Has anyone seen client sessions disconnect from a database b

We have a situation where the server/client count exceeds the number of database connections.

From promon -> R&D -> 1 -> 17:

11/30/10        Status: Servers By Broker
11:17:54

Sv                                                   Pend.   Cur.   Max.   Port
No    Pid  Type       Protocol               Logins  Users  Users  Users    Num

  0  16144 Login      TCP           42193      0      0     10  16000
2   7418 Auto       TCP             8105      0     10     10   1375
  3  11792 Auto       TCP            6270      0     10     10   1417
  4  10209 Auto       TCP            1686      0     10     10   1427
  5    628 Auto       TCP              4054      0     10     10   1435
  6   5220 Auto       TCP             3008      0     10     10   1437
  7  14091 Auto       TCP            3498      0     10     10   1438
  8   9343 Auto       TCP               908      0      9     10   1378
  9  13806 Auto       TCP             3682      0      9     10   1449
10  29497 Auto       TCP             6449      0      9     10   1450
11   1151 Auto       TCP              4466      0      9     10   1454

  1   6647 Login      TCP                         0      0      0      5  16150

for a total of 96.

promon -> 5 is:

Activity  - Sampled at 11/30/10 11:17 for 614:16:37.

Event                 Total   Per Sec   Event                 Total   Per Sec
       Commits     27883950      12.6            Undos        23425       0.0
Record Updates     14475348       6.5     Record Reads   3193942191    1444.3
Record Creates      3385250       1.5   Record Deletes      3853133       1.7
     DB Writes      2406018       1.1         DB Reads    112487662      50.9
     BI Writes      1537086       0.7         BI Reads      3096988       1.4
     AI Writes      1071652       0.5
  Record Locks    159345026      72.1     Record Waits          205       0.0
   Checkpoints          234       0.0    Buffs Flushed        11725       0.0

Rec Lock Waits    0 %    BI Buf Waits      0 %    AI Buf Waits      0 %
Writes by APW    99 %    Writes by BIW    46 %    Writes by AIW    89 %
Buffer Hits      99 %
DB Size          35 GB       BI Size    2930 MB       AI Size    5296 K
FR chain                  8486 blocks   RM chain               239742 blocks
Shared Memory 563195904         Segments      1

11 Servers, 68 Users (17 Local, 51 Remote, 12 Batch),2 Apws

and a list from proshut <db> -C list is:

a300sue1:/a4381/amfp438p > proshut amfp438p -C list | wc -l
75

...and out users who are connecting through the servers are getting "no more resources" errors.

v10.1B SP3/HPUX 11.11

Any ideas?