Hi,

On 6/11/21 10:36, al3xu5 via Dng wrote:
> I tried to install the new packages on Beowulf:
> - snetaid 1.0-1 package was istalled, but I have:
>
> $ sudo service snetaid status
> /usr/sbin/snetaid: error while loading shared libraries: libnetaid.so.1:
> cannot open shared object file: No such file or directory
>

> - indeed when trying to install the libnetaid 1.0-1 I get a dependency
>    error being required libc6>=2.29 ...

>
> Since now I have no time to rebuild snetaid sources on Beowulf, I decided
> to firstly migrate my system to Chimaera (as soon as possible) and then
> "retry" to install snetaid...

Packages have been tested only in chimaera. Anyway, i'm doing some changes in the code right now.
Notably i'm removing a double usage of the netlink socket in the daemon [*], which makes it redundant
and quite confusing (the global design is still a bit confusing, i know).

On the other hand, i also decided to restart snetaid via [**]:

execv(argv[0], params);

during the connection attempts in order to be able to redirect stdout to another tty. I've not been
able to do this redirection after reopening stdin, stdout and stderr so that the daemon or its child
processes can detach succesfully from the tty:

stdin  = fopen("/dev/null", "r");
stdout = fopen("/dev/null", "w+");
stderr = fopen("/dev/null", "w+");

Any hints at this point?

As soon as i finish these changes, i'll build an iso image with the live-sdk including the packages.
I'll let you know.

Cheers, and thanks again for your interest!

Aitor.


[*] The first one when making use of select(); the second one when making use of recvmsg() in nl_monitor.

[**] I know the risks of doing this as root, but i'm doing this in a safe mode. Note that execv will restart
snetaid keeping the same pid.


>
> _______________________________________________
> Dng mailing list
> Dng@???
> https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng