I need to read many files very fast: reading them from the disk leads to bad performance!!I copied the files into /dev/shm, being sure that they were copied in memory, but the performance didn't improve.Then I created a tmp file system in /mnt (/mnt/tmpfs) and I mounted it withmount -osize=400m tmpfs /mnt/tmpfs -t tmpfsand copied the file in. But the performance still remain almost the same.I've the doubt that I didn't copied the file in memory!The question is: Did I make the right things?I run a FC 11 64 bit on a dual procs server with 16 Gb memory
So, I was working with Ubuntu 9.10 and I wrote a C program which was working just fine. Then, I have installed Ubuntu 10.10 and tried to run this same program again. However, I get the following error: Code: Memory Mapping failed. Error: 1 Part of the code has the following: Code: volatile ulong *memory;
int fd = open("/dev/mem", O_RDWR | O_SYNC); if (fd < 0) { printf("Could not open memory. "); return(0); } code....
Does anyone have an idea what the problem could be? I do not know if there would be any significant change between Ubuntu 9.10 and 10.10 that could be generating me this problem.
I'm working with a lot of data, but always the same. I have, say 2GB that I keep loading 100 times a day from a local disk to do some computations.I was wondering if anyone knew if it is possible to read it once for all and then access it like a file but with the speed of RAM access. I would be looking for something like: Code: file2mem ~/mybigdatafile.dat ~/mybigdata_thats_now_accessed_superfast.dat And then the data is accessible in a way like with a symlink...
As i undertsand - out of 1GB of the virtual Address space for Kernel from 3GB to 4GB of the process address space, Kernel image (code, data, bss, stack, heap) resides staring @0x0 address. Vmalloc area starts either at the end of Physical ram size or at 896M. This 896M cap is mandated to ensure that minimum of 128MB is reserved as vmalloc_reserve for vmalloc,kmap etc.
Is the understanding correct? Now trying to map Physical Zones into this 1GB address space
Initial 16MB is mapped to ZONE_DMA 16MB - 896MB is mapped to ZONE_NORMAL 896MB - 1024MB is mapped to ZONE_HIGHMEM
Does this mean that Kernel image is residing in ZONE_DMA area? Any call to vmalloc() in kernel code will return address beyond 896M? insmod of any LKM will internally invoke vmalloc() to obtain contiguous area - where will this code physically located along with rest of kernel code in ZONE_DMA or in ZONE_HIGHMEM?
I have worked in linux for a long time but never managed the system until I got my own server, which is running Fedora 14. I have a 3 TB Drive and apparently can only handle 2 TB. At least the Disk Analyzer is telling me that 2TB is 100% max capacity. Also viewing disk analyzer, I am only using 50GB of my 2TB but I am out of memory in the Root file system. If I run df -h, I get he following:
Filesystem Size Used Avail Use% Mounted on /dev/mapper/vg_dev1-lv_root 50G 40G 7.2G 85% /
I have 4 computers/users and we need to put all the files on a central server. The server is running Ubuntu 10.04. What is the best way for these 4 XP users to see the files that will be stored on the server?
Or basically, how will I either share or map the files *from* Ubuntu *to* XP? Also, the users will be reading, writing, creating and deleting files on the server.
The Completely Fair Queuing (cfq) scheduler in Red Hat Enterprise Linux 5appears to have worse I/O read performance than in version 4. It appears as though the Completely Fair Queuing I/O scheduler (cfq) has a regression and thus exhibits reduced read-side throughput which can affect performance for both local and NFS mounted file systems.
One way to mitigate this is to set the cfq's slice_idle parameter to zero. To change this value, execute the following command echo 0 > slice_idle in the /sys/block directory appropriate for your situation, as shown below:
echo 0 > /sys/block/hda/queue/iosched/slice_idle
We are using NFS file systems in RHEL 5.3. I would like to know how to find which /dev/Device is being used by the NFS file systems, so that I could try setting the slice_idle to '0' to see if there is any difference in performance? In /etc/fstab I only see the actual NAS volumes for the NFS file systems.
I am looking to buy some memory for my netbook. Currently I have 1 GB of DDR3 memory. However, the specification says that 2 GB of memory is the max. However, when I do the following it says that 4GB is the max:
Latest kernel update since Fedora 2.6.33 are mapping all my NFS "shares" twice (two sets of icons, etc.). All work, but why is this happening - was fine previously.
2.6.35.6-48.fc14.x86_64 #1 SMP Fri Oct 22 15:36:08 UTC 2010 x86_64 x86_64 x86_64 GNU/Linux Currently when I press the windows key nothing happens. However, I can get some events from this key. xev displays the following event information:
Code:
KeyPress event, serial 31, synthetic NO, window 0x3800001, root 0xaa, subw 0x0, time 1744572, (161,180), root:(584,386), tate 0x0, keycode 133 (keysym 0xffeb, Super_L), same_screen YES,
I have the Cap Locks key remapped as an additional Ctrl key. (I did that using the GUI System->Preference->Keyboard). This works fine all the time except when issuing one command to emacs. If I do CapLocks+Alt+ it does nothing yet Ctrl+Alt+ indents as needed. Since CapLocks should be the same as Ctrl I do not know what is causing the problem or how to solve it.
After a while away from Linux, I recently installed Ubuntu 10.4 on my home PC. Here is the problem: The terminal STINKS in general. Unfortunately, it appears that nothing has changed in the last 3 years in this particular area. I remember now that the terminal (keyboard mappings, colors) was one of the reasons I went with Fedora for my home PC a few years ago.
Short question: WHY is Ubuntu's so different, and why hasn't it been improved? I have a standard 104-key keyboard, yet when I try to edit with VI, the arrow keys don't map properly. You'd think that user feedback alone would have fixed this over 3 years time. Anyhow, how do I fix it?
The white background, minimal color terminal isn't very good. But I really need help getting the various keys (delete, home, end, arrow keys) to map properly. I went with Ubuntu because it seemed to be more ready "out of the box" and I don't need cutting edge like Fedora offers. I wanted more stability, etc.
Have been configuring fedora 14 to connect to a windows domain server and have been successful so far....am now on mapping network drives when the user logs in via the gnome gui.
If there is a better method of mapping network drives on login
After reading up on PAM_MOUNT and using that for mapping drives on login I have been able to successfully map them, but it doesn't do this automatically on gnome login.
Problem is as follows: It works when i connect / login using the terminal but requires me to enter the password once (even when i logged into the user account on gnome).
How I want it to work:I would like the mapping to occur when i login via gnome so that i dont have to open a terminal once logged in to gnome to map the network drives. I would like it to login without having to type the password again as the user is already logged in
how to calculate (if possible) the end address of an image file in a flash memory. I'm trying to create a checksum and checkheader function and the info that I got is the file's offset, how many sector it consumes and its size. I kinda need the end address, sad thing is, I don't know how to calculate it.test.img's start address is 0, the size is 0x20000 and consumes 3 sectors.
System: F15-64bit, Intel Core i7 on Asus P6T mobo. I've upgraded to 2.6.40, and I'm regretting it!While 2.6.38 still works fine (apart from the usual random panics), 2.6.40 gives errors on boot, and reliably panics soon after login. Early in the boot I get the message "IOMMU: mapping reserved region failed" 8 times. Then boot appears to proceed as normal, at least once the nvidia blob is removed in favour of Nouveau (otherwise, forget it...).
After graphical login, the system freezes within a couple of minutes.After a text login, the system freezes within seconds with a panic, starting "BUG: Scheduling while atomic: swapper". A forum search for the IOMMU message leads to https://fedoraproject.org/wiki/Commo...IOMMU_handling but this talks about old 32-bit releases without BIOS virtualization support.
when I delete a running executable or script, it usually (for me, pretty much always, but I don't know if it will work in every case) continues to run without any problems. So I've got two questions here: Where is the running executable/script being run from? RAM memory? If stored in RAM or where ever, is there a way to extract the executable/script from that location? If it makes any difference, I'm using Ubuntu 11.04.
I have tried to write my files to 2 (2GB each) USB sticks and both turned into a Read-Only-File-System. Then I tried with my Memory Stick Duo and it also turned into that. So I give my last try with my SanDisk HC Memory Card (from photo camera) and resulted the same. I can copy the files of those portable storages into my HDD fine but not write into them or delete any files inside those portable storages.
Using 10.04 Netbook version. I am finding on my Asus EEE 901 that sometimes file copy just seems to freeze - seems to happen usually when copying from the built-in SSD memory to the plug-in SDHC memory card. I have tried reformatting the card and using a different card. It is not just this computer since I found the same thing on my last Asus which was the 900 model.
I am told that there are issues with Nautilus. Is there anything which can be done to improve this or is there anything else which I can install besides Nautilus? I am assuming that there is some issue related to Ubuntu's handling of SDHC memory cards.
It is becoming annoying because it seems to work sometimes and then not. When it happens only option seems to be to turn the netbook off and on again. Even if the file copy is cancelled the card seems to be unaccesible until rebooted.
Also after a certain point it seems that when I try and copy new files to the card, they appear to copy ok but obviously are corrupt in some way - when you try to play videos for instance they are faulty.
i am running Ubuntu Lucid x64 as a fileserver that shares its files via SFTP, NFS and Samba. Currently the hard disks are configured to go to standby if they are not needed. This works perfectly as long as no one browses the shares or my HTPC is running: That one repeatedly looks through the shares for new music or movies. In other words my problem is that the disks are spinning up a lot more often than they should have to. Additionally the spin-up time delays the response time while browsing. Since the machine has a lot of unused RAM i want to tell the kernel that it should keep the directory structure in memory. That way the disks would not need to spin up every time someone browses through the directories.
I'm a recently proud owner of a new Aspire AS3810T with Fedora 12 installed, when it boots up before entering the boot screen when it loads I see this error message:IOMMU: Mapping reserved region failedHow can i fix this error? It's quite frustrating
Samba up and running on my pc. pc runs FC12 with kde. A laptop has win vista. The pc can access the shares on the laptop but the laptop has authentication issues to access the pc. Note that windows doesnt enforce authentication forincoming network connections.Using the system-config-samba util i tried to map a windows user to the unix user "feduser". The laptop (named LAPPY) has a user (lapuser) which has on windows no password.What should I tell samba config what the windows username should be? lapuser or LAPPYlapuser doesnt work because when accessing the pc via the laptop, the authentication fails. The only auth that is successful is when choosing the same winusername as the unix username.
Secondary, id like to setup the laptop so that the user doesnt have to provide a name and password, or at least not more then once in the lifetime of the laptop. Note that you cant provide an empty password to system-config-samba. How is that possible?
Strange but not really on issue imho:the samba - KDE control module(kcmshall4) (and the smb.conf) shows 2 shares: the homedirs and the data dir the samba server configurator (system-config-samba) shows only the datadir.
To get the kernel messages of new java process, i refer the details from /proc/<java pid>/stat and /proc/<java pid>/statm files. For some java processes, I didn't find any details in the /proc/<java pid>/statm file. It has only 7 number of 0s. But /proc/<java pid>/stat file has the details. And also this kind of process will have the life time of nearly 1 minute.
Kernel version using: Linux-2.6.18-8.1.8.el5 Is there any possibility of java process without the memory details in the /proc/<java pid>/statm file? If it is possible, how to know the memory related details of that processes?
I am having a php script which is used for bulk mailing. I run the script every minute through cron job. I have mentioned the path of the php script in a .sh file and execute the .sh file through cron job. Every time i run the script it utilizes high memory which results to server crash. how to restrict memory usage for that process to be a minimum one or how to set priority to be low for that process so that it is executed when there is no high priority jobs, so that the server runs normally without going down when the script runs
I am using malloc and frees a lot in my program. It shows its allocated but when i remove it doesnt show as the memory is removed(I am using the top command to view VIRT memory usage). If this continously grows what would happen to my program (Will it go out of memory?)
I have a computer with 16GB of ram. At the moment, top shows all the RAM is taken, (NOT by cache), but the RAM used by the various processes is very far from 16GB.I have seen this problem several times, but I don't understand what is happening.My only remedy so far has been to reboot the machine.
