Re: [Dnsmasq-discuss] Very slow load time on systems with a large `ulimit -n`
On Mon, 2 Mar 2020 at 19:38, Simon Kelley wrote: > Thanks for the heads-up. > > > http://thekelleys.org.uk/gitweb/?p=dnsmasq.git;a=commit;h=0541a1adf7c25356e7bbd792a8d6ee217a0d92f3 > > followed by > > > http://thekelleys.org.uk/gitweb/?p=dnsmasq.git;a=commit;h=48755ebf093543113de96747a7f5f78e0640b333 > > fixes this, at least on Linux. > Great, thank you for the quick patch! Arthur ___ Dnsmasq-discuss mailing list Dnsmasq-discuss@lists.thekelleys.org.uk http://lists.thekelleys.org.uk/mailman/listinfo/dnsmasq-discuss
Re: [Dnsmasq-discuss] Very slow load time on systems with a large `ulimit -n`
On 02/03/2020 08:32, Arthur Darcet wrote: > Hello, > > As discussed here: https://github.com/pi-hole/FTL/issues/703 > > At startup, dnsmasq tries to close all file descriptors from 0 to > `ulimit -n` (the maximum number of open files) - here > http://thekelleys.org.uk/gitweb/?p=dnsmasq.git;a=blob;f=src/dnsmasq.c;h=573aac069b75b801ad1ddf5305af5010ca1ab88a;hb=HEAD#l151 > > On setups where this limit is high, this takes a large amount of time - > or even "hangs" indefinitely. Docker sets this limit to ~1M for instance > - and the kubernetes setup I am using (k3s) is setting it to "Infinity", > which translates to 1073741816. > > I would suggest switching over to something based on the actually open > files - like was done for the rpm package: cf > https://bugzilla.redhat.com/show_bug.cgi?id=1537564 and > https://github.com/rpm-software-management/rpm/pull/444/files for an > actual patch > > Work around is to run a `ulimit -n 1024` before starting dnsmasq - but > that's not really a clean solution. > > Thanks! > Arthur > > ___ > Dnsmasq-discuss mailing list > Dnsmasq-discuss@lists.thekelleys.org.uk > http://lists.thekelleys.org.uk/mailman/listinfo/dnsmasq-discuss > Thanks for the heads-up. http://thekelleys.org.uk/gitweb/?p=dnsmasq.git;a=commit;h=0541a1adf7c25356e7bbd792a8d6ee217a0d92f3 followed by http://thekelleys.org.uk/gitweb/?p=dnsmasq.git;a=commit;h=48755ebf093543113de96747a7f5f78e0640b333 fixes this, at least on Linux. Simon. ___ Dnsmasq-discuss mailing list Dnsmasq-discuss@lists.thekelleys.org.uk http://lists.thekelleys.org.uk/mailman/listinfo/dnsmasq-discuss
[Dnsmasq-discuss] Very slow load time on systems with a large `ulimit -n`
Hello, As discussed here: https://github.com/pi-hole/FTL/issues/703 At startup, dnsmasq tries to close all file descriptors from 0 to `ulimit -n` (the maximum number of open files) - here http://thekelleys.org.uk/gitweb/?p=dnsmasq.git;a=blob;f=src/dnsmasq.c;h=573aac069b75b801ad1ddf5305af5010ca1ab88a;hb=HEAD#l151 On setups where this limit is high, this takes a large amount of time - or even "hangs" indefinitely. Docker sets this limit to ~1M for instance - and the kubernetes setup I am using (k3s) is setting it to "Infinity", which translates to 1073741816. I would suggest switching over to something based on the actually open files - like was done for the rpm package: cf https://bugzilla.redhat.com/show_bug.cgi?id=1537564 and https://github.com/rpm-software-management/rpm/pull/444/files for an actual patch Work around is to run a `ulimit -n 1024` before starting dnsmasq - but that's not really a clean solution. Thanks! Arthur ___ Dnsmasq-discuss mailing list Dnsmasq-discuss@lists.thekelleys.org.uk http://lists.thekelleys.org.uk/mailman/listinfo/dnsmasq-discuss