When a problem occurs with GDS that cannot be resolved by referring to this Troubleshooting appendix, collect the investigation material and contact field engineers.
Investigation material collected by the pclsnap command, sdxsnap.sh, or FJQSS.
Use FJQSS if you want to collect the investigation material with other middleware products that support FJQSS.
If investigation material can be collected by neither the pclsnap command nor FJQSS for any reason such as customer requirements, collect the only essential investigation material by sdxsnap.sh.
In cases other than the above, collect investigation material by the pclsnap command.
Crash dump
For troubles related to the Management View, collect investigation material of Web-Based Admin View. For the method of collecting investigation material, see "PRIMECLUSTER Web-Based Admin View Operation Guide."
Note
Investigation material must be collected promptly. With the passage of time, necessary information may be lost.
In a cluster system, collect investigation material on all the nodes.
If the node in which the failure occurred can be restarted, please collect the crash dump manually before you restart the node. The crash dump is useful if the failure originates from the OS.
When collecting investigation material, among the by-id files (symbolic link files created under the /dev/disk/by-id directory) managed by the udev function of Linux, the by-id files of disks which are not registered in classes may not exist temporarily because OS creates the by-id files again triggered by the GDS processes for obtaining the device information.
When access to the by-id files of disks which are not registered in classes fails, retry access.
Alternatively, when collecting investigation material, do not access to the by-id files of disks which are not registered in classes.
