pcp
[Top] [All Lists]

[Bug 1336808] New: RFE: Provide uncached per-process memory metrics

To: pcp@xxxxxxxxxxx
Subject: [Bug 1336808] New: RFE: Provide uncached per-process memory metrics
From: bugzilla@xxxxxxxxxx
Date: Tue, 17 May 2016 13:31:18 +0000
Auto-submitted: auto-generated
Delivered-to: pcp@xxxxxxxxxxx
https://bugzilla.redhat.com/show_bug.cgi?id=1336808

            Bug ID: 1336808
           Summary: RFE: Provide uncached per-process memory metrics
           Product: Fedora
           Version: 23
         Component: pcp
          Assignee: nathans@xxxxxxxxxx
          Reporter: myllynen@xxxxxxxxxx
        QA Contact: extras-qa@xxxxxxxxxxxxxxxxx
                CC: brolley@xxxxxxxxxx, fche@xxxxxxxxxx, lberk@xxxxxxxxxx,
                    mgoodwin@xxxxxxxxxx, nathans@xxxxxxxxxx,
                    pcp@xxxxxxxxxxx, pholasek@xxxxxxxxxx, scox@xxxxxxxxxx



Description of problem:
Consider the following case:

# swapoff -a
# free -m
              total        used        free     shared  buff/cache   available
Mem:           7686        1522        5244         312         918        5619
Swap:             0           0           0
# awk '/VmSwap/ && $2>0 {print}' /proc/*/status | sort -nr | head -n 5
VmSwap:      252 kB
VmSwap:      128 kB
VmSwap:      120 kB
VmSwap:      100 kB
VmSwap:       92 kB
# 

VmSwap being > 0 is caused by mm counter caching and the values might be
slightly off. A precise snapshot of a process memory use can be found in
/proc/<pid>/smaps or /proc/<pid>/pagemap. pgmap(1), included in libpagemap,
gets its data (USS/PSS/SWAP/RES/SHR) from /proc/<pid>/pagemap and should be
accurate.

It would be nice to have such uncached metrics available from PCP as well.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug 
https://bugzilla.redhat.com/token.cgi?t=PkzTt0saDg&a=cc_unsubscribe
<Prev in Thread] Current Thread [Next in Thread>