https://flyte.org logo
Join Slack
Powered by
Hey guyz, I did a setup of flyte on GCP. I'm able ...
# flyte-support
e
Hey guyz, I did a setup of flyte on GCP. I'm able to access the UI but on running pyflyte run --remote sample_workflow.py main --x 10 --y 20 I'm getting this error,
Copy code
grpc._channel._InactiveRpcError: <_InactiveRpcError of RPC that terminated with:
        status = StatusCode.UNAVAILABLE
        details = "failed to connect to all addresses; last error: UNKNOWN: ipv4:172.25.0.46:443: Ssl handshake failed: SSL_ERROR_SSL: error:1000007d:SSL routines:OPENSSL_internal:CERTIFICATE_VERIFY_FAILED"
        debug_error_string = "UNKNOWN:failed to connect to all addresses; last error: UNKNOWN: ipv4:172.25.0.46:443: Ssl handshake failed: SSL_ERROR_SSL: error:1000007d:SSL routines:OPENSSL_internal:CERTIFICATE_VERIFY_FAILED {grpc_status:14, created_time:"2023-02-07T18:47:06.951145+05:30"}"
Any body has an idea what is the problem here? P.s I have TLS disabled on ingress.
a
Hi @echoing-river-3611! Do you have 
admin.insecure: True
set in your 
~/.flyte/config.yaml
?
e
no it's set to false
Copy code
admin:
  endpoint: "dns:///"
#  authType: Pkce
  insecure: false
  insecureSkipVerify: true
console:
  endpoint: ""
logger:
  show-source: true
  level: 0
if i set it to true i'm not even able to use 
flytectl get projects
@agreeable-kitchen-44189 any suggestion on this?
a
Not really, sorry!
e
<!here> Anybody else can help with this?
a
@echoing-river-3611 what IdP are you using for authentication?
Looking at previous messages in this channel, some users have experienced this error and adding the redirect URI to the IdP configuration has been helpful.
e
We've deployed flyte in GKE with gsa-flyteadmin service account apart from that we're not using any IdP
f
This is unauthenticated
Cc @high-accountant-32689 / @stocky-notebook-88311 seems to be similar problem
Grpc is having some ssl trouble in case of your own org signed certs
In python and flytectl works
@echoing-river-3611 problem is we are unable to reproduce this
Can you write pip show Grpcio
e
Copy code
Name: grpcio
Version: 1.51.1
Summary: HTTP/2-based RPC framework
Home-page: <https://grpc.io>
Author: The gRPC Authors
Author-email: <mailto:grpc-io@googlegroups.com|grpc-io@googlegroups.com>
License: Apache License 2.0
Location: "<package-path>"
Requires: 
Required-by: flytekit, grpcio-status
@freezing-airport-6809 ^ o/p of pip show grpcio
h
@echoing-river-3611, just to confirm, those are self-signed certs you're using, right? If not, how did you generate the certs?
s
Yes. This is the same issue.
Self signed cert is not liked by python grpc client
e
what do we mean by self-asigned cert? cert to access the cluster?
f
When you access the console from the browser, do you get "Your connection is not private"?
s
by self signed I meant that the one signed by my org CA, not Mozilla or Digicert and the likes
e
that's only the case if i user https://<dns>/console but i use http://
f
I see, when you access using 
https
, can you click on the 
not secure
Can you click on the 
certificate is not valid
it should show you the CA
e
yaa it says certificate is not valid
f
If you click on Certificate is not valid, it will show you the CA:
e
yaa it shows something
f
What does 
Issue by
-> 
Common Name (CN)
say?
e
k8s ingress controller cert
f
ok, that seems like a self signed cert
Do you have the option to download the ca-bundle ?
e
Nope
f
What ingress controller are you using? Are you able to put a custom cert there?
e
ngnix ingress
btw i was able to download .cer fike
*file
f
ok, in your config file, you can try pointing to it:
Copy code
admin:
  
  caCertFilePath: /path/to/ca-bundle.crt
e
but flytectl is working with insecureSkipVerify : true
it just pyflyte cli that's causing the issue
do they not use same config?
f
they are.. but with the bundle you can enable the verify
We had the same issues with pyflyte and we were able to fix using the proper certs
e
certificate is valid for ingress.local not <dns> error
🤦 1
f
i am curious, if you have a non-ssl endpoint available, why is SSL happening?
e
the dns is only accessible over a VPN is this bcz of that?
and also one more question when i explicitly mentioned insecureSkipVerify: true why pyflyte not taking that into consideration?
f
are you using the non-ssl port in flyte configs?
Python GRPC doesn't behave well with skipping certificate verification..
I would try this: use the non-ssl endpoint with 
admin.insecure: True
e
hey, can we package the whole module using 
pyflyte package
i tried it and it throws an error i.e. submodule not found? P.s We're still not able to resolve ssl issue so can't use pyflyte register bcz of that
t
pyflyte package
should work. Could you share the exact error?
e
ModuleNotFoundError: No module named 'utils'
utils is a submodule in the module
t
Does the package you're sending to 
pyflyte package
contain 
utils
?
e
Copy code
pyflyte --pkgs dynamic_workflow package --image <image-name> -f -o dynamic_workflow.tgz
command for reference
Does the package you're sending to 
pyflyte package
contain 
utils
?
yes
@tall-lock-23197 am i using the wrong command or is there a another way to package the whole module?
t
Can you do a relative import? 
from .utils ...
e
the workflow did ran on doing relative import. but propeller is throwing this error 
Error when trying to reconcile workflow
 and in the dynamic task it was suppose to create 5 different task as i have given 5 input it did create those but it's been 20 min it is still stuck in queued and propeller is throwing above mentioned error and only one pod for dynamic task was created expecting 5
a
Hi @echoing-river-3611! Sorry, this is slightly unrelated, but do you also see 
rpc error: code = InvalidArgument desc = missing project
somewhere in your propeller logs?
e
yess
Copy code
[Feb 8, 2023 6:20:04 PM GMT+5]
{"json":{"exec_id":"ahlrb9l95q2h4gg8fhgf","node":"n0/dn0","ns":"development","res_ver":"55914809","routine":"worker-14","wf":"flyteexamples:development:dynamic_workflow.main.dynamic_workflow"},"level":"error","msg":"handling parent node failed with error: InvalidArgument: Invalid fields for event message, caused by [rpc error: code = InvalidArgument desc = missing project]","ts":"2023-02-08T12:50:04Z"}
[Feb 8, 2023 6:20:04 PM GMT+5]
{"json":{"exec_id":"ahlrb9l95q2h4gg8fhgf","node":"n0/dn0","ns":"development","res_ver":"55914809","routine":"worker-14","wf":"flyteexamples:development:dynamic_workflow.main.dynamic_workflow"},"level":"error","msg":"failed Execute for node. Error: InvalidArgument: Invalid fields for event message, caused by [rpc error: code = InvalidArgument desc = missing project]","ts":"2023-02-08T12:50:04Z"}
[Feb 8, 2023 6:20:04 PM GMT+5]
{"json":{"exec_id":"ahlrb9l95q2h4gg8fhgf","node":"n0","ns":"development","res_ver":"55914809","routine":"worker-14","wf":"flyteexamples:development:dynamic_workflow.main.dynamic_workflow"},"level":"error","msg":"handling dynamic subnodes failed with error: InvalidArgument: Invalid fields for event message, caused by [rpc error: code = InvalidArgument desc = missing project]","ts":"2023-02-08T12:50:04Z"}
[Feb 8, 2023 6:20:04 PM GMT+5]
{"json":{"exec_id":"ahlrb9l95q2h4gg8fhgf","node":"n0","ns":"development","res_ver":"55914809","routine":"worker-14","wf":"flyteexamples:development:dynamic_workflow.main.dynamic_workflow"},"level":"error","msg":"failed Execute for node. Error: InvalidArgument: Invalid fields for event message, caused by [rpc error: code = InvalidArgument desc = missing project]","ts":"2023-02-08T12:50:04Z"}
[Feb 8, 2023 6:20:04 PM GMT+5]
{"json":{"exec_id":"ahlrb9l95q2h4gg8fhgf","ns":"development","res_ver":"55914809","routine":"worker-14","wf":"flyteexamples:development:dynamic_workflow.main.dynamic_workflow"},"level":"warning","msg":"Error in handling running workflow [InvalidArgument: Invalid fields for event message, caused by [rpc error: code = InvalidArgument desc = missing project]]","ts":"2023-02-08T12:50:04Z"}
[Feb 8, 2023 6:20:04 PM GMT+5]
{"json":{"exec_id":"ahlrb9l95q2h4gg8fhgf","ns":"development","res_ver":"55914809","routine":"worker-14","wf":"flyteexamples:development:dynamic_workflow.main.dynamic_workflow"},"level":"error","msg":"Error when trying to reconcile workflow. Error [InvalidArgument: Invalid fields for event message, caused by [rpc error: code = InvalidArgument desc = missing project]]. Error Type[*errors.EventError]","ts":"2023-02-08T12:50:04Z"}
[Feb 8, 2023 6:20:04 PM GMT+5]
E0208 12:50:04.456196 1 workers.go:102] error syncing 'development/ahlrb9l95q2h4gg8fhgf': InvalidArgument: Invalid fields for event message, caused by [rpc error: code = InvalidArgument desc = missing project]
error logs ^
a
Ok, I feel like this might be related to this thread. I am currently trying to debug this in propeller
cc @hallowed-mouse-14616
e
so inshort dynamic workflow always faces this issue? though i tried it on sandbox it worked
just on remote cluster ^
a
I’ve just tried and seems like the error is in flytekit somewhere. Downgrading to 1.2.7 worked for me
I think the offending PR is this one. cc @high-accountant-32689
e
Hey @agreeable-kitchen-44189 does map_task also face same issue?
f
I think this is dynamic. I am surprised this made through the as of indeed a bug. It should not need a project
Project/domain should not matter for dynamic tasks
e
I meant i know dynamic has this problem but bcz map_task also spawns multiple pods will that also face same problem
f
We will follow up, please downgrade for now
Cc @high-accountant-32689
e
Copy code
{"json":{"exec_id":"as6j9bshk8qx5rfc6x7f","node":"n0","ns":"development","res_ver":"56079374","routine":"worker-0","wf":"flyteexamples:development:dynamic_workflow.main.dynamic_workflow"},"level":"error","msg":"failed Execute for node. Error: InvalidArgument: Invalid fields for event message, caused by [rpc error: code = InvalidArgument desc = missing project]","ts":"2023-02-08T16:14:31Z"}
{"json":{"exec_id":"as6j9bshk8qx5rfc6x7f","ns":"development","res_ver":"56079374","routine":"worker-0","wf":"flyteexamples:development:dynamic_workflow.main.dynamic_workflow"},"level":"warning","msg":"Error in handling running workflow [InvalidArgument: Invalid fields for event message, caused by [rpc error: code = InvalidArgument desc = missing project]]","ts":"2023-02-08T16:14:31Z"}
@freezing-airport-6809 @agreeable-kitchen-44189 still the same error even after downgrading flytekit to 1.2.7. do i also need to downgrade flytepropeller / <all the deployments> docker image to the same version?
h
@echoing-river-3611, can you double-check you're running flytekit 1.2.7 again? This should not happen if you're using flytekit<1.3.0
e
yup it's 1.2.7
h
ok, and how are you registering your workflows+tasks?
e
Copy code
pyflyte --pkgs dynamic_workflow.main package --image <image-name> -f -o dynamic_workflow.tgz

flytectl register files --project flyteexamples --domain development --archive dynamic_workflow.tgz --version v2
^ @high-accountant-32689
h
sorry, I meant the version of flytekit installed in the image.
e
oh checking that one sec
oh my bad i didn't change flytekit version in image after changing that too to 1.2.7 I no longer face the error thanks for the help
👍 1
h
@echoing-river-3611, @agreeable-kitchen-44189, we just released flytekit 1.3.2 which should have a fix for this. We'll be following up on https://github.com/flyteorg/flyte/issues/3324 for the long term fix.
🙌 2
🙌🏽 1
f
I would also say, we are extremely sorry to have released this. this was a complete miss and break in our process
❤️ 1
e
Hey, isn't there any way of doing a insecure 
grpc
connection through 
pyflyte
?
h
@echoing-river-3611,you should be able to set 
insecure: true
in the config file. What error are you seeing?
e
Copy code
grpc._channel._InactiveRpcError: <_InactiveRpcError of RPC that terminated with:
        status = StatusCode.UNAVAILABLE
        details = "failed to connect to all addresses; last error: INTERNAL: Trying to connect an http1.x server"
        debug_error_string = "UNKNOWN:Failed to pick subchannel {created_time:"2023-02-09T18:57:01.276539+05:30", children:[UNKNOWN:failed to connect to all addresses; last error: INTERNAL: Trying to connect an http1.x server {grpc_status:14, created_time:"2023-02-09T18:57:01.27653+05:30"}]}"
sorry just saw the message this is the error i'm getting for 
pyflyte
if i put 
insecure:true
and for 
flytectl
,
Copy code
rpc error: code = Unavailable desc = connection closed before server preface received
but error with 
flytectl
goes away if i use 
insecure: false
and 
insecureSkipVerify: true
but than i get 
openssl handshake failed
for 
pyflyte
in that case P.s 
TLS
disabled Hope I was able to explain the problem😅
@high-accountant-32689 Any suggestion?
f
@echoing-river-3611 this is what we were discussing - seems you have a self signed cert. thr cert does not follow the rfc and python Grpc client does not allow this
@high-accountant-32689 did you have other suggestions
382 Views
Open in Slack
PreviousNext
Powered by