Tonneson DokuWiki

$ ipcs -l
         Shared Memory Limits 
   max number of segments = 4096               // SHMMNI        
   max seg size (kbytes) = 32768               // SHMMAX
   max total shared memory (kbytes) = 8388608  // SHMALL
   min seg size (bytes) = 1
          Semaphore Limits 
   max number of arrays = 1024                 // SEMMNI
   max semaphores per array = 250              // SEMMSL
   max semaphores system wide = 256000         // SEMMNS
   max ops per semop call = 32                 // SEMOPM
   semaphore max value = 32767
          Messages: Limits 
   max queues system wide = 1024               // MSGMNI
   max size of message (bytes) = 65536         // MSGMAX
   default max size of queue (bytes) = 65536   // MSGMNB

• Edit /etc/sysctl.conf
• Run sysctl -p

There are several systems that require additional steps following a kernel upgrade to restore services.

Note: The most important step when performing a kernel upgrade is to use rpm -ivh not rpm -Uvh!! Using the -i option will install a new kernel and leave the existing kernel in place, so you have a rollback option. Also, most of the steps below require the devel kernel, so make sure you have installed that version also.

VMWare - on VMWare-hosted virtual machines the following additional steps are required. If this is not done the system will not boot.

• On kernels 4.4 and greater the SCSI driver needs to be LSILogic.
• While server is off, Edit Settings and change SCSI driver to LSILogic
• After boot the network will not work until VMTools are re-run - /usr/bin/vmware-config-tools.pl
• Restart network - service network restart

SAN - on SAN-attached servers the HBA driver and {PowerPath} drivers need to be re-installed. If this is not done the system will not see the SAN disk. Perform these steps after the server is running on the new kernel.

• Re-install HBA driver
• Re-install {PowerPath}
• Start {PowerPath} - service {PowerPath} start
• Reboot

ServiceGuard - on servers with {ServiceGuard} (SVG) there are two kernel modules that need to be re-installed. If this is not done the server will not join the cluster. Perform these steps after the server is running on the new kernel.

• cd /usr/src/pident/drivers
• make modules
• make modules_install
• cd /usr/local/cmcluster/drivers
• make modules
• make modules_install

GFS - on systems with Red Hat Cluster Services (RHCS) and Global File System (GFS) there are several kernel modules and kernel-dependent packages to re-install. If this step is not done the server will not join the RHCS cluster and will be unable to mount GFS volumes. Perform these steps before booting to the new kernel.

• Comment out SAN filesystems in /etc/fstab to prevent them from mounting on reboot
• Prevent RHCS and GFS from starting on reboot: chkconfig ccsd off, chkconfig cman off, chkconfig fenced off, chkconfig clvmd off, chkconfig gfs off
• If SVG is also present, prevent SVG from starting on reboot by editing /usr/local/cmcluster/conf/cmcluster.rc, change AUTOSTART_CMCLD=0
• Reboot to new kernel
• Follow SAN procedure above to re-install SAN drivers
• If SVG is also present, follow SVG procedure above to re-install SVG drivers
• Install RHCS and GFS kernel rpms
• Reboot
• Manually start RHCS and GFS: service ccsd start, service cman start, service fenced start, service clvmd start, service gfs start
• If successful set service to auto-start: chkconfig ccsd on, chkconfig cman on, chkconfig fenced on, chkconfig clvmd on, chkconfig gfs on
• If present manually start SVG: cmrunnode
• If successful set cluster to auto-start: edit /usr/local/cmcluster/conf/cmcluster.rc, change AUTOSTART_CMCLD=1