[UPDATE: No bug. PEBKAC error; bad file permissions were the culprit.]
IDS 14.10.FC5 on CentOS 7 x86_64
We recently switched from using GSKit/CSM key databases to OpenSSL/p12 key databases. Since then, when running locally on the host where the DB resides, regular users can still connect using client.p12/client.sth, but PAM users who attempt to connect get :
$ ls -la $INFORMIXDIR/etc/client.???
-rw-------. 1 informix informix 4798 Jan 15 21:10 /informix/products/TJGTEST/etc/client.p12
-rw-------. 1 informix informix 193 Jan 15 21:10 /informix/products/TJGTEST/etc/client.sth
$ echo "SELECT COUNT(*) FROM systables;" | dbaccess tjgdb
28014: Secure Sockets Layer error: cannot initialize GSKit environment/GSK_ERROR_BAD_KEYFILE_PASSWORD (GSK.
If I switch back to client.kdb/client.sth, it works fine:
$ ls -la $INFORMIXDIR/etc/client.???
-rw-r--r--. 1 informix informix 88 Feb 6 2020 /informix/products/TJGTEST/etc/client.crl
-rw-r--r--. 1 informix informix 10088 Feb 6 2020 /informix/products/TJGTEST/etc/client.kdb
-rw-r--r--. 1 informix informix 88 Feb 6 2020 /informix/products/TJGTEST/etc/client.rdb
-rw-r--r--. 1 informix informix 193 Feb 6 2020 /informix/products/TJGTEST/etc/client.sth
$ echo "SELECT COUNT(*) FROM systables;" | dbaccess tjgdb
Database selected.
(count(*))
77
1 row(s) retrieved.
Database closed.
This seems to
only be the case for implicit local connections involving PAM users. Connecting remotely, or using dbaccess->Connection->Connect and manually typing the username and password also works.
I suspect this is a bug with the OpenSSL implementation.
I have a case open,
TS004916260------------------------------
TOM GIRSCH
------------------------------
#Informix