Red Hat Talks Up debuginfod As The New Debug Info Web Server

[RahulSundaram]

His point is that there's no way to run a proxy on a network that has not interaction at all with the outside

Not true. A internal infrastructure can act as a proxy and as I pointed out in my original repo, debuginfod itself is optional. You can continue using debuginfo packages if you wish

[DrYak]

Yes, you already said that. What is your point?

His point is that there's no way to run a proxy on a network that has not interaction at all with the outside (as opposed to network that ARE connected to the outside through a proxy). In some premices, there is NO internet end-point to which the proxy could connect and be proxying to the closed internal network.

On the other hand, as others have pointed out, it is possible to run a debuginfod server yourself. Thus on such "never touches the outside" networks, it would be possible to run a local instance for debuging purpose and feed it with the good old "station wagon full of harddrives or optical discs" method.

[Damian_Ivanov]

There are many private networks which are not even physcaly connected to the public network.
If they will stop provide debuginfo packages it will be yet another reason to move away from RH.

And do not mention that OS like Solaris does not need any dist debuinfo resources to diagnose issues to the level allowing support to figure out what is wrong.

kloczek Is that was dtrace does? Are there no debuginfo packages on Solaris?

[RahulSundaram]

Thre are many networs which are phisically not connected to public network.

Yes, you already said that. What is your point?

[fche]

There are many private networks which are not even physcaly connected to the public network.
If they will stop provide debuginfo packages it will be yet another reason to move away from RH.

There is no plan to stop providing debuginfo packages. If anything, debuginfod makes it easier to operate in a private network, because if you can transfer the debuginfo packages to an internal server, you can run your own debuginfod server against that set of rpms, and then all your internal machines can get at it. It's designed so that you can run servers for yourself, your team, your org ... and still federate among them or to distros.

[fche]

So a copy of what Clear Linux has for already a long time?

Yeah, a similar flavour but debuginfod is more complex & ambitious in some ways. Thanks for bringing our attention to it!

https://sourceware.org/ml/elfutils-devel/2020-q1/msg00064.html

[kloczek]

Such private networks typically use a proxy to use the resources they want from outside their private network. This doesn't change anything for them given this is optional anyway

Thre are many networs which are phisically not connected to public network.

[RahulSundaram]

There are many private networks which are not even physcaly connected to the public network.
If they will stop provide debuginfo packages it will be yet another reason to move away from RH.

Such private networks typically use a proxy to use the resources they want from outside their private network. This doesn't change anything for them given this is optional anyway

[DrYak]

......

...except that the final production release will be taken over by Lennart Poettering and rolled out ad `systemd-debuginfod`.



(ducks and runs away)

[kloczek]

There are many private networks which are not even physcaly connected to the public network.
If they will stop provide debuginfo packages it will be yet another reason to move away from RH.

And do not mention that OS like Solaris does not need any dist debuinfo resources to diagnose issues to the level allowing support to figure out what is wrong.