|2009-11-23 13:58, comment #10:|
Also being considered as mirror bug to GSVG issue bug #59215.
Low risk with TD 23rd May 2010.
|2009-11-19 13:14, comment #9:|
This has been passed to Alain Roy of VDT in order to get a
rebuild of the offending globus part(s).
|2009-11-18 12:24, comment #8:|
Adding Linda in CC - this clearly has some security implications.
|2009-11-08 07:43, comment #7:|
Please track this bug. I've upped the priority.
It's casuing problems for the CERN syslog tracker which is a core part of the CERN Security Team's infrastructue.
Luis Fernando Muñoz Mejías is the contact here.
|2009-08-05 13:51, comment #6:|
BTW, this is not just an "enhancement", but a real bug: it prevents tracing events in any central location.
|2009-08-05 13:30, comment #5:|
Essentially, the problem here is that the GRAM Gatekeeper insists in telling syslog that its application name (or syslogtag, naming depends on the syslog daemon) is "GRAM gatekeeper", with whitespaces.
This is not allowed by syslog practices and standard draft (RFC3164 and 5424 respectively), and thus syslog daemons are allowed to mess things up. The GRAM gatekeeper code must change the ident argument to something without spaces when calling to openlog.
According to the syslog standard draft, this field has the following form:
APP-NAME = NILVALUE / 1*48PRINTUSASCII
PRINTUSASCII = %d33-126
Whitespace is %d32.
|2009-04-20 07:36, comment #4:|
yes, we are running with mixed rpms:
[root@gridce01 ~]# rpm -qa|grep kernel
[root@gridce01 ~]# rpm -qa | grep vdt_globus
P.S. The bug is not critical, I realized cause we send syslog messages to a relational database and these messages do not make it cause the server is not properly filled in.
|2009-04-20 06:37, comment #3:|
Hi Laura, I see you commented that $ARCH is x86_64 - The CE isn't released for x86_64. Are you running with mixed RPMs (some i386, some x96_64)? Can you also send me the version of gatekeeper installed (rpm -qa | grep vdt_globus)
[nb - as we basically rebundle the VDT releases for gLite, this probably will have to be pushed upstream]
|2009-04-17 09:46, comment #2:|
Sorry, architecture is x86_64.
|2009-04-17 09:38, comment #1:|
I think this one is for you, could you please have a look? Thanks a lot!
|2009-04-17 09:23, original submission:|
I can see on /var/log/messages the following entries related to the gatekeeper.
Feb 6 14:23:26 GRAM gatekeeper: /opt/globus/sbin/globus-gatekeeper pid=14288 starting at Fri Feb 6 14:23:26 2009
Feb 6 14:23:26 GRAM gatekeeper: GRAM contact: ce1jtest.ft.uam.es:2119:/DC=es/DC=irisgrid/O=uam/CN=ce1jtest.ft.uam.es
According to syslog specs, second field shouldn't contain the hostname?