From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.8) Gecko/20050511 Firefox/1.0.4 Description of problem: When starting CUPS service in a cluster configuration with clumanger the service starts two cups daemons but doesn't ends correctly, and the processes: sh /etc/init.d/cups start and clusvcmgrd stays in memory until I manually stop the cups service from /etc/init.d/cups stop. The service is defined in the cluster.xml file as: <service checkinterval="0" failoverdomain="None" id="1" name="Printer" userscript="/etc/init.d/cups"> <service_ipaddresses/> </service> </services> The script works well when called directly from a shell in the machine, Version-Release number of selected component (if applicable): clumanager-1.2.16-1 How reproducible: Always Steps to Reproduce: 1. Defining a service in the cluster with init script /etc/init.d/cups 2. Enable the service in one node with: clusvcadm -e <Service_Name> 3. Check the status of the service and the log to test that actually is not working as it should be if the CUPS service was started manually without cluster control: ps -ef | grep cups 4. Tail /var/log/cluster to see log messages from cluster software 5. ps -ef | grep clus to see the processes of the cluster running Actual Results: 3. ps -ef | grep cups root 2648 2640 0 13:15 ? 00:00:00 /bin/sh /etc/init.d/cups start root 2653 2648 0 13:15 ? 00:00:00 initlog -q -c cupsd root 2654 2653 0 13:15 ? 00:00:00 cupsd root 2655 2654 1 13:15 ? 00:00:00 cupsd root 2728 28267 0 13:15 pts/0 00:00:00 grep cups 4. tail /var/log/cluster Jun 20 12:50:42 pcXXX clusvcmgrd[1592]: <debug> [C] Pid 1592 handling start request for service Printer Jun 20 12:50:42 pcXXX clusvcmgrd[1592]: <debug> Starting service 1 - flags 0x00000001 Jun 20 12:50:42 pcXXX clusvcmgrd[1592]: <debug> Handling start request for service Printer Jun 20 12:50:42 pcXXX clusvcmgrd[3273]: <debug> [M] Pid 1592 -> start for service Printer Jun 20 12:50:42 pcXXX clusvcmgrd[1592]: <notice> Starting disabled service Printer Jun 20 12:50:42 pcXXX clusvcmgrd[1592]: <debug> Exec of script /usr/lib/clumanager/services/service, action start, service Printer Jun 20 12:50:42 pcXXX clusvcmgrd: [1593]: <notice> service notice: Starting service Printer ... Jun 20 12:50:42 pcXXX clusvcmgrd: [1593]: <notice> service notice: Running user script '/etc/init.d/cups start' Jun 20 12:51:39 pcXXX clusvcmgrd: [1593]: <notice> service notice: Started service Printer ... 5. ps -ef | grep clus root 3273 1 0 Jun16 ? 00:00:25 /usr/sbin/clusvcmgrd root 2576 785 0 13:15 pts/0 00:00:00 clusvcadm -e Printer root 2577 3273 0 13:15 ? 00:00:00 /usr/sbin/clusvcmgrd Expected Results: 3. ps -ef | grep cups root 2161 1 3 13:09 ? 00:00:00 cupsd 4. tail /var/log/cluster Jun 20 12:50:42 pcXXX clusvcmgrd[1592]: <debug> [C] Pid 1592 handling start request for service Printer Jun 20 12:50:42 pcXXX clusvcmgrd[1592]: <debug> Starting service 1 - flags 0x00000001 Jun 20 12:50:42 pcXXX clusvcmgrd[1592]: <debug> Handling start request for service Printer Jun 20 12:50:42 pcXXX clusvcmgrd[3273]: <debug> [M] Pid 1592 -> start for service Printer Jun 20 12:50:42 pcXXX clusvcmgrd[1592]: <notice> Starting disabled service Printer Jun 20 12:50:42 pcXXX clusvcmgrd[1592]: <debug> Exec of script /usr/lib/clumanager/services/service, action start, service Printer Jun 20 12:50:42 pcXXX clusvcmgrd: [1593]: <notice> service notice: Starting service Printer ... Jun 20 12:50:42 pcXXX clusvcmgrd: [1593]: <notice> service notice: Running user script '/etc/init.d/cups start' Jun 20 12:51:39 pcXXX clusvcmgrd: [1593]: <notice> service notice: Started service Printer ... Jun 20 12:51:39 pcXXX clusvcmgrd[1592]: <debug> Exec of script for service Printer returned 0 5. ps -ef | grep clus root 3273 1 0 Jun16 ? 00:00:25 /usr/sbin/clusvcmgrd Additional info: The version of CUPS used: cups-1.1.17-13.3.13 It seems as if the init script hangs executing /usr/sbin/cupsd when it is called by the service manager of the cluster, and the the clusvcmgrd daemon stays in memory waiting for the success exit status of the script. I have tested this with a start clause that only executes /usr/bin/cupsd and then exit 0. It works smoothly when called from a root shell but hangs when called from service manager of the cluster manager suite (clusvcadm -e Printer).
Please update to 1.2.26.1, which fixes an issue where signals were blocked in started services.
(In reply to comment #1) > Please update to 1.2.26.1, which fixes an issue where signals were blocked in > started services. Sorry I can't find such version of the package, is it published for IA32 i386? Where can I find it or its associated SRPM? I found 1.2.22-2 but it still has the same behaviour with cups script. Thanks for your help.
https://rhn.redhat.com/network/software/packages/details.pxt?pid=313963
Upppssss, I can't get this link with my login account, it gives me a permission error. I have no systems or channels configured in my RHN profile because I'm using it only for testing purposes. Is there any other way or link where I could get the package to test if this solves my problem? Thanks in advance for your help
Hmm, your sales engineer can give you official packages if you're an evaluation customer. In the mean time, there are a few options: (a) This patch should apply against your 1.2.22 tree: https://bugzilla.redhat.com/bugzilla/attachment.cgi?id=109260 (b) You can use the older 1.2.24-0.1 packages (which include this fix) from the original bugzilla concerning blocked signals in services: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=143867 (c) You can rebuild the source RPM from updates.redhat.com: http://updates.redhat.com/enterprise/3ES/en/RHCS/SRPMS/clumanager-1.2.26.1-1.src.rpm
I have updated to version 1.2.24 from the second link and it works. Thanks, I will look forwad latest version *** This bug has been marked as a duplicate of 143867 ***
well, I have the same error: <debug> Exec of script /usr/lib/clumanager/services/service, action stop, service oracle_cluster and I have clumanager 1.2.26.1-1 ? , my service is the oracle database script. N.B: I use Bonding methode for acrive passive network.
Fixing product name. Clumanager on RHEL3 was part of RHCS3, not RHEL3