Skip to content

G
gitlab-workhorse
Switch branch/tag
  1. 09 Dec, 2019 3 commits
  3. 06 Dec, 2019 1 commit
    • Stan Hu's avatar
      Log duration_ms when Workhorse hits a bad gateway · 43ce41ec
      Stan Hu authored 
      Previously when the upstream application (e.g. Unicorn) timed out,
Workhorse would report an error message of the form:

```json
{
  "correlation_id": "k4mKg0snXG3",
  "error": "badgateway: failed after 61s: EOF",
  "level": "error",
  "method": "GET",
  "msg": "error",
  "time": "2019-12-02T22:05:09-08:00",
  "uri": "/admin"
}
```

Since Sentry attempts to group similar events by some fingerprint that
includes the message, the changes in timings would trick Sentry into
creating a new event. To avoid that, the new log message now looks like:

```json
{
  "correlation_id": "k4mKg0snXG3",
  "duration_ms": 60599,
  "error": "badgateway: failed to receive response: EOF",
  "level": "error",
  "method": "GET",
  "msg": "error",
  "time": "2019-12-02T22:05:09-08:00",
  "uri": "/admin"
}
```

The inclusion of `duration_ms` and the standardized `error` message makes
it easier for users to filter and parse this error.
      43ce41ec
  5. 03 Dec, 2019 1 commit
  7. 27 Nov, 2019 6 commits
  9. 21 Nov, 2019 1 commit
  11. 20 Nov, 2019 2 commits
    • Nick Thomas's avatar
      Merge branch 'nuget-object-storage-upload-route' into 'master' · f407257d
      Nick Thomas authored 
      Add NuGet route for package uploads

See merge request gitlab-org/gitlab-workhorse!441
      f407257d
    • Nick Thomas's avatar
      Set a time limit on git upload-pack requests · f2ad577a
      Nick Thomas authored 
      When a client does a git fetch over HTTP, workhorse performs an access
check based on the HTTP request header, then reads the entire request
body into a temporary file before handing off to Gitaly to service it.
However, the client has control over how long it takes to read the
request body. Since the Gitaly RPC only happens once the request body
is read, people can set up a connection before their access is revoked
and use it to gain access to code committed days or weeks later.

To resolve this, we place an overall limit of 10 minutes on receiving
the `upload-pack` request body. Since this is over HTTP, the client is
using the `--stateless-rpc` mode, and there is no negotiation between
client and server. The time limit is chosen fairly arbitrarily, but it
fits well with the existing 10MiB limit on request body size, implying
a transfer speed of just 17KiB/sec to be able to fill that buffer and
get a "request too large" error instead of "request too slow".

Workhorse does not expose the `upload-archive` endpoint directly to the
user; the client in that case is always gitlab-rails, so there is no
vulnerability there.

The `receive-pack` endpoint is theoretically vulnerable, but Gitaly
performs a second access check in the pre-receive hook which defeats
the attack, so no changes are needed.

The SSH endpoints are similarly vulnerable, but since those RPCs are
bidirectional, a different approach is needed.
      f2ad577a
  13. 19 Nov, 2019 2 commits
  15. 11 Nov, 2019 2 commits
  17. 05 Nov, 2019 2 commits
  19. 25 Oct, 2019 2 commits
  21. 15 Oct, 2019 3 commits
  23. 14 Oct, 2019 1 commit
  25. 11 Oct, 2019 4 commits
  27. 10 Oct, 2019 4 commits
  29. 09 Oct, 2019 2 commits
  31. 07 Oct, 2019 2 commits
    • Nick Thomas's avatar
      Merge branch 'preserve-cache-headers-in-sendurl' into 'master' · 560ad42f
      Nick Thomas authored 
      Preserve original HTTP cache headers in sendurl

See merge request gitlab-org/gitlab-workhorse!428
      560ad42f
    • Sean McGivern's avatar
      Preserve original HTTP cache headers in sendurl · cfd8e4e8
      Sean McGivern authored 
      There are a few cases for serving uploads:

1. File storage. Rails asks Workhorse to serve the file.
2. Object storage. Rails redirects (via SendFileUpload) to the object
   storage host.
2. Object storage with `proxy_download` enabled. Rails asks Workhorse to
   proxy the download from the object storage host.

Rails also sets caching headers for uploads. In case 1, the reverse
proxy will keep those headers. In case 2, the headers are whatever the
object storage provider sets.

Case 3 is changed here. Previously, it would use the cache headers from
the object storage provider. Now, it keeps the cache headers from Rails
instead. This is better because:

1. Cache headers on the object storage provider can be hard to
   configure.
2. Even if we ask users to manually configure them, they may get it
   wrong and inadvertently allow private resources to be cached by
   proxies.
3. Even if we ask users to manually configure them and they get it
   right, they will also need to track any updates the Rails application
   makes to the cache headers it sends.

We could solve these by trying to automatically set the metadata policy
on the object storage bucket, which would also help with case 2
above. However, that has its own pitfalls. We could, for instance, say
that `uploads/-/system/user` is public with an expiry of five minutes,
and that's fairly straightforward. But then if we need to update that
policy in future to make it public with an expiry of one minute, we are
introducing coordination issues.

This would get even more complicated if we allowed caching uploads from
public projects. If the project's visibility changed, we'd need to
update the object storage metadata too. So it's a tricky problem, and
this is a relatively small code change to at least solve one case.
      cfd8e4e8
  33. 30 Sep, 2019 2 commits
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7