[OpenWrt-Devel] ddns-scripts "sleep 10" before updating?

Christian Schoenebeck christian.schoenebeck at gmail.com
Fri Jan 16 18:25:10 EST 2015


Am 16.01.2015 um 17:50 schrieb Catalin Patulea:
> On Fri, Jan 16, 2015 at 4:36 AM, Christian Schoenebeck
> <christian.schoenebeck at gmail.com> wrote:
>> Did you really ever run into problems because of 10 seconds delay after an outage of thousands of seconds ?
> No, but I don't want to wait until I have a problem. ddns-scripts
> should do the best it can to update the record quickly.
> 
>> What do your ddns provider think about multiple updates per minute, if the connection toggle due to errors on your wan connection. Most providers start blocking for hours after to many updates.
> First you said the sleep was because netifd runs the script too early.
> That turned out to be false. Now you're saying it's to be nice to the
> ddns provider - I don't think it's a problem and sleep is the wrong
> way to solve it anyway.
> 
> As John Crispin recently said in a different patch review, "using
> magic delays is a no go":
> https://lists.openwrt.org/pipermail/openwrt-devel/2015-January/030561.html
> 
> That is exactly what this sleep 10 is - some arbitrary value that
> doesn't seem to have a specific purpose, added to fix one minority
> edge case - or maybe no one remembers why it was added.
> 
>> For the moment please comment out the 4 lines starting at line 219 of dynamic_dns_updater.sh
> No. I do not maintain one router, I maintain several of them. I will
> not go around editing dynamic_dns_updater.sh on them all, and have to
> do this again every time I flash a new version.
> 
> I do not want to maintain a diff compared to openwrt master either,
> because that will just give merge headaches in the future.
> 
> Please, what is the reason sleep 10 is still there? Was it because of
> the early netifd call - then let's just remove it. If it is only to
> fix the satellite user case, then maybe that use should be the one to
> modify dynamic_dns_updater.sh on their router. But right now, for the
> majority of cases, it adds a 10 second delay for no reason. I don't
> think this is right.
> 
Removed in 2.1.0-5
Christian
_______________________________________________
openwrt-devel mailing list
openwrt-devel at lists.openwrt.org
https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel



More information about the openwrt-devel mailing list