From: Rothwell, Giles (ELS) (G.Rothwell@elsevier.co.uk)
Date: Thu May 08 2003 - 07:04:41 EDT
I am hoping someone can point me in the right direction. We are currently
putting together an application which effectively enriches a set of XML data
from a reference database held in Oracle The application is running in Java
and querying the database via JDBC. The problem we are having is with a new
8-way V1280 which is hosting the Oracle Database. The initial test runs are
giving a slower performance than a dual cpu Intel machine on which the app
was originally developed. During the tests the DB server seems to be
suffering a very high level of "mutex contention". I have included below
part of the part of the lockstat output, but to be honest I am not entirely
sure what I am looking at. There seems to be very little information
available on the subject. From what I have found it sounds like some work is
required on the application, but I don't want to go all guns blazing at the
developers without soemthing more concrete or helpful.
Does anyone know of any good sources of info on this, or seem anything
similar.
Many Thanks in advance..
Giles Rothwell
Team Leader
Unix Systems Management
Elsevier Science
Spin lock spin: 48189 events in 5.001 seconds (9636 events/sec)
Count indv cuml rcnt spin Lock Caller
----------------------------------------------------------------------------
---
3830 8% 8% 1.00 590 cpu[2]+0x78 disp_getbest+0xc
3593 7% 15% 1.00 581 cpu[3]+0x78 disp_getbest+0xc
3478 7% 23% 1.00 107 cpu[1]+0x78 disp+0x94
3405 7% 30% 1.00 120 cpu[8]+0x78 disp+0x94
3277 7% 36% 1.00 728 cpu[8]+0x78 disp_getbest+0xc
3185 7% 43% 1.00 519 cpu[1]+0x78 disp_getbest+0xc
2998 6% 49% 1.00 69 cpu[2]+0x78 disp+0x94
2995 6% 56% 1.00 650 cpu[11]+0x78 disp_getbest+0xc
2893 6% 62% 1.00 629 cpu[10]+0x78 disp_getbest+0xc
2736 6% 67% 1.00 573 cpu[0]+0x78 disp_getbest+0xc
2672 6% 73% 1.00 125 cpu[3]+0x78 disp+0x94
2560 5% 78% 1.00 686 cpu[9]+0x78 disp_getbest+0xc
2144 4% 83% 1.00 113 cpu[0]+0x78 disp+0x94
2008 4% 87% 1.00 116 cpu[9]+0x78 disp+0x94
1979 4% 91% 1.00 74 cpu[10]+0x78 disp+0x94
1925 4% 95% 1.00 87 cpu[11]+0x78 disp+0x94
546 1% 96% 1.00 1056 sleepq_head+0x698 cv_signal+0x38
160 0% 96% 1.00 175 cpu[2]+0x78 setfrontdq+0xf8
158 0% 97% 1.00 208 cpu[8]+0x78 setfrontdq+0xf8
142 0% 97% 1.00 147 cpu[1]+0x78 setfrontdq+0xf8
135 0% 97% 1.00 851 cp_default+0x30 disp_getbest+0xc
123 0% 97% 1.00 170 cpu[3]+0x78 setfrontdq+0xf8
119 0% 98% 1.00 239 cpu[9]+0x78 setfrontdq+0xf8
118 0% 98% 1.00 174 cpu[10]+0x78 setfrontdq+0xf8
117 0% 98% 1.00 159 cpu[11]+0x78 setfrontdq+0xf8
103 0% 98% 1.00 998 turnstile_table+0x838 turnstile_lookup+0x50
78 0% 99% 1.00 190 cpu[8]+0x78 setbackdq+0x174
76 0% 99% 1.00 150 cpu[3]+0x78 setbackdq+0x174
75 0% 99% 1.00 157 cpu[0]+0x78 setfrontdq+0xf8
73 0% 99% 1.00 141 cpu[1]+0x78 setbackdq+0x174
65 0% 99% 1.00 144 cpu[2]+0x78 setbackdq+0x174
59 0% 99% 1.00 190 cpu[11]+0x78 setbackdq+0x174
51 0% 99% 1.00 253 sleepq_head+0x698 cv_block+0xa0
51 0% 99% 1.00 136 cpu[9]+0x78 setbackdq+0x174
43 0% 100% 1.00 166 cpu[10]+0x78 setbackdq+0x174
40 0% 100% 1.00 134 cpu[0]+0x78 setbackdq+0x174
36 0% 100% 1.00 705 turnstile_table+0x838
turnstile_interlock+0x7c
24 0% 100% 1.00 1197 turnstile_table+0xff8 turnstile_lookup+0x50
19 0% 100% 1.00 326 turnstile_table+0x9f8 turnstile_lookup+0x50
17 0% 100% 1.00 217 sleepq_head+0x728 cv_block+0xa0
17 0% 100% 1.00 797 turnstile_table+0xd78 turnstile_lookup+0x50
8 0% 100% 1.00 3092 turnstile_table+0xcc8 turnstile_lookup+0x50
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 23:26:21 EDT