Closed ifraixedes closed 5 months ago
I don't understand why there are 2 tests failing with these new satellite versions.
Because of the errors it looks something related with storj-up
failures:
---- integration_upload_multipart_abort_and_list_parts_cursor stdout ----
thread 'integration_upload_multipart_abort_and_list_parts_cursor' panicked at 'write data multipart 0: Custom { kind: Other, error: Uplink(Internal("uplink: stream: metaclient: not enough nodes: not enough nodes: requested from cache 10, found 0\n\tstorj.io/uplink/private/metaclient.(*Client).BeginSegment:1047\n\tstorj.io/uplink/private/storage/streams.(*Store).PutPart:466\n\tstorj.io/uplink/private/stream.NewUploadPart.func1:45\n\tgolang.org/x/sync/errgroup.(*Group).Go.func1:57")) }', uplink/tests/uploads_test.rs:304:10
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
error: test failed, to rerun pass `--test uploads_test`
---- integration_upload_multipart_commit stdout ----
make: *** [Makefile:31: test-integration] Error 101
thread 'integration_upload_multipart_commit' panicked at 'write data multipart 1: Custom { kind: Other, error: Uplink(Internal("uplink: stream: metaclient: not enough nodes: not enough nodes: requested from cache 10, found 0\n\tstorj.io/uplink/private/metaclient.(*Client).BeginSegment:1047\n\tstorj.io/uplink/private/storage/streams.(*Store).PutPart:466\n\tstorj.io/uplink/private/stream.NewUploadPart.func1:45\n\tgolang.org/x/sync/errgroup.(*Group).Go.func1:57")) }', uplink/tests/uploads_test.rs:145:10
@elek @dlamarmorgan do you have any idea?
It says not enough nodes
. Storagenodes may not be started. I would execute the test locally and check the log of the docker-compose.
@elek I see that the Storagenodes fail to ping the satellite.
All the storagenode reports in their logs:
uplink-rust-storagenode-6 | 2023-04-13T17:27:14.665Z ERROR contact:service contact/service.go:106 ping satellite failed {"Process": "storagenode", "Satellite ID": "12whfK1EDvHJtajBiAUeajQLYcWqxcQmdYQU5zX5cCf6bAxfgu4", "attempts": 4, "error": "ping satellite: failed to ping storage node, your node indicated error code: 0, rpc: tcp connector failed: rpc: dial tcp [::]:28967: connect: cannot assign requested address", "errorVerbose": "ping satellite: failed to ping storage node, your node indicated error code: 0, rpc: tcp connector failed: rpc: dial tcp [::]:28967: connect: cannot assign requested address\n\tstorj.io/storj/storagenode/contact.(*Service).pingSatelliteOnce:147\n\tstorj.io/storj/storagenode/contact.(*Service).pingSatellite:101\n\tstorj.io/storj/storagenode/contact.(*Chore).updateCycles.func1:87\n\tstorj.io/common/sync2.(*Cycle).Run:160\n\tstorj.io/common/sync2.(*Cycle).Start.func1:77\n\tgolang.org/x/sync/errgroup.(*Group).Go.func1:75"}
Do you know what is doing this? The only changes in the docker-comose.yml
generated by storj-up
are bumping the edge
and the storj
image tag to the last published version.
@ifraixedes The docker compose file in this repo looks to be generated from a very old version of storj-up (perhaps even the v1 version). There are several ENV variables that have changed since the compose file in here was created. If possible my suggestion would be to recreate the docker compose file using the latest version of storj-up. If you need help with this, I can assist.
@dlamarmorgan
The docker compose file in this repo looks to be generated from a very old version of storj-up (perhaps even the v1 version). There are several ENV variables that have changed since the compose file in here was created. If possible my suggestion would be to recreate the docker compose file using the latest version of storj-up.
I regenerated yesterday with the v1.1.0 which is the one that right now we are using https://github.com/storj-thirdparty/uplink-rust/blob/main/Makefile#L57
I'd love to migrate to a higher version, but rather than targeting a specific commit hash, I'd like to target a specific tag. Could we release a new tag version?
A new version tag that could compile with Go 1.20 without disabling QUIC would be awesome and we could close https://github.com/storj/up/issues/83
The current workaround for that it's using -tags noquic
(https://github.com/storj-thirdparty/uplink-rust/blob/main/Makefile#L52)
Bump the versions of the storj-up satellite and edge to match the last published tag version.
Uncomment an end-to-end test assertion that was commented because the previous satellite version had a bug and it made the test not pass.
Instruction in one of the Makefile recipes had to be adjusted to work with the new images.
Also, fix a few typos.
Closes #50