scyld-sysinfo -- Capture the system state information.
[--up | -i NODES]
The tool works best when executed by a cluster administrator who is either
user root or a user with
The executing user must have write access to the current working directory.
The tool captures elements of the current system state into a subdirectory of
the current working directory with the name
(using a 2-digit Year-Month-Day).
This "capture" subdirectory is compressed by default into a gzip'ed tarball;
alternatively, the optional
--no-tar argument skips that compression and
allows the administrator to explore the "capture" subdirectory to view exactly
what information the tool has captured.
The administrator can employ a blacklist file containing a list of files and
directories to not capture,
passing this blacklist path to the tool with the
The administrator can also use
--no-tar and manually delete captured files
and subdirectories within
then manually compress the final captured information for archival or for
sending the file to others for examination.
The tool also optionally captures sysinfo state for compute nodes, for either all up nodes or a specific node or list of nodes.
If the optional
-d DIR_SUBSTR string is specified,
then the directory name contains that alphanumeric string,
-m MESSAGE is specified,
then the MESSAGE string is retained as the contents of the file
DESCRIPTION at the top of the output directory.
-m MESSAGE is not specified,
then the script queries the user for optional multi-line input that is
retained as file
DESCRIPTION in the output directory.
In the rare event that the tool aborts while capturing data,
note that a partial capture is still available as the subdirectory
sysinfo-$(hostname)-YY-MM-DD in the current working directory.
- -h, --help
Print usage message and exit. Ignore trailing args, parse and ignore preceding args.
scyld-sysinfoversion and Scyld package versions.
- --no-save BLACKLIST
Do not save files/directories listed in file BLACKLIST.
Leave the output as a subdirectory, not as a gzip'ed tarball.
Optionally capture the state of all up compute nodes.
- -i NODES
Optionally capture the state of a specific node or nodes.
- -d DIR_SUBSTR
Insert the alphanumeric string DIR_SUBSTR into the output directory/tarball name.
- -d MESSAGE
If specified, then the MESSAGE string is retained as the contents of file
DESCRIPTIONat the top of the output directory.
Capture the state of the current node into a gzip'ed tarball, executed as user root.
Capture the state of the current node into a human-readable subdirectory of the current working directory.
scyld-sysinfo -I -d UMich
The output directory name for the head node "headnode1" is "sysinfo-UMich-headnode1-YY-MM-DD".
scyld-sysinfo -m "dhcpd fails with network error"
The output directory contains the file DESCRIPTION that contains the specified string.
Capture the state of the current head node and all the up compute nodes.
scyld-sysinfo -i n0-10
Capture the state of the current head node and compute nodes n0 through n10.
scyld-sysinfo -i n0,n2,n100
Capture the state of the current head node and compute nodes n0, n2, and n100.
Upon successful completion, scyld-sysinfo returns 0.
On failure, an error message is printed to
scyld-sysinfo returns nonzero.