================================================================================ Revert a VXVM disk back to LVM control: chpv -C powerpath6 ================================================================================ comparable mount options for VXFS -vs- rbrw mount option for AIX's JFS. The following mount options will disable vxfs caching. -mincache=direct & convosync=direct. If there is entry in /etc/filesystems mount -V vxfs -o mincache=direct,convosync=direct /mount _point Or use the full path if the disk group is under VCS controlled. mount -V vxfs -o remount, mincache=direct,convosync=direct \ /dev/vx/dsk/dg_name/vol_name /moun _point ================================================================================ #setup virtual of 10.10.10.10 on en1 - use 'alias' keyword #put entry in /etc/hosts for "10.10.10.10 somevirtualname" ifconfig en1 inet 10.10.10.10 netmask 255.255.255.0 alias #remove ifconfig en1 inet 10.10.10.10 netmask 255.255.255.0 -alias #get rid of the virtual ifconfig en1 10.10.10.10 delete ================================================================================ #ERROR CmdClient - Command Server may not be running in this system export PATH=$PATH:/opt/VRTSvcs/bin: CmdServer ps -ef | grep -i Cmd ================================================================================ see LLT kernel values scls -l llt name type major idnum idname minpsz maxpsz lowat hiwat llt mux 39 2802 llt 1 16384 1024 10240 ================================================================================ set disks on both sides to show PVIDs (this is so you can import a vg) chdev -l hdiskpower0 -a pv=yes for DISK in $( lsdev -Cc disk -Fname | grep power) do echo $DISK chdev -l $DISK -a pv=yes done ================================================================================ root@server1# more /etc/llttab set-node server1 set-cluster 20 link en0 /dev/dlpi/en:0 - ether - - # en1 is for jeopardy issue - we only have one HB due to DR link en1 /dev/dlpi/en:1 - ether - - link-lowpri en2 /dev/dlpi/en:2 - ether - - root@server1# /opt/*llt/getmac /dev/dlpi/en:1 /dev/dlpi/en:1 00:09:6B:AE:F1:8D root@server1# /opt/VRTSllt/dlpiping -vs /dev/dlpi/en:1 & root@server2# /opt/VRTSllt/dlpiping -vc /dev/dlpi/en:1 00:09:6B:AE:F1:8D dlpiping: dlrecv do_get failed dlpiping: recv ECHO_RES failed ================================================================================ 0516-1281 synclvodm: WARNING, lv control block of fn_trans_rl1 has been over written. 0516-622 synclvodm: Warning, cannot write lv control block data. If you are using raw - this is OK as the LVs have their LVCB (logical volume control block) in the first 512 bytes of the LV. If not raw, then run "synclvodm". If raw, *** DO NOT *** run "synclvodm". ================================================================================ Document ID: 251109 http://support.veritas.com/docs/251109 E-Mail this document to a colleague How can LLT and GAB manually be unloaded and loaded on IBM AIX ? -------------------------------------------------------------------------------- Details: Steps to stop and unload Global Atomic Broadcast (GAB) 1. Check if GAB is loaded: # /usr/sbin/lsdev -CHc gab name status location description gab Available N/A 2. Check if GAB is running: # /sbin/gabconfig -l GAB Driver Configuration Driver state : ConfiguredPartition arbitration: Disabled Control port seed : Enabled Halt on process death: Disabled Missed heartbeat halt: Disabled Halt on rejoin : Disabled Keep on killing : Disabled Restart : Disabled Node count : 0 Disk HB interval (ms): 1000 Disk HB miss count : 4 IOFENCE timeout (ms) : 15000 Stable timeout (ms) : 5000 3. Stop GAB and verify: # /sbin/gabconfig -U # /sbin/gabconfig -l GAB Driver Configuration Driver state : UnconfiguredPartition arbitration: Disabled Control port seed : Disabled Halt on process death: Disabled Missed heartbeat halt: Disabled Halt on rejoin : Disabled Keep on killing : Disabled Restart : Disabled Node count : 0 Disk HB interval (ms): 1000 Disk HB miss count : 4 IOFENCE timeout (ms) : 15000 Stable timeout (ms) : 5000 4. Unload GAB and verify: # /usr/sbin/rmdev -l gab gab Defined # /usr/sbin/lsdev -CHc gab name status location description gab Defined N/A Steps to stop and unload Low Latency Transport (LLT) 1. Check if LLT is loaded: # /usr/sbin/strload -q -d llt llt: yes 2. Check if LLT is running: # /sbin/lltconfig LLT is running 3. Stop LLT and verify: # /sbin/lltconfig -U lltconfig: this will attempt to stop and reset LLT. Confirm (y/n)? y # /sbin/lltconfig LLT is not running 4. Unload LLT and verify: # /usr/sbin/strload -u -d llt # /usr/sbin/strload -q -d llt llt: no Steps to load and start LLT (provided LLT is not loaded) 1. Load LLT and verify: # /usr/sbin/strload -d llt # /usr/sbin/strload -q -d llt llt: yes 2. Start LLT and verify: # /sbin/lltconfig -c # /sbin/lltconfig LLT is running Steps to load and start GAB (provided GAB is not loaded) 1.Load GAB and verify: # /usr/sbin/mkdev -l gab gab Available # /usr/sbin/lsdev -CHc gab name status location description gab Available N/A 2. Start GAB and verify: # /sbin/gabconfig -c # /sbin/gabconfig -l GAB Driver Configuration Driver state : ConfiguredPartition arbitration: Disabled Control port seed : Enabled Halt on process death: Disabled Missed heartbeat halt: Disabled Halt on rejoin : Disabled Keep on killing : Disabled Restart : Enabled Node count : 0 Disk HB interval (ms): 1000 Disk HB miss count : 4 IOFENCE timeout (ms) : 15000 Stable timeout (ms) : 5000 ================================================================================ How to verify if DLPI, LLT and GAB are loaded on IBM AIX -------------------------------------------------------------------------------- Details: Data Link Provider Interface (DLPI) and Low Latency Transport (LLT) The DLPI and LLT drivers are configured in /etc/pse.conf. The Portable Stream Environment is loaded from this file at boot time. These are the lines in /etc/pse.conf that enable DLPI and LLT: # # *** tab delimited fields **** # d+ dlpi en /dev/dlpi/en # streams dlpi ethernet driver d llt # LLT driver Please refer to VERITAS TechNote 251126 (link in the Related Documents section of this TechNote) to learn about the /etc/pse.conf file. To check if both drivers are loaded use the strload command: # /usr/sbin/strload -q -d dlpi dlpi: yes # /usr/sbin/strload -q -d llt llt: yes Global Atomic Broadcast (GAB) GAB is loaded as a driver from /usr/lib/drivers at boot time and is registered as a custom device in the device configuration database. GAB is associated with the device class "gab". To test if GAB is running execute the following command: # /usr/sbin/lsdev -CHc gab name status location description gab Available N/A The status "Available" indicates that GAB is running. Otherwise the status would be "Defined". ================================================================================ After installing VERITAS Cluster Server (tm), when LLT is started with "lltconfig -c", it reports the error "LLT:lltconfig:15000: open /dev/dlpi/en failed: A file or directory in the path name does not exist." -------------------------------------------------------------------------------- Exact Error Message LLT:lltconfig:15000: open /dev/dlpi/en failed: A file or directory in the path name does not exist. Details: This would tend to indicate that the Installation Guide has not been followed correctly. Here are the steps to correct the issue. 1. Look at the file /etc/pse.conf and check that the line for /dev/dlpi/en has been uncommented, e.g.: # grep dlpi /etc/pse.conf #d+ dlpi en /dev/dlpi/en # streams dlpi ethernet driver #d+ dlpi et /dev/dlpi/et # streams dlpi 802.3 driver #d+ dlpi tr /dev/dlpi/tr # streams dlpi token ring driver #d+ dlpi fi /dev/dlpi/fi # streams dlpi FDDI driver Also check to ensure that the line "d llt #LLT driver" has been inserted into the "#PSE Modules" section. 2. Ensure that dlpi is not currently loaded: # strload -q -d dlpi dlpi: yes If it is currently loaded then you need to unload it: # strload -u -d dlpi # strload -q -d dlpi dlpi: no 3. Ensure that the file /dev/dlpi does not exist (it will be removed when the streams driver dlpi is unloaded): # ls -ld /dev/dlpi ls: 0653-341 The file /dev/dlpi does not exist. 4. Reboot the server for the changes to take effect Here is the relevant information from the VERITAS Cluster Server 3.5 for AIX Installation Guide, page 10: Modifying /etc/pse.conf to Enable the LLT Driver Before installing VCS, you must make two modifications to the /etc/pse.conf file to enable the LLT driver. On each system, use the following procedure. 1. Check to see if the Ethernet driver is configured in the /etc/pse.conf file: # egrep 'ethernet driver' /etc/pse.conf 2. In the output, examine the line containing the "ethernet driver" expression: #d+ dlpi en /dev/dlpi/en # streams dlpi ethernet driver 3. If the comment symbol ("#") precedes line, the Ethernet driver is not configured. Using vi or another text editor, edit the file: # vi /etc/pse.conf 4. Find the section in the file labeled "#PSE drivers" and look for the line shown in step 2. Uncomment the line by removing the initial "#" symbol. 5. Move to the section of the file labeled "# PSE Modules." 6. Within that section, add the line: d llt # LLT driver This entry enables the LLT driver after each system reboot. 7. Save and close the file. 8. If you uncommented the /dev/dlpi/en entry, reboot the system to configure the driver. The system does not need to be rebooted if the only change was the addition of the llt entry (step 6) in the PSE Modules section. 9. Repeat step 1 through step 8 on each sysdem in the cluster. If you uninstall VCS and LLT, disable the LLT driver by removing the line you have added to /etc/pse.conf in step 6. ================================================================================ What is the /etc/pse.conf file on IBM AIX and what do the entries for the Data Link Provider Interface (DLPI) and Low Latency Transport (LLT) mean ? -------------------------------------------------------------------------------- Details: The /etc/pse.conf file is used to configure the Portable Streaming Environment on IBM AIX. The entries in the /etc/pse.conf file represent STREAMS drivers and modules that are loaded at boot time. On IBM AIX, DLPI and LLT are loaded as STREAMS drivers. That means both drivers have no entries in either the Predefined Device Database nor in the Custom Device Database. STREAMS drivers (and modules) work slightly different from character I/O device drivers in terms of communication methods. STREAMS drivers use only open/close system calls, and process any other communication through message queues (send/receive to/from processes and hardware). Once configured in /etc/pse.conf, after a reboot, LLT and DLPI device files will be found in the /dev directory: # ls -la /dev | grep llt crw-rw-rw- 1 root system 16, 51 04 Oct 12:29 llt # ls -la /dev/dlpi total 16 drwxr-xr-x 2 root system 256 21 May 15:08 . drwxrwxr-x 7 root system 8192 04 Oct 12:35 .. crw-rw-rw- 1 root system 16, 50 21 May 15:08 en The following lines in /etc/pse.conf are necessary to get DLPI and LLT loaded: d+ dlpi en /dev/dlpi/en # streams dlpi ethernet driver d llt # LLT driver What do the entries in each field mean ? d Specifies that the entry refers to a driver (rather than a module) d+ The "+" specifies that the driver can be configured more than once, e.g. there can be a DLPI driver loaded for ethernet and one for token ring. dlpi, llt Driver names used to build the device files, resp. directories in /dev (if not specified otherwise with "alternate node") en Optional argument passed to extension's config routine. /dev/dlpi/en Alternate node to create - if not specified, the device file will be created as /dev/. Note: It may be possible to avoid a reboot after modifying /etc/pse.conf by using the following commands: # /usr/sbin/strload -f /etc/pse.conf ( to load all drivers/modules ) or # /usr/sbin/strload -d (to load a specific driver/module) Note: Care must be taken with these commands and they should only be used if you are familiar with them. ================================================================================ Where does the Group Atomic Broadcast kernel extension write its messages on IBM AIX ? -------------------------------------------------------------------------------- Details: On IBM AIX, the Group Atomic Broadcast (GAB) informational and error messages are written to the /var/adm/streams/error.mm-dd file, where mm-dd stands for the day and month the file was created. Example: 000070 14:06:53 110e802a 0 ... 2803 0 GAB:20035: Port h attempting to kill process due to client process failure 000071 14:06:53 110e803c 0 ... 2803 0 GAB:20032: Port h closed 000072 14:07:11 110e8741 0 ... 2803 0 GAB:20036: Port h gen da90d904 membership 01 000073 14:17:38 110f7c3a 0 ... 2803 0 GAB:20032: Port h closed What does each column of these entries mean ? 000073 14:17:38 110f7c3a 0 ... 2803 0 GAB:20032: Port h closed 000073 - error message sequence number 14:17:38 - message entry time 110f7c3a - message entry time in ticks since last reboot 0 - flag indicating message severity (F denotes fatal) and message processing (T denotes message send to a tracing process, and N denotes message send to an administrator account) 2803 - STREAMS module ID 0 - Sub module ID GAB:20032: Port h closed - message text Background information: STREAMS messages are written to /var/adm/streams/error.mm-dd by the strerr daemon. Messages are generated by the strlog system call. See /usr/include/sys/strlog.h for definitions, and the strerr man page for more information. GAB is actually not a STREAMS driver or module. It runs as an AIX kernel extension but makes use of the STREAMS error logging facility. ================================================================================ Where does the Low Latency Transport module write its messages on IBM AIX ? -------------------------------------------------------------------------------- Details: On IBM AIX, the Low Latency Transport (LLT) informational and error messages are written to the /var/adm/streams/error.mm-dd file whereby mm-dd stands for the day and month the file was created. Example: 000050 14:41:58 00011e50 0 ... 2802 0 LLT:10032: link 1 (en:3) node 1 inactive 14 sec (902/1803) 000051 14:41:59 00011ea0 0 ... 2802 0 LLT:10032: link 0 (en:0) node 1 inactive 15 sec (901/1803) 000052 14:41:59 00011eb4 0 ... 2802 0 LLT:10032: link 1 (en:3) node 1 inactive 15 sec (902/1803) 000053 14:42:00 00011f04 0 ... 2802 0 LLT:10032: link 0 (en:0) node 1 inactive 16 sec (901/1803) 000054 14:42:00 00011f04 0 ... 2802 0 LLT:10033: link 0 (en:0) node 1 expired What does each column of these entries mean ? 000054 14:42:00 00011f04 0 ... 2802 0 LLT:10033: link 0 (en:0) node 1 expired 000054 - error message sequence number 14:42:00 - message entry time 00011f04 - message entry time in ticks since last reboot 0 - flag indicating message severity (F denotes fatal) and message processing (T denotes message send to a tracing process, and N denotes message send to administrator account) 2802 - streams module ID 0 - Sub module ID LLT:10033: link 0 (en:0) node 1 expired - message text Background information: Streams messages are written to /var/adm/streams/error.mm-dd by the strerr daemon. Messages are generated by the strlog system call. See /usr/include/sys/strlog.h for definitions, and the strerr man page for more information. To STOP streams logs, remove /var/adm/streams directory. ================================================================================