Extremely High Load Avg

[Larry Becke]

Ever hear of a system with a load average of > 100 (try 152.xx at times)

[Larry Becke]

last pid: 5327; load averages: 130.56, 108.92, 100.12 syst 10:39:14
853 processes: 836 sleeping, 14 running, 3 zombie
CPU states: 42.7% usr, 0.0% nice, 57.3% sys, 0.0% idle, 0.0% block, 0.0% intr
Memory: Real: 623M/805M act/tot Virtual: 627M/856M act/tot Free: 1.4G

TTY PID USERNAME PRI NICE SIZE RES STATE TIME CPU COMMAND
pts/2 5001 informix 241 0 37K 66K run 0:02 8.52% tar
tcb 1421 mkluger 241 0 96K 243K run 1:06 7.15% unld_po_data.4
? 2843 fspinf 156 0 2663K 747K sleep 38:45 4.56% oninit
? 2845 fspinf 154 0 2655K 739K sleep 25:54 4.54% oninit
? 2743 ccspinf 241 0 923K 163K run 277:53 4.15% oninit
? 2796 omdpinf 156 0 1146K 1217K sleep 7:03 3.69% oninit
? 4666 root 241 0 144K 99K run 0:09 2.85% awk
? 5405 jwunsch 154 0 256K 180K sleep 0:09 2.84% sshd
? 2342 netscape 154 0 4892K 5157K sleep 18:08 2.71% ns-httpd

[Anupam Anshu_1]

Looks like you have given an output from top(1). Could you please run:

$ top -f file

And attach this file and post it here. Also tell us which version of HP-UX you are using.

Regards,

Anshu

[KapilRaj]

vmstat 2 ... would give little more and u can see whether cpu,memory or io is the problem

though it looks like cpu here.. Could u pl post the o/p ?

Kaps

[Chris Wilshaw]

The only time I've seen a similar issue was when one of the disks on a server was failing and recovering constantly. The processes all got stacked up waiting for clear I/O to the device.

Check syslog.log for any errors.

In my case, the server was mirrored, so when the faulty disk was pulled out, the server recovered and processed the outstanding requests on the other disk in the pair.

[Larry Becke]

vmstat 2 output

procs memory page faults cpu
r b w avm free re at pi po fr de sr in sy cs us sy id
10 0 0 154690 355411 56 24 1 0 35 0 1 526 6070 5156 51 14 35
10 0 0 151865 355299 42 36 0 0 16 0 0 181 5311 15594 78 22 0
10 0 0 151865 355210 84 31 0 0 28 0 0 205 5709 14934 61 39 0
179 0 0 155706 355420 195 60 0 0 63 0 0 240 8756 16547 39 61 0

[Larry Becke]

Here's the top -d 1 10000 > top.file file (the version of top that I am using does not have the -f parameter)

[KapilRaj]

Looks like u r running short of memory ... Add a bit more . That's the reason the number of blocked processes are more than the runq at times.

That means the processor is free but it can not do a usefull process 'cause of memory.

Regds,

Kaps

[Larry Becke]

Yet the system is only running 60% used of available memory.

[Larry Becke]

Which piece of info is telling you that there isn't enough memory?