How to measure on Linux the peak memory of an application after has ended

Question

How do I measure the peak memory of an application running on Linux?

I run this app in batch so I cannot use RSS since, it reports the current memory. I need the peak memory that the application has used to be reported at the end.

The VmPeak is not a solution either, since it reports the allocated memory and also does not calculates from the real Ram but also from the Hard Disk.

Answer 1

Here are 2 methods for tracking a processes' peak memory usage.

Syrupy

I haven't used this tool but it sounds like what you're looking for. It's called Syrupy.

Description

Syrupy is a Python script that regularly takes snapshots of the memory and CPU load of one or more running processes, so as to dynamically build up a profile of their usage of system resources.

Example

$ syrupy.py myprog

  PID DATE        TIME     ELAPSED  CPU   MEM    RSS   VSIZE
14634 2008-10-10  20:45:25   00:00  0.0   0.0   2996    6680
14634 2008-10-10  20:45:26   00:01  105   0.2   7804   12592
14634 2008-10-10  20:45:27   00:02  103   0.2   8996   13776
14634 2008-10-10  20:45:28   00:03  103   0.2  10468   15348
14634 2008-10-10  20:45:29   00:04  103   0.3  11412   16396
14634 2008-10-10  20:45:30   00:05  104   0.3  12492   17444

/usr/bin/time -v

Yes ironically the GNU time command can give you the peak memory usage of a process. It reports the peak memory like so: Maximum resident set size (kbytes).

Example

$ /usr/bin/time -v ~/projects/prime_numbers/eratosthenes_prime_sieve.pl 10 1000000
...

    Command being timed: "/home/saml/projects/prime_numbers/eratosthenes_prime_sieve.pl 10 1000000"
    User time (seconds): 1.12
    System time (seconds): 0.05
    Percent of CPU this job got: 54%
    Elapsed (wall clock) time (h:mm:ss or m:ss): 0:02.19
    Average shared text size (kbytes): 0
    Average unshared data size (kbytes): 0
    Average stack size (kbytes): 0
    Average total size (kbytes): 0
    Maximum resident set size (kbytes): 79304
    Average resident set size (kbytes): 0
    Major (requiring I/O) page faults: 0
    Minor (reclaiming a frame) page faults: 20014
    Voluntary context switches: 83
    Involuntary context switches: 274
    Swaps: 0
    File system inputs: 0
    File system outputs: 0
    Socket messages sent: 0
    Socket messages received: 0
    Signals delivered: 0
    Page size (bytes): 4096
    Exit status: 0

References

• Peak memory usage of a linux/unix process
• peak memory measurement of long running process in linux

Answer 2

Even though the topic is quite old, I want to share another project that emerged from the cgroups Linux kernel feature.

https://github.com/gsauthof/cgmemtime:

cgmemtime measures the high-water RSS+CACHE memory usage of a process and its descendant processes.

To be able to do so it puts the process into its own cgroup.

For example process A allocates 10 MiB and forks a child B that allocates 20 MiB and that forks a child C that allocates 30 MiB. All three processes share a time window where their allocations result in corresponding RSS (resident set size) memory usage.

The question now is: How much memory is actually used as a result of running A?

Answer: 60 MiB

cgmemtime is the tool to answer such questions.

Usage examples would be:

$ sudo ./cgmemtime --setup -g <myusergroup> --perm 775

$ ./cgmemtime ./testa x 10 20 30
Parent PID is 27189
Allocating 10 MiBs
New Child: 27193
Allocating 20 MiBs
New Child: 27194
Allocating 30 MiBs
Child user:    0.000 s
Child sys :    0.005 s
Child wall:    6.006 s
Child high-water RSS                    :      11648 KiB
Recursive and acc. high-water RSS+CACHE :      61840 KiB

$ ./cgmemtime python -c 'print range(100000)[48517]'
48517
Child user:    0.014 s
Child sys :    0.014 s
Child wall:    0.029 s
Child high-water RSS                    :       9948 KiB
Recursive and acc. high-water RSS+CACHE :       5724 KiB