Corevector
From Jaymzworld
Contents |
NAME
corevector - set/clear/list core file vector settings
SYNOPSIS
corevector
corevector -?
corevector -l
corevector -c�all | warm | cold2 | reload | insane | single | trap
corevector -s [-f] all | warm | cold2 | reload | insane | single | trap
DESCRIPTION
This command is used to set, clear, or list the server's core file vector settings. When the core files are set to be taken they are generated prior to the execution of the CommunicaMgr (MV) restart (warm, cold2 and reload) that is requested. For the other settings (insane, single and trap) the core file(s) are generated after these actions are detected.
When the core files are to be taken, the server's state of health (SOH) is lowered by the prc_mgr to request a server interchange (assuming the server is ACTIVE). It then waits N time period to allow the interchange to occur (all OP processes are stopped), it checks to ensure it is now on the standby server and then generates the core files. The core files are only taken on the STANDBY server unless the -f (force) argument is specified. The same code sequence is executed but the core files are generated even if a server interchange does not occur. In a simplex server configuration the core files are taken (-f argument is not necessary).
On a server interchange an MV restart is executed on the new standby (STB) server and new active (ACT) server. These initial interchange requested restarts do not generated core files.
- no parameters - List the core vector settings.
-
- -?
- Detailed help information is displayed.
- -l
- List the core vector settings.
- -f
- The force (-f) option is only valid with set (-s) option. Force the core file(s) to be taken on the ACTIVE server if a server interchange was not successful. Note: core files are usually taken on the STANDBY server so as not to impact customer service.
- -c
- Clear (-c) corevector settings using the valid arguments noted below.
- -s
- Set (-s) corevector settings using the valid arguments noted below.
Corevector command valid arguments: all coredump files specified by all command arguments warm coredump on warm restart request cold2 coredump on cold2 restart request reload coredump on reload (reboot) restart request insane coredump on system insane condition, e.g., ALLOC_BUF single coredump on single process restart and killing of TERMINAL processes trap coredump of process that trapped, e.g., SIGSEGV signal 11
- Example of the command output
init@jupiter2> corevector -s warm single
init@jupiter2> corevector -sf cold2 reboot
init@jupiter2> corevector -c reboot
init@jupiter2> corevector -sf trap
init@jupiter2> corevector -l
Current corevector settings:
warm
cold2 - force
single
trap - force
WARNINGS
This command should be used with extreme caution. Core dumps could take anywhere from 1 minute to 4.5 minutes.
On the Linux platform after the process core file(s) is generated the process dies. There is currently no procedure to have the Linux kernel take a process core file and keep the binary in memory allowing any control of restarting the binary in main().
Each server has their own core vector settings, i.e., the corevector settings are not shadowed or filesync'd from one server to the next.
DIAGNOSTICS
- The following messages may be displayed
-
Error in reading ecs.conf file - EXITING no changes made - File may be missing or file permission error
-
set_regval err=-1, errno�=13 check ecs.conf file - File may be missing or file permission error
-
register with border process err=3 - Border process error, core vector has been set.
-
Error in notifying CommunicaMgr software but the core vector has been set - CommunicaMgr software may be down.
