : Steve (photo@stargw.net) [IPTC] Sub-location : Pattaya [IPTC] Credit : Steve (photo@stargw.net) [IPTC] CopyrightNotice : Steve (photo@stargw.net) [IPTC] Contact : Steve (photo@stargw.net) [IPTC] ApplicationRecordVersion : 4
OneDrive Version:
exiftool -a -s -G1 -XMP:All -IPTC:All 20181225-060457-SW1-000-one.jpg [XMP-rdf] About : uuid:faf5bdd5-ba3d-11da-ad31-d33d75182f1b [XMP-iptcExt] LocationShownSublocation : Pattaya [XMP-dc] Creator : Steve (photo@stargw.net) [XMP-dc] Rights : Steve (photo@stargw.net) [XMP-dc] Subject : Steve, Thailand [XMP-dc] Title : Ayutthaya, the Ancient Capital of Thailand [XMP-dc] Description : Ayutthaya, the Ancient Capital of Thailand [XMP-exif] DateTimeOriginal : 2018:12:25 06:04:57+00:00 [XMP-microsoft] LastKeywordXMP : Steve, Thailand [XMP-microsoft] LastKeywordIPTC : Steve, Thailand [IPTC] CodedCharacterSet : UTF8 [IPTC] Caption-Abstract : Ayutthaya, the Ancient Capital of Thailand [IPTC] Keywords : Steve, Thailand [IPTC] By-line : Steve (photo@stargw.net) [IPTC] Sub-location : Pattaya [IPTC] Credit : Steve (photo@stargw.net) [IPTC] CopyrightNotice : Steve (photo@stargw.net) [IPTC] Contact : Steve (photo@stargw.net) [IPTC] ApplicationRecordVersion : 4
Where did that XMP-microsoft come from? Looking at the raw XMP XML:
exiftool -XMP -b one.jpg 20181225-060457-SW1-000-one.jpg <rdf:Description rdf:about="uuid:faf5bdd5-ba3d-11da-ad31-d33d75182f1b" xmlns:MicrosoftPhoto="http://ns.microsoft.com/photo/1.0/"> <exif:DateTimeOriginal>2018-12-25T06:04:57+00:00</exif:DateTimeOriginal> </rdf:Description> <rdf:Description rdf:about="uuid:faf5bdd5-ba3d-11da-ad31-d33d75182f1b" xmlns:MicrosoftPhoto="http://ns.microsoft.com/photo/1.0/"> <MicrosoftPhoto:LastKeywordXMP> <rdf:Bag> <rdf:li>Steve</rdf:li> <rdf:li>Thailand</rdf:li> </rdf:Bag> </MicrosoftPhoto:LastKeywordXMP> </rdf:Description> <rdf:Description rdf:about="uuid:faf5bdd5-ba3d-11da-ad31-d33d75182f1b" xmlns:MicrosoftPhoto="http://ns.microsoft.com/photo/1.0/"> <MicrosoftPhoto:LastKeywordIPTC> <rdf:Bag> <rdf:li>Steve</rdf:li> <rdf:li>Thailand</rdf:li> </rdf:Bag> </MicrosoftPhoto:LastKeywordIPTC> </rdf:Description> </rdf:RDF>
Issues seem to occur when the original files metadata is changed, and then uploaded. OneDrive seems to merge back old tags, or reapply metadata that its holding elsewhere. For example the IPTC:Keywords in the original file used to be "Steve, Thailand" but I changed this a while ago to just "Steve". But OneDrive is retaining "Steve, Thailand". Or it could be completely random. There are probably other changes going on, but I have seen enough to worry me. The file I uploaded is not the file in OneDrive, and thats a problem for me.
In summary JPG files uploaded to OneDrive are changed by Microsoft. MetaData and Thumbnails are changed. Oddly this does not happen to all files all the time. There seems to be no underlying reason why some files are changed and not others. I can copy the file, rename it and upload it and the behaviour may be different. Its frequent enough that it is a significant problem.
I have been using rclone as a one way sync from my computer to OneDrive. When Microsoft changes the file, in the next rclone the originals are uploaded again. It would be even worse if I was using a two way sync or a Windows based sync. Microsoft would change the files, and your computer would then download the changed file. The originals would be lost (although they will be available to restore for a while).
I was using Microsoft OneDrive as a backup solution, but if files are being changed by Microsoft on OneDrive I cannot guarentee the integrity of the files on OneDrive! Note, I have only experienced this behaviour on JPG Image files. But I wonder what other files might get changed? Docs? Txt? Lack of File Integrity makes OneDrive useless as a backup solution.
05 May 2021
I opened case #:1022032721 with Microsoft.
26 Jul 2021
After some initial investigation, Microsoft went quiet and stopped responding to my requests for updates on the ticket.
Nov 2021
When my OneDrive subscription ended, I cancelled my subscription and
have been wary of cloud providers ever since!
Jul 2023
I regularly get emails from other OneDrive customers that experience the same issues I saw. So it looks like its still an issue, Microsoft knows about it but keeps quiet. I think a lot of people are unaware of what is going on behind the scenes to their files.
Contact Me: