Not handling OC_FORM_RESULT_NEWGROUP correctly was the reason for the incorrect DNS processing. Thanks for the hint! Fabian Am 16.04.2014 um 20:16 schrieb Kevin Cernekee <cernekee at gmail.com>: > On Wed, Apr 16, 2014 at 10:50 AM, Fabian J?ger > <fabian.jaeger at chungwasoft.com> wrote: >> Dear all, >> I have a user of Shimo (using openconnect internally) wo claims that CISCO_SPLIT_INC_* environment variables are not set correctly in 5.99 and that it was working with 5.02. >> >> Is there anything know related to such an issue? The corresponding thread in our forum can be found here: http://www.chungwasoft.com/support/community/discussion/936/openconnect-not-setting-routes-correctly#Item_3 > > From the linked post: > >> The values for INTERNAL_IP4_ADDRESS are also different when I connect using 3336 and 2587 > >> Also - there are X-CSTP-Split-Include variables set when connecting using 2587 that do not exist in 3336 as well > > Can we conclude that the gateway is actually sending different > X-CSTP-* data in the headers? So maybe a different authgroup is being > selected in 2587 versus 3336? > > Perhaps the user wanted "BCSI-SV-Split" but is actually getting logged > into "BCSI-SV-FULL"? > > Is it possible that sending the correct platform name "mac-intel" now > enables XML POST and this requires implementation of > OC_FORM_RESULT_NEWGROUP in the Shimo UI? > > If possible it would be good to run the stock openconnect CLI with > --dump-http-traffic for both cases, and compare the results. > -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/pkcs7-signature Size: 3088 bytes Desc: not available URL: <http://lists.infradead.org/pipermail/openconnect-devel/attachments/20140523/b6596713/attachment.p7s>
