fixed data race at nodes deletion

[hellt]

No description provided.

[hellt]

@networkop there was a data race when you delete the lab with >2 nodes. In the meantime I will change the chan to use copies of the GenericContainer structs as a quick fix, but it would be nice to get to the root of it and understand the cause of the data race

[networkop]

@hellt do you have steps to reproduce?

[hellt]

@networkop , I got help from @karimra to have a fix that keeps the pointer channel
The culprit was with a pointer to the loop variable that got sent into the channel. As I understood Karims explanations, that variable is kept the same throughout the loop cycle, so we had the same pointer passed all over to the workers

https://stackoverflow.com/questions/49123133/sending-pointers-over-a-channel

in bf13970 I applied the suggested quick fix

[networkop]

so i guess this looks like a long-term fix? or did you still want me to find an alternative @hellt ?

[hellt]

that is a long term one, yes no actions are needed

…

On Sun, May 23, 2021 at 10:17 PM Michael Kashin ***@***.***> wrote: so i guess this looks like a long-term fix? or did you still want me to find an alternative @hellt <https://github.com/hellt> ? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#425 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABLKV5P3H7L6TUUHVIBE5XLTPFPELANCNFSM45GS5XSA> .