STKM
The STKM report, as shown in Figure 1, lists the server token manager statistics. LOCALUSR is the local system (the server). ZEROLINK is a "special client" used to handle zero link count files and vnode inactivations. Table 1 describes the contents of the report.
Server Token Manager (STKM) Statistics
--------------------------------------
Maximum tokens: 30724 Allocated tokens: 30720
Tokens In Use: 27687 File structures: 27696
Token obtains: 3542592 Token returns: 3485439
Token revokes: 1309562 Async Grants: 0
Garbage Collects: 666 Thrash Resolutions: 0
Thrashing Files: 8
Usage Per System:
System Tokens Obtains Returns Revokes Async Grt
-------- ---------- ---------- ---------- ---------- ----------
NP1 3781 897812 894887 502842 0
NP2 15147 1233561 1188354 415917 0
NP3 3 912 909 0 0
NP4 8756 1410737 1402062 504757 0
ZEROLINK 0 0 0 0 0
LOCALUSR 0 0 0 0 0
Thrashing Objects:
Inode Uniquifier File system
---------- ---------- --------------------
19305 181700 PLEX.ZFS.SMALL2
1 1 ZFSAGGR.BIGZFS.DHH.FS4.EXTATTR
711 184733 PLEX.ZFS.SMALL2
1 1 ZFSAGGR.BIGZFS.DHH.FS14.EXTATTR
1 1 ZFSAGGR.BIGZFS.DHH.FS1.EXTATTR
13 1 ZFSAGGR.BIGZFS.DHH.FS4.EXTATTR
11 1 ZFSAGGR.BIGZFS.DHH.FS14.EXTATTR
21761 8528 ZFSAGGR.BIGZFS.DHH.FS6.EXTATTR
Field | Contents |
---|---|
Maximum tokens: | Lists the token limit at the server which is defined by the IOEFSPRM configuration option token_cache_size. The server runs garbage collection to ensure that token maximum is not exceeded. In some cases, the system workload might cause the token maximum to be exceeded, such as when there are many open files. |
Allocated tokens: | Number of tokens allocated in server memory. Tokens are allocated as needed, up to maximum tokens. |
File structures: | Number of file structures. |
Tokens In Use: | Number of tokens currently held by all clients and the local system. If this number approaches maximum tokens, then consider increasing the token_cache_size setting. |
Token obtains: | Total number of token obtains by all clients and local system. |
Token revokes: | Total number of token revokes by all clients and local system. |
Token returns: | Total number of token returns by all clients and local system. |
Async grants: | Number of asynchronously granted tokens to all clients and local system. Asynchronous grant is used during file deletion processing when the file is still opened by some process in the sysplex, and in support of NFS V4 share modes. |
Garbage collects: | Number of garbage collections of tokens. Garbage collection is used to keep the total number of client/local system tokens below the maximum whenever possible. If this number gets high, consider increasing the token_cache_size setting. |
Thrashing files: | Number of files or directories that are thrashing. |
Thrashing resolutions: | Number of trashing situations resolved. |
The report indicates how many tokens each system currently has, how many token obtains and token returns each system has done, and how many times each system has had some tokens revoked.
The report also contains a list of objects that are undergoing thrashing. Thrashing means that the system that owns the file system containing the object needed to keep revoking tokens for the object because multiple systems were repeatedly writing to it. The list contains the inode and uniquifier of the object and the file system that contains it.