Safe to add TC_ENABLE=1 to tc_stop() function?

[Anders Peter Fugmann]

Bastian Kleineidam wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hi,
> 
> I see that in "/etc/init.d/fiaif tc-stop" there is a TC_ENABLE=1 before
> calling tc_stop. Shouldn't this be always set when calling tc_stop()?
> For example when fiaif is started with traffic shaping enabled, then
> the config is changed to disable it. After that a call to "fiaif stop"
> will call tc_stop with TC_ENABLED=0, which does not remove the qdiscs.
tc-start and tc-stop are debug commands, I use when debugging traffic 
shaping. These functions overrides the global traffic shaping settings.

If you also override the ENABLE_TC when calling 'fiaif stop', then fiaif 
might remove qdiscs from a 3. party traffic-shaping, which is not that 
nice.

Therefore I am a little hesitant saying yes here.

If you question is based on Bug#343146 in the debian bug database, then 
the error described is another one.

The new version of FIAIF does not support the CBQ based traffic shaping 
any more, but as the syntax check for fiaif configuration files is 
rather simple, it sees the line "TC_TYPE=CBQ" in the zone configuration 
file and complains. The configuration file should be updated to read
"TC_TYPE=HTB" (actually it should be obsoleted, as the user is forced to 
use HTB now). There will be a new version of FIAIF soon, where I will 
ignore the line. Also documentation needs to be updated.


> 
> So is it safe always to have TC_ENABLED=1 in tc_stop() function to
> ensure the qdiscs are removed, or could some other part of traffic
> shaping break with this setting?
No other part of FIAIF would break, but 3. party programs might.

> 
> 
> Regards,
>   Bastian
Regards
Anders Fugmann

P.s. Thanks for the good work for maintaining FIAIF in debian.