API request failed caused by: [56] Failure when receiving data - POST is failed

[aklevleeva]

Environment

self-hosted (https://develop.sentry.dev/self-hosted/)

What are you trying to accomplish?

Steps to Reproduce

curl https://github.com/getsentry/sentry-cli/releases/download/2.8.0/sentry-cli-Windows-x86_64.exe -o sentry-cli.exe -L
sentry-cli.exe --log-level=debug --url CUSTOM_URL login --auth-token TOKEN
sentry-cli.exe --log-level=debug --url CUSTOM_URL releases new NEW_RELEASE --org sentry -p TEST
sentry-cli.exe --log-level=debug --url CUSTOM_URL upload-dif -o sentry -p treat --wait OUTPUT --include-sources

Expected Result
auth - success
new release created - success
All pdb files are uploaded - success

Actual Result
auth - success
new release created - success
All pdb files are uploaded - failed with the below errors

Uploaded 881 missing debug information files
--
06-Nov-2023 11:27:17 | error: API request failed
06-Nov-2023 11:27:17 | caused by: [56] Failure when receiving data from the peer (seek callback returned error 2)
06-Nov-2023 11:27:17 | DEBUG   2023-11-06 11:27:17.720935800 +00:00 skipping update nagger because session is not attended

How are you getting stuck?

This error is a temporary one and we don't experience it constantly. Is started to fail unexpectedly and we don't see any related errors or significant load increase in the sentry logs.

Where in the product are you?

APIs

Link

No response

DSN

No response

Version

23.10.1

[getsantry]

Assigning to @getsentry/support for routing ⏲️

[getsantry]

Assigning to @getsentry/support for routing ⏲️

[getsantry]

Routing to @getsentry/product-owners-releases for triage ⏲️

[getsantry]

Routing to @getsentry/product-owners-settings-auth for triage ⏲️

[getsantry]

Routing to @getsentry/product-owners-releases for triage ⏲️

[malwilley]

@aklevleeva do you know why API endpoint is failing and with what response? Since you are using self-hosted it will be difficult for us to debug your problem without knowing that.

[scttcper]

can you try again on a newer version of sentry cli. the newest version is 2.21.3

[aklevleeva]

Hi @malwilley, @scttcper

It is a flow issue - caught time from time. There is no reason to fail the endpoint with such error. I have provided the response and the error in the description. There are 3 commands and the error occurs at the last one.
I will try to execute the test with the latest cli version to make sure it can bring the permanent solution.
Please let me know if you have other ideas about what else could be checked.

Thanks, Aneliya

[krzys-h]

I'm running into the same issue on our Windows build VM. I think it happened always, but somehow seems to have gotten worse today, for no particular reason. The API endpoint doesn't seem to matter, I can replicate the issue with both sentry-cli info, sentry-cli releases list as well as during symbol upload from within the Unity sentry plugin.

(sorry for this being a screenshot, I don't have an easy way to copy text out of that VM)

Things I tried to far:

• Upgrading sentry-cli to 2.30.2 - no change
• Upgrading sentry-selfhosted to 24.3.0 - no change
• Using sentry-cli on a different machine in the same network (a laptop running Linux) - the issue doesn't happen there, so perhaps it's Windows-specific? (since @aklevleeva seems to be using Windows as well)
• Running with debug log level - nothing indicating why the request fails
  
• Sending the request manually with curl (or rather the PowerShell Invoke-WebRequest cmdlet which is aliased to curl on Windows) - that seems to work, suggesting a sentry-cli issue
• Connecting to the public sentry instance instead of the self-hosted one - doesn't seem to happen on sentry.io, only on self-hosted
• I don't see anything interesting in the server-side logs - only a normal 200 response to the API request, which somehow doesn't reach the client machine

tl;dr - seems to be a Windows-only sentry-cli issue that happens only on self-hosted instances.

In case it turns out to be relevant: our self-hosted instance is using basically the default config, the only change is that we are using traefik as a reverse proxy instead of exposing the port directly:

   nginx:
     <<: *restart_policy
-    ports:
-      - "$SENTRY_BIND:80/tcp"
+    #ports:
+    #  - "$SENTRY_BIND:80/tcp"
+    labels:
+      - traefik.enable=true
+      - traefik.http.routers.sentry.rule=Host(`sentry.***`)
+      - traefik.http.routers.sentry.entrypoints=websecure

[JaredEzz]

I'm encountering this during local debugging on windows as well

[hubertdeng123]

I'm going to transfer this to sentry-cli then

[szokeasaurusrex]

Hi all, looks like the error message is originating from the curl library, so perhaps there is some underlying bug there.

It looks like the latest CLI release is using an older version of curl. We bumped curl to the latest version a week ago in 9cf9afb, and I will release a new Sentry CLI version early next week which includes that change.

Let's see whether bumping the dependency solves the problem before investigating further.

[szokeasaurusrex]

As of Sentry CLI v2.32.2, we are using curl version 0.4.46. Please let me know whether upgrading to Sentry CLI v2.32.2 fixes this problem!

[getsantry]

This issue has gone three weeks without activity. In another week, I will close it.

But! If you comment or otherwise update it, I will reset the clock, and if you remove the label Waiting for: Community, I will leave it alone ... forever!

---

"A weed is but an unloved flower." ― Ella Wheeler Wilcox 🥀