Andrew, after a closer look I see l2tp and ppp configuration file could be in the form 'hostname + "." + config_file' in the test directory.
ikev1-l2tp-01/north.xl2tpd.conf they are not copied using the function that got fixed in 1e0b14ec0b. Now I am confused why this commit added 'hostname + "." + config_file' That seems wrong to me. Why do we need it? the test directory should only allow one config. If not found fall back to host/ and then all/ May be there were bugs before 1e0b14ec0b as it is now it looks bad to me. I don't have a good fix. So I will leave it for now. eff59a46350f is only a hack to prevent a common error. On Mon, Sep 21, 2020 at 07:12:18PM +0200, Antony Antony wrote: > Hi Andrew, > > I do not quite follow your arguments. > > when did wel allow west.ipsec.secrets to work? It should not work! I had > quick look at v3.30 west.ipsec.secrets would not be copied instead > /etc/baseconfig/host/etc/ipsec.secerts will be copied as /etc/ipsec.secrets. > > For host specific secerts file only allow west.secrets > > May be you are confusing with weststrongswan.conf some other combinations. > it is good to clean up the logic. > > I added support for ipsec.secrets, which be coppied all test hosts, and not > copy when there is ipsec.secrets and west.secrets (any host specific) files. > > Most likely Paul still dislike ipsec.secrets:( as far I am concerned that is > a different discussion for another day. > > On Mon, Sep 21, 2020 at 12:37:57PM -0400, Andrew Cagney wrote: > > FYI, I removed the code because it seemed arbitrary. It would reject some, > > but > > not all combinations of: > > west.ipsec.secrets > > westipsec.secrets > > west.secrets > > ipsec.secrets > > I figured reducing this list to just: > > west.ipsec.secrets > > ipsec.secrets > > (and perhaps only allowing one) + logging the result was for a later pass. > > > > On Mon, 21 Sep 2020 at 11:34, Antony Antony <ant...@vault.libreswan.fi> > > wrote: > > > > New commits: > > commit eff59a46350f2e638f1ef5051ab6f7a29033e5cf > > Author: Antony Antony <ant...@phenome.org> > > Date: Mon Sep 21 15:24:13 2020 +0000 > > > > testing: swan-prep break on conflicting config files > > > > fix 1e0b14ec0b20c3 this is a nice improvement, however, allowing > > conflicting files and picking the first one is less than ideal. > > We would end up with test cases with reduntant config files and > > endless > > confusion. Even with this fix I suspect 1e0b14ec0b20c3 would allow > > more > > config files than what a test would use. > > > > This is a first step to refuse to copy config file when there are > > conflicting files. The checks could stricter than this. > > > > _______________________________________________ > > Swan-commit mailing list > > swan-com...@lists.libreswan.org > > https://lists.libreswan.org/mailman/listinfo/swan-commit > > > > > _______________________________________________ > > Swan-dev mailing list > > Swan-dev@lists.libreswan.org > > https://lists.libreswan.org/mailman/listinfo/swan-dev > > _______________________________________________ > Swan-dev mailing list > Swan-dev@lists.libreswan.org > https://lists.libreswan.org/mailman/listinfo/swan-dev _______________________________________________ Swan-dev mailing list Swan-dev@lists.libreswan.org https://lists.libreswan.org/mailman/listinfo/swan-dev
