Contexte Métier
A CAS server is experiencing memory pressure. A CAS administrator needs to quickly investigate if any users are storing excessively large tables in their personal caslibs. The goal is to get a server-wide overview of personal data storage without checking users one by one.
Préparation des Données
Simulate a multi-user environment. 'analyst_a' has small work tables. 'data_scientist_b' has loaded a very large table, which is the likely cause of the memory issue. 'user_c' has an empty caslib.
Copié !
| 1 | /* Data setup requires admin privileges */ |
| 2 | PROC CAS; |
| 3 | /* Analyst A's small tables */ |
| 4 | DATA casuser(analyst_a).lookup_codes(promote=yes); |
| 5 | DO i=1 to 100; code=i; OUTPUT; END; |
| 6 | RUN; |
| 7 | |
| 8 | /* Data Scientist B's large table */ |
| 9 | DATA casuser(data_scientist_b).large_dataset(promote=yes); |
| 10 | DO i=1 to 5000000; x=rand('NORMAL'); y=rand('UNIFORM'); OUTPUT; END; |
| 11 | RUN; |
| 12 | |
| 13 | /* User C has no tables, just an active session */ |
| 14 | QUIT; |
Étapes de réalisation
1
Enable Server-Wide Access: Run the accessPersonalCaslibs action to see into all user caslibs.
Copié !
| 1 | PROC CAS; |
| 2 | ACCESSCONTROL.accessPersonalCaslibs; |
| 3 | RUN; |
2
Global Table Listing: Run the table.tableInfo action with a caslib pattern or for all known users to get a list of all tables in all personal caslibs. Here we check the three simulated users explicitly.
Copié !
| 1 | PROC CAS; |
| 2 | /* In a real script, you might loop through a list of users */ |
| 3 | TABLE.tableInfo / caslib="CASUSER(analyst_a)"; |
| 4 | TABLE.tableInfo / caslib="CASUSER(data_scientist_b)"; |
| 5 | TABLE.tableInfo / caslib="CASUSER(user_c)"; |
| 6 | RUN; |
Résultat Attendu
The administrator successfully executes table.tableInfo against multiple user caslibs in a single session. The results will show the small table for 'analyst_a', the very large table for 'data_scientist_b' (identifiable by its row count and size), and an empty result set for 'user_c'. This confirms the action is effective for server-wide diagnostics.