# How do I repair a repository with missing objects? **URL:** \ **Category:** Support\ **Created:** [June 3, 2023, 7:26pm UTC](https://kopia.discourse.group/t/how-do-i-repair-a-repository-with-missing-objects/2067 "2023-06-03T19:26:20Z")\ **Posts on this page:** 6\ **Page:** 1
**Author:** ![RunicGear](https://avatars.discourse-cdn.com/v4/letter/r/919ad9/32.png) [@RunicGear](https://kopia.discourse.group/u/RunicGear)\ **Post date:** [June 3, 2023, 7:26pm UTC](https://kopia.discourse.group/t/how-do-i-repair-a-repository-with-missing-objects/2067/1 "2023-06-03T19:26:20Z")
I have a `filesystem` repository on a USB external hard drive that had some errors. I suspect a bumped cable. (I do plan to change this later.) When I run `kopia snapshot verify --verify-files-percent=100`, I receive: ```plaintext Processed 22199 objects. ERROR unable to add 33142a2bc4e80c95d96fd49ed8862b61.0.3.0 to contents: can't get mod time for file "/var/cache/kopia/contents/33/142a2bc4e80c95d96fd49ed8862b61.0.3.0.f": stat /var/cache/kopia/contents/33/142a2bc4e80c95d96fd49ed8862b61.0.3.0.f: no such file or directory Processed 22450 objects. ``` I have a S3 remote that I regularly `sync-to`, and my first thought was to reverse that sync. The first time I did so, it wanted to re-download almost the entire repository. I let that run, ran maintenance, and found the problem was still present. Doing it again now seems much more reasonable: ```plaintext Source: S3: endpoint-redacted.com bucket-redacted Destination: Filesystem: /home/backups/data/kopia NOTE: By default no BLOBs are deleted, pass --delete to allow it. Looking for BLOBs to synchronize... Found 11815 BLOBs in the destination repository (269.6 GB) Found 11933 BLOBs (269.8 GB) in the source repository, 268 (3.3 GB) to copy Found 0 BLOBs to delete (0 B), 11665 in sync (266.5 GB) ``` However, this did not result in restoring the missing blobs. When I tried using `kopia repository repair filesystem --path ...`, I received ```plaintext looking for format blob... format blob already exists, not recovering, pass --recover-format=yes ``` I have not lost any data on the source systems. I don’t need to restore (today…), but I want to bring the repository back into a clean state for the future. How can I get the repository back into a clean state? ---
**Author:** ![RunicGear](https://avatars.discourse-cdn.com/v4/letter/r/919ad9/32.png) [@RunicGear](https://kopia.discourse.group/u/RunicGear)\ **Post date:** [June 9, 2023, 5:07pm UTC](https://kopia.discourse.group/t/how-do-i-repair-a-repository-with-missing-objects/2067/2 "2023-06-09T17:07:14Z")
I’ve now also tried ```sh kopia snapshot fix invalid-files --verify-files-percent=100 --commit ``` to no effect. ```sh kopia snapshot verify --verify-files-percent=100 ``` still shows a ton of ```plaintext ERROR unable to add d816xxxxxxxxxxxxxxxxxxxxxxxx78ee.1100.3.0 to contents: can't get mod time for file "/var/cache/kopia/contents/d8/16xxxxxxxxxxxxxxxxxxxxxxxx78ee.1100.3.0.f": stat /var/cache/kopia/contents/d8/16xxxxxxxxxxxxxxxxxxxxxxxx78ee.1100.3.0.f: no such file or directory ``` and yet running `fix invalid-files` again shows the same errors and leaves the snapshots `unchanged`. Why would the repository still be trying to reference invalid objects after they were removed from all snapshots? Why is `fix invalid-files` both encountering missing objects and not stubbing them out as requested? ---
**Author:** ![ZebCorp](https://avatars.discourse-cdn.com/v4/letter/z/5fc32e/32.png) [@ZebCorp](https://kopia.discourse.group/u/ZebCorp)\ **Post date:** [June 10, 2023, 1:59pm UTC](https://kopia.discourse.group/t/how-do-i-repair-a-repository-with-missing-objects/2067/3 "2023-06-10T13:59:10Z")
Well, I get the same errors, while verification: ERROR cache unable to add f692fc351f8e0755ddb4827fb5cdaf90.1100.2.0 to contents: can’t get mod time for file “C:\Users\xxx\AppData\Local\kopia\05e5535357e49791\contents/f6/92fc351f8e0755ddb4827fb5cdaf90.1100.2.0.f”: CreateFile C:\Users\xxx\AppData\Local\kopia\05e5535357e49791\contents/f6/92fc351f8e0755ddb4827fb5cdaf90.1100.2.0.f: The system cannot find the file specified. ERROR cache unable to add 27b174791e77a6a598304cd2ae1c8f6a.1100.2.0 to contents: can’t get mod time for file “C:\Users\xxx\AppData\Local\kopia\05e5535357e49791\contents/27/b174791e77a6a598304cd2ae1c8f6a.1100.2.0.f”: CreateFile C:\Users\xxx\AppData\Local\kopia\05e5535357e49791\contents/27/b174791e77a6a598304cd2ae1c8f6a.1100.2.0.f: The system cannot find the file specified. But what it means exactly, I’m not sure. Something with cache (?) because those missing (?) files are in the local cache directory. ---
**Author:** ![RunicGear](https://avatars.discourse-cdn.com/v4/letter/r/919ad9/32.png) [@RunicGear](https://kopia.discourse.group/u/RunicGear)\ **Post date:** [June 11, 2023, 3:19pm UTC](https://kopia.discourse.group/t/how-do-i-repair-a-repository-with-missing-objects/2067/4 "2023-06-11T15:19:13Z")
If the files exist for you, it may not be the same problem I have. If your repository is not otherwise corrupt, try deleting your cache and seeing what happens. I find it odd that it switches from windows `\` path separators to linux `/` ones half-way though. I wonder if that’s related/expected. ---
**Author:** ![ZebCorp](https://avatars.discourse-cdn.com/v4/letter/z/5fc32e/32.png) [@ZebCorp](https://kopia.discourse.group/u/ZebCorp)\ **Post date:** [June 11, 2023, 3:54pm UTC](https://kopia.discourse.group/t/how-do-i-repair-a-repository-with-missing-objects/2067/5 "2023-06-11T15:54:11Z")
Repository seems to be OK. Apart of those errors, nothing else happens. Repository sits on the usb, so cache is always created from scratch before backup/verification. ---
**Author:** ![basldfalksjdf](https://avatars.discourse-cdn.com/v4/letter/b/ce73a5/32.png) [@basldfalksjdf](https://kopia.discourse.group/u/basldfalksjdf)\ **Post date:** [June 19, 2023, 4:09pm UTC](https://kopia.discourse.group/t/how-do-i-repair-a-repository-with-missing-objects/2067/6 "2023-06-19T16:09:06Z")
There may not be anything wrong with your reporisotiry. `snapshot verify` appears to be broken since v0.12.1. See [snapshot verify shows processed 0 objects, with or without percent parameters · Issue #2967 · kopia/kopia · GitHub](https://github.com/kopia/kopia/issues/2967)