Related
I use a module (that I cannot modify) which contains a method that I need to use. This method returns 10GB of data, but also allocates 8GB of memory that it does not release. I need to use this method at the start of a script that runs for a long time, and I want to make sure the 8GB of memory are released after I run the method. What are my options here?
To be clear, the 8GB do not get reused by the script - i.e. if I create a large numpy array after running the method, extra memory is allocated for that numpy array.
I have considered running the method in a separate process using the multiprocessing module (and returning the result), but run into problems serializing the large result of the method - 10GB cannot be pickled by the default pickler, and even if I force multiprocessing to use pickle version 4 pickling has a very large memory overhead. Is there anything else I could do without being able to modify the offending module?
Edit: here is an example
from dataloader import dataloader1
result = dataloader1.get("DATA1")
As I understand it, dataloader is a Python wrapper around some C++ code using pybind11. I do not know much more about its internal workings. The code above results in 18GB being used. If I then run
del result
10GB gets freed up correctly, but 8GB continues being used (with seemingly no python objects existing any more).
Edit2: If I create a smallish numpy array (e.g. 3GB), memory usage stays at 8GB. If I delete it and instead create a 6GB numpy array, memory usage goes to 14GB and comes back down to 8GB after I delete it. I still need the 8GB released to the OS.
can you modify the function?
If the memory is held by some module, try to reload that module, (importlib.reload) which should release the memory.
If the memory is not released by th gc, it is probably because an object is store in the class that created it, so an option is to find what is this big attribute in the class (by profiling) instance and assigned it to None which may cause the gc to release the memory.
Python uses 2 different mechanisms to free memory.
Reference Counting which is employed primarily and deallocates memory as soon as it is no longer needed (eg. object lost from scope).
Garbage Collector, which is secondary and is used to collect objects with cyclic references (a -> b -> c -> a). This can be triggered using a method. Otherwise Python itself will decide, when to free memory.
However I would highly suggest profiling and chaning the code so that it does not use as much memory. Perhaps look into streams, or use a database.
I have a few related questions regarding memory usage in the following example.
If I run in the interpreter,
foo = ['bar' for _ in xrange(10000000)]
the real memory used on my machine goes up to 80.9mb. I then,
del foo
real memory goes down, but only to 30.4mb. The interpreter uses 4.4mb baseline so what is the advantage in not releasing 26mb of memory to the OS? Is it because Python is "planning ahead", thinking that you may use that much memory again?
Why does it release 50.5mb in particular - what is the amount that is released based on?
Is there a way to force Python to release all the memory that was used (if you know you won't be using that much memory again)?
NOTE
This question is different from How can I explicitly free memory in Python?
because this question primarily deals with the increase of memory usage from baseline even after the interpreter has freed objects via garbage collection (with use of gc.collect or not).
I'm guessing the question you really care about here is:
Is there a way to force Python to release all the memory that was used (if you know you won't be using that much memory again)?
No, there is not. But there is an easy workaround: child processes.
If you need 500MB of temporary storage for 5 minutes, but after that you need to run for another 2 hours and won't touch that much memory ever again, spawn a child process to do the memory-intensive work. When the child process goes away, the memory gets released.
This isn't completely trivial and free, but it's pretty easy and cheap, which is usually good enough for the trade to be worthwhile.
First, the easiest way to create a child process is with concurrent.futures (or, for 3.1 and earlier, the futures backport on PyPI):
with concurrent.futures.ProcessPoolExecutor(max_workers=1) as executor:
result = executor.submit(func, *args, **kwargs).result()
If you need a little more control, use the multiprocessing module.
The costs are:
Process startup is kind of slow on some platforms, notably Windows. We're talking milliseconds here, not minutes, and if you're spinning up one child to do 300 seconds' worth of work, you won't even notice it. But it's not free.
If the large amount of temporary memory you use really is large, doing this can cause your main program to get swapped out. Of course you're saving time in the long run, because that if that memory hung around forever it would have to lead to swapping at some point. But this can turn gradual slowness into very noticeable all-at-once (and early) delays in some use cases.
Sending large amounts of data between processes can be slow. Again, if you're talking about sending over 2K of arguments and getting back 64K of results, you won't even notice it, but if you're sending and receiving large amounts of data, you'll want to use some other mechanism (a file, mmapped or otherwise; the shared-memory APIs in multiprocessing; etc.).
Sending large amounts of data between processes means the data have to be pickleable (or, if you stick them in a file or shared memory, struct-able or ideally ctypes-able).
Memory allocated on the heap can be subject to high-water marks. This is complicated by Python's internal optimizations for allocating small objects (PyObject_Malloc) in 4 KiB pools, classed for allocation sizes at multiples of 8 bytes -- up to 256 bytes (512 bytes in 3.3). The pools themselves are in 256 KiB arenas, so if just one block in one pool is used, the entire 256 KiB arena will not be released. In Python 3.3 the small object allocator was switched to using anonymous memory maps instead of the heap, so it should perform better at releasing memory.
Additionally, the built-in types maintain freelists of previously allocated objects that may or may not use the small object allocator. The int type maintains a freelist with its own allocated memory, and clearing it requires calling PyInt_ClearFreeList(). This can be called indirectly by doing a full gc.collect.
Try it like this, and tell me what you get. Here's the link for psutil.Process.memory_info.
import os
import gc
import psutil
proc = psutil.Process(os.getpid())
gc.collect()
mem0 = proc.memory_info().rss
# create approx. 10**7 int objects and pointers
foo = ['abc' for x in range(10**7)]
mem1 = proc.memory_info().rss
# unreference, including x == 9999999
del foo, x
mem2 = proc.memory_info().rss
# collect() calls PyInt_ClearFreeList()
# or use ctypes: pythonapi.PyInt_ClearFreeList()
gc.collect()
mem3 = proc.memory_info().rss
pd = lambda x2, x1: 100.0 * (x2 - x1) / mem0
print "Allocation: %0.2f%%" % pd(mem1, mem0)
print "Unreference: %0.2f%%" % pd(mem2, mem1)
print "Collect: %0.2f%%" % pd(mem3, mem2)
print "Overall: %0.2f%%" % pd(mem3, mem0)
Output:
Allocation: 3034.36%
Unreference: -752.39%
Collect: -2279.74%
Overall: 2.23%
Edit:
I switched to measuring relative to the process VM size to eliminate the effects of other processes in the system.
The C runtime (e.g. glibc, msvcrt) shrinks the heap when contiguous free space at the top reaches a constant, dynamic, or configurable threshold. With glibc you can tune this with mallopt (M_TRIM_THRESHOLD). Given this, it isn't surprising if the heap shrinks by more -- even a lot more -- than the block that you free.
In 3.x range doesn't create a list, so the test above won't create 10 million int objects. Even if it did, the int type in 3.x is basically a 2.x long, which doesn't implement a freelist.
eryksun has answered question #1, and I've answered question #3 (the original #4), but now let's answer question #2:
Why does it release 50.5mb in particular - what is the amount that is released based on?
What it's based on is, ultimately, a whole series of coincidences inside Python and malloc that are very hard to predict.
First, depending on how you're measuring memory, you may only be measuring pages actually mapped into memory. In that case, any time a page gets swapped out by the pager, memory will show up as "freed", even though it hasn't been freed.
Or you may be measuring in-use pages, which may or may not count allocated-but-never-touched pages (on systems that optimistically over-allocate, like linux), pages that are allocated but tagged MADV_FREE, etc.
If you really are measuring allocated pages (which is actually not a very useful thing to do, but it seems to be what you're asking about), and pages have really been deallocated, two circumstances in which this can happen: Either you've used brk or equivalent to shrink the data segment (very rare nowadays), or you've used munmap or similar to release a mapped segment. (There's also theoretically a minor variant to the latter, in that there are ways to release part of a mapped segment—e.g., steal it with MAP_FIXED for a MADV_FREE segment that you immediately unmap.)
But most programs don't directly allocate things out of memory pages; they use a malloc-style allocator. When you call free, the allocator can only release pages to the OS if you just happen to be freeing the last live object in a mapping (or in the last N pages of the data segment). There's no way your application can reasonably predict this, or even detect that it happened in advance.
CPython makes this even more complicated—it has a custom 2-level object allocator on top of a custom memory allocator on top of malloc. (See the source comments for a more detailed explanation.) And on top of that, even at the C API level, much less Python, you don't even directly control when the top-level objects are deallocated.
So, when you release an object, how do you know whether it's going to release memory to the OS? Well, first you have to know that you've released the last reference (including any internal references you didn't know about), allowing the GC to deallocate it. (Unlike other implementations, at least CPython will deallocate an object as soon as it's allowed to.) This usually deallocates at least two things at the next level down (e.g., for a string, you're releasing the PyString object, and the string buffer).
If you do deallocate an object, to know whether this causes the next level down to deallocate a block of object storage, you have to know the internal state of the object allocator, as well as how it's implemented. (It obviously can't happen unless you're deallocating the last thing in the block, and even then, it may not happen.)
If you do deallocate a block of object storage, to know whether this causes a free call, you have to know the internal state of the PyMem allocator, as well as how it's implemented. (Again, you have to be deallocating the last in-use block within a malloced region, and even then, it may not happen.)
If you do free a malloced region, to know whether this causes an munmap or equivalent (or brk), you have to know the internal state of the malloc, as well as how it's implemented. And this one, unlike the others, is highly platform-specific. (And again, you generally have to be deallocating the last in-use malloc within an mmap segment, and even then, it may not happen.)
So, if you want to understand why it happened to release exactly 50.5mb, you're going to have to trace it from the bottom up. Why did malloc unmap 50.5mb worth of pages when you did those one or more free calls (for probably a bit more than 50.5mb)? You'd have to read your platform's malloc, and then walk the various tables and lists to see its current state. (On some platforms, it may even make use of system-level information, which is pretty much impossible to capture without making a snapshot of the system to inspect offline, but luckily this isn't usually a problem.) And then you have to do the same thing at the 3 levels above that.
So, the only useful answer to the question is "Because."
Unless you're doing resource-limited (e.g., embedded) development, you have no reason to care about these details.
And if you are doing resource-limited development, knowing these details is useless; you pretty much have to do an end-run around all those levels and specifically mmap the memory you need at the application level (possibly with one simple, well-understood, application-specific zone allocator in between).
First, you may want to install glances:
sudo apt-get install python-pip build-essential python-dev lm-sensors
sudo pip install psutil logutils bottle batinfo https://bitbucket.org/gleb_zhulik/py3sensors/get/tip.tar.gz zeroconf netifaces pymdstat influxdb elasticsearch potsdb statsd pystache docker-py pysnmp pika py-cpuinfo bernhard
sudo pip install glances
Then run it in the terminal!
glances
In your Python code, add at the begin of the file, the following:
import os
import gc # Garbage Collector
After using the "Big" variable (for example: myBigVar) for which, you would like to release memory, write in your python code the following:
del myBigVar
gc.collect()
In another terminal, run your python code and observe in the "glances" terminal, how the memory is managed in your system!
Good luck!
P.S. I assume you are working on a Debian or Ubuntu system
I have a few related questions regarding memory usage in the following example.
If I run in the interpreter,
foo = ['bar' for _ in xrange(10000000)]
the real memory used on my machine goes up to 80.9mb. I then,
del foo
real memory goes down, but only to 30.4mb. The interpreter uses 4.4mb baseline so what is the advantage in not releasing 26mb of memory to the OS? Is it because Python is "planning ahead", thinking that you may use that much memory again?
Why does it release 50.5mb in particular - what is the amount that is released based on?
Is there a way to force Python to release all the memory that was used (if you know you won't be using that much memory again)?
NOTE
This question is different from How can I explicitly free memory in Python?
because this question primarily deals with the increase of memory usage from baseline even after the interpreter has freed objects via garbage collection (with use of gc.collect or not).
I'm guessing the question you really care about here is:
Is there a way to force Python to release all the memory that was used (if you know you won't be using that much memory again)?
No, there is not. But there is an easy workaround: child processes.
If you need 500MB of temporary storage for 5 minutes, but after that you need to run for another 2 hours and won't touch that much memory ever again, spawn a child process to do the memory-intensive work. When the child process goes away, the memory gets released.
This isn't completely trivial and free, but it's pretty easy and cheap, which is usually good enough for the trade to be worthwhile.
First, the easiest way to create a child process is with concurrent.futures (or, for 3.1 and earlier, the futures backport on PyPI):
with concurrent.futures.ProcessPoolExecutor(max_workers=1) as executor:
result = executor.submit(func, *args, **kwargs).result()
If you need a little more control, use the multiprocessing module.
The costs are:
Process startup is kind of slow on some platforms, notably Windows. We're talking milliseconds here, not minutes, and if you're spinning up one child to do 300 seconds' worth of work, you won't even notice it. But it's not free.
If the large amount of temporary memory you use really is large, doing this can cause your main program to get swapped out. Of course you're saving time in the long run, because that if that memory hung around forever it would have to lead to swapping at some point. But this can turn gradual slowness into very noticeable all-at-once (and early) delays in some use cases.
Sending large amounts of data between processes can be slow. Again, if you're talking about sending over 2K of arguments and getting back 64K of results, you won't even notice it, but if you're sending and receiving large amounts of data, you'll want to use some other mechanism (a file, mmapped or otherwise; the shared-memory APIs in multiprocessing; etc.).
Sending large amounts of data between processes means the data have to be pickleable (or, if you stick them in a file or shared memory, struct-able or ideally ctypes-able).
Memory allocated on the heap can be subject to high-water marks. This is complicated by Python's internal optimizations for allocating small objects (PyObject_Malloc) in 4 KiB pools, classed for allocation sizes at multiples of 8 bytes -- up to 256 bytes (512 bytes in 3.3). The pools themselves are in 256 KiB arenas, so if just one block in one pool is used, the entire 256 KiB arena will not be released. In Python 3.3 the small object allocator was switched to using anonymous memory maps instead of the heap, so it should perform better at releasing memory.
Additionally, the built-in types maintain freelists of previously allocated objects that may or may not use the small object allocator. The int type maintains a freelist with its own allocated memory, and clearing it requires calling PyInt_ClearFreeList(). This can be called indirectly by doing a full gc.collect.
Try it like this, and tell me what you get. Here's the link for psutil.Process.memory_info.
import os
import gc
import psutil
proc = psutil.Process(os.getpid())
gc.collect()
mem0 = proc.memory_info().rss
# create approx. 10**7 int objects and pointers
foo = ['abc' for x in range(10**7)]
mem1 = proc.memory_info().rss
# unreference, including x == 9999999
del foo, x
mem2 = proc.memory_info().rss
# collect() calls PyInt_ClearFreeList()
# or use ctypes: pythonapi.PyInt_ClearFreeList()
gc.collect()
mem3 = proc.memory_info().rss
pd = lambda x2, x1: 100.0 * (x2 - x1) / mem0
print "Allocation: %0.2f%%" % pd(mem1, mem0)
print "Unreference: %0.2f%%" % pd(mem2, mem1)
print "Collect: %0.2f%%" % pd(mem3, mem2)
print "Overall: %0.2f%%" % pd(mem3, mem0)
Output:
Allocation: 3034.36%
Unreference: -752.39%
Collect: -2279.74%
Overall: 2.23%
Edit:
I switched to measuring relative to the process VM size to eliminate the effects of other processes in the system.
The C runtime (e.g. glibc, msvcrt) shrinks the heap when contiguous free space at the top reaches a constant, dynamic, or configurable threshold. With glibc you can tune this with mallopt (M_TRIM_THRESHOLD). Given this, it isn't surprising if the heap shrinks by more -- even a lot more -- than the block that you free.
In 3.x range doesn't create a list, so the test above won't create 10 million int objects. Even if it did, the int type in 3.x is basically a 2.x long, which doesn't implement a freelist.
eryksun has answered question #1, and I've answered question #3 (the original #4), but now let's answer question #2:
Why does it release 50.5mb in particular - what is the amount that is released based on?
What it's based on is, ultimately, a whole series of coincidences inside Python and malloc that are very hard to predict.
First, depending on how you're measuring memory, you may only be measuring pages actually mapped into memory. In that case, any time a page gets swapped out by the pager, memory will show up as "freed", even though it hasn't been freed.
Or you may be measuring in-use pages, which may or may not count allocated-but-never-touched pages (on systems that optimistically over-allocate, like linux), pages that are allocated but tagged MADV_FREE, etc.
If you really are measuring allocated pages (which is actually not a very useful thing to do, but it seems to be what you're asking about), and pages have really been deallocated, two circumstances in which this can happen: Either you've used brk or equivalent to shrink the data segment (very rare nowadays), or you've used munmap or similar to release a mapped segment. (There's also theoretically a minor variant to the latter, in that there are ways to release part of a mapped segment—e.g., steal it with MAP_FIXED for a MADV_FREE segment that you immediately unmap.)
But most programs don't directly allocate things out of memory pages; they use a malloc-style allocator. When you call free, the allocator can only release pages to the OS if you just happen to be freeing the last live object in a mapping (or in the last N pages of the data segment). There's no way your application can reasonably predict this, or even detect that it happened in advance.
CPython makes this even more complicated—it has a custom 2-level object allocator on top of a custom memory allocator on top of malloc. (See the source comments for a more detailed explanation.) And on top of that, even at the C API level, much less Python, you don't even directly control when the top-level objects are deallocated.
So, when you release an object, how do you know whether it's going to release memory to the OS? Well, first you have to know that you've released the last reference (including any internal references you didn't know about), allowing the GC to deallocate it. (Unlike other implementations, at least CPython will deallocate an object as soon as it's allowed to.) This usually deallocates at least two things at the next level down (e.g., for a string, you're releasing the PyString object, and the string buffer).
If you do deallocate an object, to know whether this causes the next level down to deallocate a block of object storage, you have to know the internal state of the object allocator, as well as how it's implemented. (It obviously can't happen unless you're deallocating the last thing in the block, and even then, it may not happen.)
If you do deallocate a block of object storage, to know whether this causes a free call, you have to know the internal state of the PyMem allocator, as well as how it's implemented. (Again, you have to be deallocating the last in-use block within a malloced region, and even then, it may not happen.)
If you do free a malloced region, to know whether this causes an munmap or equivalent (or brk), you have to know the internal state of the malloc, as well as how it's implemented. And this one, unlike the others, is highly platform-specific. (And again, you generally have to be deallocating the last in-use malloc within an mmap segment, and even then, it may not happen.)
So, if you want to understand why it happened to release exactly 50.5mb, you're going to have to trace it from the bottom up. Why did malloc unmap 50.5mb worth of pages when you did those one or more free calls (for probably a bit more than 50.5mb)? You'd have to read your platform's malloc, and then walk the various tables and lists to see its current state. (On some platforms, it may even make use of system-level information, which is pretty much impossible to capture without making a snapshot of the system to inspect offline, but luckily this isn't usually a problem.) And then you have to do the same thing at the 3 levels above that.
So, the only useful answer to the question is "Because."
Unless you're doing resource-limited (e.g., embedded) development, you have no reason to care about these details.
And if you are doing resource-limited development, knowing these details is useless; you pretty much have to do an end-run around all those levels and specifically mmap the memory you need at the application level (possibly with one simple, well-understood, application-specific zone allocator in between).
First, you may want to install glances:
sudo apt-get install python-pip build-essential python-dev lm-sensors
sudo pip install psutil logutils bottle batinfo https://bitbucket.org/gleb_zhulik/py3sensors/get/tip.tar.gz zeroconf netifaces pymdstat influxdb elasticsearch potsdb statsd pystache docker-py pysnmp pika py-cpuinfo bernhard
sudo pip install glances
Then run it in the terminal!
glances
In your Python code, add at the begin of the file, the following:
import os
import gc # Garbage Collector
After using the "Big" variable (for example: myBigVar) for which, you would like to release memory, write in your python code the following:
del myBigVar
gc.collect()
In another terminal, run your python code and observe in the "glances" terminal, how the memory is managed in your system!
Good luck!
P.S. I assume you are working on a Debian or Ubuntu system
I've been looking at a in-memory database -- and it got me thinking, how does Python handle IO that's not tied to a connection (and even data that is); for example, hashes, sets, etc.; is this a config somewhere, or is it dynamically managed based on resources; are there "easy" ways to view the effect resources are having on a real program, and simulate what the performance hit would be differing hardware setups?
NOTE: If it matters, Redis is the in-memory data store I'm looking at; there's an implementation of a wrapper for Redis datatypes so they mimic the datatypes found in Python.
Python allocates all memory that the application asks for. There is not much room for policy. The only issue is when to release memory. (C)Python immediately releases all memory that is not referenced anymore (this is also not tunable). Memory that is referenced only from itself (ie. cycles) are released by the garbage collector; this has tunable settings.
It is the operating system's decision to write some of the memory into the pagefile.
Not exactly what you're asking for, but Dowser is a Python tool for interactively browsing the memory usage of your running program. Very useful in understanding memory usage and allocation patterns.
http://www.aminus.net/wiki/Dowser
I have a Python program that dies with a MemoryError when I feed it a large file. Are there any tools that I could use to figure out what's using the memory?
This program ran fine on smaller input files. The program obviously needs some scalability improvements; I'm just trying to figure out where. "Benchmark before you optimize", as a wise person once said.
(Just to forestall the inevitable "add more RAM" answer: This is running on a 32-bit WinXP box with 4GB RAM, so Python has access to 2GB of usable memory. Adding more memory is not technically possible. Reinstalling my PC with 64-bit Windows is not practical.)
EDIT: Oops, this is a duplicate of Which Python memory profiler is recommended?
Heapy is a memory profiler for Python, which is the type of tool you need.
The simplest and lightweight way would likely be to use the built in memory query capabilities of Python, such as sys.getsizeof - just run it on your objects for a reduced problem (i.e. a smaller file) and see what takes a lot of memory.
In your case, the answer is probably very simple: Do not read the whole file at once but process the file chunk by chunk. That may be very easy or complicated depending on your usage scenario. Just for example, a MD5 checksum computation can be done much more efficiently for huge files without reading the whole file in. The latter change has dramatically reduced memory consumption in some SCons usage scenarios but was almost impossible to trace with a memory profiler.
If you still need a memory profiler: eliben already suggested sys.getsizeof. If that doesn't cut it, try Heapy or Pympler.
You asked for a tool recommendation:
Python Memory Validator allows you to monitor the memory usage, allocation locations, GC collections, object instances, memory snapshots, etc of your Python application. Windows only.
http://www.softwareverify.com/python/memory/index.html
Disclaimer: I was involved in the creation of this software.