Hello !
We have an AlphaServer 8400 with eight EV5 _at_ 625 MHz cpus, 4 GB memory,
DU 4.0B eco #8, configured as an Ingres 6.4 database server.
Two nights ago, we upgraded it to 4.0E, and replaced the cpus with
six EV6 _at_ 525 MHz. Since then, we've been experiencing severe performance
problems. We ran sys_check and it reported "There is excessive page-in
activity on the system ( 4800 )". We also see a high context switch rate,
and an average 20% system time (with 80% user time, little or no idle itme)
The system has normally about 170 processes, we reduced this to 150
by tweezing some Ingres parameters, but to no avail.
Here's a sample of vmstat 5 output
Virtual Memory Statistics: (pagesize = 8192)
procs memory pages intr cpu
r w u act free wire fault cow zero react pin pout in sy cs us sy
id
21174 36 478K 765 33K 9414 2439 2796 0 3109 0 689 36K 8K 74 24
2
19170 40 478K 702 33K 2970 698 1300 0 786 0 993 25K 8K 79 20
2
18177 34 478K 724 33K 346 19 269 0 15 0 749 14K 7K 80 18
2
16177 36 478K 746 33K 2216 316 1302 0 340 0 736 19K 7K 80 19
1
20172 37 478K 761 33K 362 80 176 0 65 0 867 15K 8K 78 20
2
17174 38 478K 697 33K 850 21 538 0 41 0 866 22K 8K 80 19
1
17174 38 478K 765 33K 847 22 501 0 41 0 773 18K 8K 79 19
2
18176 35 478K 789 33K 1859 281 1259 0 288 0 733 16K 7K 79 20
2
19174 36 478K 684 33K 320 42 194 0 47 0 781 19K 7K 82 17
1
16176 37 478K 731 33K 133 0 128 0 0 0 840 18K 7K 83 16
1
15178 36 478K 757 33K 141 0 139 0 0 0 832 17K 7K 84 15
1
17178 34 478K 755 33K 525 54 193 0 85 0 651 17K 7K 84 15
1
18182 36 478K 744 33K 7829 1745 2909 0 2246 0 659 25K 7K 82 18
0
20180 36 478K 760 33K 291 0 269 0 0 0 739 18K 7K 84 15
1
19177 35 476K 2124 33K 3878 1166 1012 0 1352 0 618 43K 6K 83 16
1
18177 34 478K 739 33K 2623 526 1448 0 583 0 689 20K 7K 84 15
1
19173 38 478K 821 33K 815 90 248 0 149 0 892 17K 8K 79 19
1
20175 34 478K 766 33K 1041 38 571 0 55 0 712 16K 7K 81 17
2
19175 36 477K 1504 33K 1603 34 917 0 64 0 867 19K 8K 78 21
1
17178 35 478K 758 33K 3423 719 1575 0 613 0 688 18K 7K 82 17
2
16179 35 478K 783 33K 677 150 283 0 130 0 585 21K 6K 84 15
1
And here's our sysconfigtab :
ipc:
shm-max = 1073741824
shm-mni = 131072
shm-seg = 1024
sem-mni = 4096
sem-msl = 500
sem-opm = 1250
sem-ume = 900
vm:
ubc-maxpercent = 80
vm-ubcseqstartpercent = 10
vm-mapentries = 16384
vm-vpagemax = 131072
vm-page-free-target=768
proc:
max-proc-per-user = 2048
maxusers = 2048
max-threads-per-user = 2048
vfs:
bufcache = 1
name-cache-hash-size=8192
generic:
msgbuf_size = 65536
message-buffer-size = 65536
inet:
tcbhashsize = 400
advfs:
AdvfsPreallocAccess=6112
AdvfsCacheMaxPercent = 3
AdvfsMaxDevQLen = 48
If anybody has a clue on what's happening, please let me know, I will
be glad to provide further information. Many thanks in advance.
Miguel Fliguer - MINIPHONE S.A.
Buenos Aires, Argentina
Received on Tue Jul 27 1999 - 15:04:54 NZST