-
Notifications
You must be signed in to change notification settings - Fork 590
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[FeG] Gx/Gy interfaces over SCTP with connections suddenly closed, then no interface/service auto-recovered #12067
Comments
Why don't you have local_address defined? You don't need to defin the IP, but you need the port. Local ports for GX, GY and S6a should be different. You could use I also suggest to update your complete setup to 1.7. The version is not official yet, but we area already testing it and it works properly. Note 1.5 is almost a year old and some of the bugs you may be experiencing are already fixed there.
|
Thanks @uri200. It's applied. I see in your example no gy.init_method. May we change it also? Side note: when trying via the API, I got
So performed the local_address addition via the web UI, that made me lose:
Now:
|
@uri200 update: the ABORTs are not solved. I'm not sure about the SHUTDOWNs because the high frequency of ABORTs is not allowing them to happen now. |
An almost relationship between:
And:
Extra:
|
can you share FeG pcaps? Maybe on a private channel. |
Sent @emakeev. Thanks! |
@crbertoldo I don't think the unit method has any impact here . I would suggest to use
To filter out the logs. Otherwise you will be collecting other services logs that can be mistaken. BTW, solution to this issue may be found at |
Thanks @uri200 |
@crbertoldo should we close the issue now? |
Yes @emakeev . Thanks again |
Your Environment
Describe the Issue
We have two issues that cause the described by the title of this issue:
/proc/net/sctp/assocs
the missing association(s). Restarting the FeG containers recover the interface(s).To Reproduce
Since the deployment is sorta particular, we suggest using our deployment (lab and/or prod) to take any information needed.
Expected behavior
The issues can be related to the network issues (like the drops for 1 above), to workload vs. provisioned computational resources, etc., but we would expect both logs showing why these SHUTDOWNs/ABORTs happen and better mechanisms by session_proxy to handle such "natural" failures (if so).
Additional context
command: envdir /var/opt/magma/envdir /var/opt/magma/bin/session_proxy -logtostderr=true -v=10
client, server
is not passed to connection.gomagma/feg/gateway/diameter/connection_manager.go
Line 63 in beac53c
The text was updated successfully, but these errors were encountered: