cpu usage per user view fails after update to version 5

[sansay]

One of the views in my Splunk health dashboard which showed the users' consumption of cpu due to searches no longer work now that we have updated to version 5.0.4.

Here is the query: index=os sourcetype=ps host=lv-snp-prdsplunk* OR host=lvp-snp-prsplnki* splunkd search pctCPU | multikv fields pctCPU USER | rex field=raw "--user=(?[^-]+)(?=)" | search USER=splunk SPL_USER=* | timechart sum(pctCPU) by SPL_USER useother=false limit=100

After reducing my query to: index=os sourcetype=ps host=lv-snp-prdsplunk* OR host=lvp-snp-prsplnki* splunkd pctCPU | multikv fields pctCPU USER | rex field=raw "--user=(?[^-]+)(?=)"

I realized that the both string "search" and value "--user=..." have disappeared.
Anybody knows another way to do this?

[sansay · Accepted solution]

Well I finally found the time to investigate this issue. Here is the solution:
index=os sourcetype=ps host=lvp-snp-prsplnki* splunkd search NOT splunk-system-user | multikv fields pctCPU USER | rex field=raw "--user=(?.+)--pro" | where IsNotNull(USER) | timechart span=1m useother=false limit=0 sum(pctCPU) by SPL_USER

It seems that many things in the logs changed since this used to work.