RHEL6: pmiostat fails to replay a PCP archive log created on a RHEL6 system

Environment

• RHEL6 (any update)
• pcp-3.9.5-1.el6 or earlier
• pmiostat from pcp-3.10, e.g. as installed on your workstation.

Issue

• pmiostat fails with the following error when replaying archives collected on RHEL6 systems

pmiostat -a myarchive
pmiostat: pmLookupName: Unknown metric name ['disk.dev.write_rawactive']

Resolution

• This is resolved in PCP-3.10 and later, which includes the pmiostat tool
• Note that the pmiostat tool is NOT included in pcp-3.9.x. It first shipped in pcp-3.10 and later.
• Also note that pmiostat from pcp-3.10 or later can replay archives created on a system running pcp-3.9 or later (and also monitor such systems in live mode).
• on earlier versions of PCP such as 3.9.x on RHEL6, there is a typo in the pmlogconf configuration file for pmiostat, where disk.dev.read_rawactive appears twice but disk.dev.write_rawactive doesn't appear at all (also also for disk.all.write_rawactive). The second occurrence of disk.dev.read_rawactive needs to be changed to disk.dev.write_rawactive (also also for disk.all.write_rawactive).
• to make the change, edit /var/lib/pcp/config/pmlogconf/tools/iostat and make the above change at or near line 43
• or apply the following simple patch to that file :

# diff -u /var/lib/pcp/config/pmlogconf/tools/iostat.orig /var/lib/pcp/config/pmlogconf/tools/iostat
--- /var/lib/pcp/config/pmlogconf/tools/iostat.orig	2015-02-02 10:54:29.342987503 +1100
+++ /var/lib/pcp/config/pmlogconf/tools/iostat	2015-02-02 10:54:33.615000182 +1100
@@ -40,7 +40,7 @@
         disk.dev.write                  # total: Total writes completed successfully
         disk.dev.write_merge            # merged: grouped writes (resulting in one I/O)
         disk.dev.blkwrite               # sectors: Sectors written successfully
-        disk.dev.read_rawactive         # ms: milliseconds spent writing
+        disk.dev.write_rawactive         # ms: milliseconds spent writing

• after making the above change, the pmlogger service needs to be restarted - this will gather up all the pmlogconf configuration files, create a new /etc/pcp/pmlogger/config.default and then restart pmlogger

# service pmlogger restart

See Also

• Index of Performance Co-Pilot (PCP) articles, solutions, tutorials and white papers
• Content from pcp.io is not included.PCP Quick Reference Guide
• Introduction to storage performance analysis with PCP
• Content from www.pcp.io is not included.Performance Co-Pilot User's and Administrator's Guide

Root Cause

• typo in the iostat configuration file
• fixed in pcp-3.10 and later

This solution is part of Red Hat’s fast-track publication program, providing a huge library of solutions that Red Hat engineers have created while supporting our customers. To give you the knowledge you need the instant it becomes available, these articles may be presented in a raw and unedited form.