I am not an OS expert, and I am having trouble understanding my server's memory usage. I need your advices to understand the following:
My server has 8 GB RAM and operates as web server. PHP, mySQL and Apache processes consume the majority of the memory. When I issue the command "free" after the system is rebooted, I would normally see something along these lines:
total used free shared buffers cached
Mem: 8059080 2277924 5781156 0 948 310852
-/+ buffers/cache: 1966124 6092956
Swap: 4194296 0 4092668
Obviously, sooner or later the free memory would drop and the cached memory would increase and I assume there is nothing wrong with that since the OS decides to cache it.
What I don't understand is about 1-2 days later after the machine is rebooted, I would slightly see an increase in the used swap memory. Does not this mean that the server does not have free memory anymore and using IO instead? How can I understand which processes cause this?
I am asking this question to stackoverflow users because if I ask it to my hosting provider, I am sure they would ask more money to increase RAM.
Thanks.
This is perfectly normal. When the machine starts up, a large number of services also start up. As they run their startup code, read their configuration, and so on, they dirty some pages of memory. Many of these services will never run again. By writing this data to swap, the operating system accomplishes two things:
First, if it ever does encounter memory pressure, it can discard the pages without having to write them first, since it has already written them. Second, it can discard the pages to make more free memory to enlarge the cache.
The alternative is to keep information that hasn't been touched in days in physical memory. And that just doesn't make sense.
Related
I am writing a server app which I want to efficiently use ALL available physical RAM of the machine when possible. The plan is that it will allocate physical pages using AWE until it detects that 99% of physical memory and stop when 1% is free, and any time physical memory drops below 1% free, it will free physical pages it doesn't need.
However when I put this plan into practice, Windows seems to think any time it has 99% of RAM in use it would be a good idea to free up more physical memory, and so it starts paging all sorts of stuff to disk, and my system crashes.
How can I tell Windows it is OK to have 99% of RAM in use and it doesn't need to try to page stuff back to disk until it reaches whatever its default perceived ideal level of usage is (I guess it will be something like 90%...)
Note: Raymond says 'Unless you are designing a system where you are the only program running on the computer, this is a bad idea.'
In this server scenario this is basically intended to be the only app running on the computer. But unfortunately there are some OS/background tasks that need to run...
But certainly I don't expect there is any other process on the computer indulging in this 'use all but 1% of RAM' behavior...?
Update: I've done more experimentation and started to wonder if I'm somewhat asking the wrong question. My assumption that windows is being overeager may be wrong. Perhaps the question should instead have been 'how can I determine how much physical RAM my process can safely use without compromising overall responsiveness on the machine'?
You can't. The Windows memory manager runs at a lower level than your program and knows nothing about your program (and even if it did, it has no reason to assume your program is the good citizen you claim it to be. What if your program crashes, or has an off-by-one error in a loop that mallocs? What about other programs that need memory while yours is running? What about the thousand other scenarios that the guys who wrote the Windows MM encountered when they were writing it?)
Don't try to be cleverer than Windows. A more productive use of your time would be to consider if your application really needs to allocate 99% physical memory up front.
I am running Cloudera Hadoop on my laptop and Oracle VirtualBox VM.
I have given 5.6 GB out of mine 8 and six from eight cores as well.
And still I am not able to keep it up and running.
Even without load services would not stay up and running and when I try a query at least Hive will be down within 20 minutes. And sometimes they go down like dominoes: one after another.
More memory seemed to help some: with 3GB and all services, Hue was blinking with red colors when the Hue itself managed to get up. And after rebooting it would takes 30 - 60 minutes before I manage to get the system up enough to even try running anything on it.
There has been two sensible notes (that I have managed to find):
- Warning of swapping.
- Crashing note when the system used 26 GB of virtual memory which was not enough.
My dataset is less than one megabyte, so it is hard to understand why the system would go up to dozens of gigabytes, but for whatever was reason for that has passed: now the system is running more steadily around the 5.6 GB that I have given to it after closing down a few services: see my answer to myself.
And still it is just more stable. Right after I got a warning of swapping and the Hive went down again. What could be reason for more-or-less all Hadoop services going down if the VM starts to swap?
I don't have enough reputation to post the picture to here, but when Hive went down again it was swapping 13 pages / second and utilizing 5.9 GB / 5.6 GB. So basically my system starts crashing more-or-less right after it start to swap. "428 pages were swapped to disk in the previous 15 minute(s)"
I have used default installation options as far as hard drive is concerned.
Only addition is a shared folder between Windows and VM. That works somewhat strangely locking files all the time, so I used it just like FTP and only for passing files from one system to another. Thus I can go days without using it, but systems still crash, so that is not the cause either.
Now that the system is mostly up, services crash still about twice a day: Service Monitor and Hive are quite even with their crashing frequency. After those come Activity Monitor and Event Server, which appear to crash always together. I believe Yarn crashes as well, but it gets up on its own. Last time Hive crashed first, and then it got followed by Service Monitor, Hive (second time), Activity Monitor and Event Server all.
As swap is disk, perhaps the problem is with disk:
# cat /etc/fstab
# swapoff -a
# badblocks -v /dev/VolGroup/lv_swap
Checking blocks 0 to 8388607
Checking for bad blocks (read-only test): done
Pass completed, 0 bad blocks found.
# badblocks -vw /dev/VolGroup/lv_swap
Checking for bad blocks in read-write mode
From block 0 to 8388607
Testing with pattern 0xaa: done
Reading and comparing: done
Testing with pattern 0x55: done
Reading and comparing: done
Testing with pattern 0xff: done
Reading and comparing: done
Testing with pattern 0x00: done
Reading and comparing: done
Pass completed, 0 bad blocks found.
So nothing wrong with swap disk and I have not noticed any disk error anywhere else either.
Note that you could check file system from Windows side also. But I expect that if you make Windows to fix your Linux file system, you have good chances of destroying your Linux with that, so I did my checks somewhat pessimistically, because AFAIK these commands are safe to execute.
About half of the services kept going down, so giving more specifics would be a long story.
I succeeded to get the system more stable by closing down flume, hbase, impala, ks_indexer, oozie, spark and sqoop. And by increasing more memory to some remaining services that complained they had not been given enough memory.
Also I fixed couple of thing on the Windows side, I am not sure which one of these helped:
- MsMpEng.exe kept my hard drive busy. I didn't have permissions to kill it, but I decreased its priority to lowest possible.
- CcmExec.exe got to loop on my DVD and kept reading it for forever. This I solved by taking the DVD out from the drive. Then later on I killed the process tree to keep it from bothering for a while.
I found these using Windows resource manager.
The VM requires 4GB: http://www.cloudera.com/content/cloudera-content/cloudera-docs/DemoVMs/Cloudera-QuickStart-VM/cloudera_quickstart_vm.html You should use that.
I am not clear whether you are using the QuickStart VM though. It's set up to run just the essential services and tuned to conserve memory rather than exploit lots of memory.
It sounds like you are running your own installation, on one virtual machine, on your Windows machine. You may be running an entire cluster's worth of services on one desktop machine. Each of these services has master, worker processes, monitoring processes, etc. You don't need most of them.
You also probably have left memory settings at default suitable for a server-class machine of 16+ GB RAM. Remember these services usually run across many machines, not all on one.
Finally, you're clearly swapping, and that makes things incredibly slow. Remember this is all through a VM too!
Bottom line, use the QuickStart VM if you really want a 1-machine cluster tuned correctly. If you want a real cluster or more services, you need more hardware.
Also consider: cloudera.com/live contains a full CDH 5.1 cluster + sample data, running on demand on AWS. Of course, the advantage of the VM is that you can BYOD, but if you're simply looking for a hands-on Hadoop experience, Live is a great option.
Hypervisors and Memory Management
I have been using virtual machines for years and never really had any issues. I have primarily used VMWare's free single ESXi host and had nothing but success. Because I have never had any issues I have never delved in much deeper. I have however always been very wary of loading the system up and get a lot of spare resources handy.
I have recently purchased a new server and we have decided to give Hyper-V a try and see how that goes. We have a fairly small team but utilise lots of servers for testing etc.
My question relates to memory and how much I need to leave free or available for the host machine to run appropriately.
Setup:Dell Server 24 Cores: 48GB Ram
When I run taskmgr in the windows host instance I see the following:
Physical Memory: 49139
Cached: 14933
Available: 17743
Free: 2982
What exactly do these figures mean? What is the difference between free and available?
My server uses hardly any CPU resources ever and has 10 Production servers running on it without a single user complaint ever about speed of the services.
Am I able to run up another server with 2GB ram effectivly leaving 982MB free? or am I starting to push my requirements a little?
Thanks for the help.
You shouldn’t use the host partition for anything other than Hyper-V (although you can run security and infrastructure software such as management agents, backup agents and firewalls). Therefore, that 2GB recommendation assumes you aren’t going to run any extra applications or server roles in the parent partition.
Hyper-V doesn’t let you allocate memory directly to the host partition. It essentially uses whatever memory is left over. Therefore, you have to remember to leave 2GB of your host server’s memory not allocated so it’s available for the parent partition.
Source
I am writing a simple memory manager for my application which will free any excess memory being used by my modules, so that it's available again for use. My modules allocate memory using alloc_page & kmem_cache_alloc and so the memory manager uses put_page & kmem_cache_free respectively to free up the memory which is not in use.
The problem I am facing is, even after I free the memory using put_page & kmem_cache_free, my modules are not able to get the free memory. I have written a test code which allocates a lot of memory in loop and when out of memory sleeps on memory manager to free up the memory. Memory manager successfully executes free code and wakes up the sleeping process as memory should be available now. Interestingly the alloc_page / kmem_cache_alloc calls still fail to allocate memory. Now, I am clueless why it is happening so seeking help.
I figured out the problem, it was with the API I was using. If I use
__free_page instead of put_page to free the memory pages, everything
works absolutely fine.
We have some Win32 console applications running on Windows Server 2003 Service Pack 2 that regularly fail with this:
Error 1450 (ERROR_NO_SYSTEM_RESOURCES): "Insufficient system resources exist to complete the requested service."
All the documentation we've found suggests it is linked to the number of Free System Page Table Entries running out. We have 16GB RAM in these machines and use the /3GB Operating System switch to squeeze the Windows kernel into 1GB and allow our processes access to 3GB of address space. This drastically reduces the total number of Free System Page Table Entries, so combined with our heavy use of MapViewOfFile() it is perhaps not surprising that the kernel page table entries are running out.
However, when using Performance Monitor to view the Free System Page Table Entries counter, the value is around 36,000 on reboot and doesn't go down when our application starts. I find it hard to believe that our application, which opens many large memory-mapped files, doesn't have any effect on the kernel page table. If we can't believe the counter, it's much more difficult to test the effect of any system changes we make.
There is a promising Knowledge Base article, The Performance tool does not accurately show the available Free System Page Table entries in Windows Server 2003, but it says the problem has been fixed in Service Pack 1, and we are already on Service Pack 2.
Has anyone else struggled with or solved this issue?
Update: I have checked !sysptes in windbg (debugging the kernel) and the value matches the performance counter, around 36,000. I guess this is most likely to mean that there really are that many free page table entries and Windows is telling the truth. It does leave the question of why we're getting 1450 errors though, if the PTEs are not running out.
Further update: We never did get to the bottom of why the 1450 errors were occurring. However, instead we upgraded the OS on these servers to 64-bit Windows. This allows the existing 32-bit applications (without recompilation) to access a full 4GB of virtual address space, and lets the kernel memory area with those pesky Page Table Entries be as big as it likes too. I don't think we've had a 1450 error since.
Can you try the windbg command "!sysptes" to get System PTE Information? I'm not sure if you can do this with live kernel debug, you may have to get a memory dump.
I'm not sure why you assume that ERROR_NO_SYSTEM_RESOURCES is caused only by running out of free System Page Table Entries ? As far as I know, such generic error codes are used for more than one resource type. And in fact, the first Google hit suggests that running out of file cache memory may cause it too. (KB on an XP bug, which tripped this error mode).
In your case, I'd be checking the "Handle Count". Another possible problem is address space fragmentation. If you you want to create a 1GB file mapping view, you need 1GB of free address space, and it has to be contiguous. If you map a 1GB file, a 800 MB file, and a 1GB file, close the 800MB one and open a 900MB file, the 900MB file may not fit in the hole that's left.
MS has 2 ways to allow there 32 bit OS to "deal" with hardware that has 4 GB or more of RAM.
Option 1: is what you did with the /3GB Switch in the Boot.ini.
Option 1 Pros and Cons:
(CONS) This option sucks 1 GB from the normal 2 GB kernel area - hence making the OS struggle to meet the demands of both Paged Pool allocations and kernel stack allocations. So a person might think that using the /3GB Switch will help their, but really this option is screwing the 32 bit Window OS into a slow death.
(CONS) But, This gives my App 3GB.... WRONG (Hence this is a CON) The catch is that ONLY application that have been recompiled from the vendor to be "/3GB Switch aware" can really use the extra 1 GB. Hence the whole use of the /3GB Switch is a really BAD J.O.K.E on everyone.
Read this link for a much better write-up:
http://blogs.technet.com/askperf/archive/2007/03/23/memory-management-demystifying-3gb.aspx
Option 2: Use the /PAE switch in the Boot.ini.
Option 2 Pros and Cons:
(PROS) This really this only option if you have a more then 4GB of RAM. It tricks a application by placing the complete application memory footprint in RAM. Normally, only a application "Working Set" memory is in RAM and the remaining application memory requirements go into Windows Pagefile. What is a application total memory requirements?? - it called "Virtual Size".
In my world, I have a big fat Java based IBM Product that I deal with. The server that is running the "application" has 16 GB of RAM. I simply add the /PAE switch and watch (thanks to sysinternals Processes Explorer) application paging requests go from 200 KB per sec to up to 4MB per sec.
Question: "Why"?
Answer: The whole application is in RAM.
Question: "Does the application know that it is completely running in RAM?
Answer: No - It is running that same old way that it was always run, "THINKING" that it's has part of itself as the "Working Set" memory living in RAM and the remaining application memory requirements go into Windows Pagefile.
Yes, it is that flipping GOOD.
Please Note: Microsoft has done a poor job telling anyone about the great Windows OS option. Duh
Try it and report back to stackoverflow....