CentOS 5 :: Why Do Files Created Via Samba Have Nobody Ownership?
Apr 21, 2011
files and directories are NOT being created with consistent ownership and permissions: when created via Samba, they are created with user/group = nobody, and when created via the OS, they are being created with user/group = root.This causes problems with our automated tools that access the server (via Samba) and do a variety of file system operations (which need root privileges).How do I cofigure Samba so files/directories are created with user/group = root?
i have created an folder inside my redhat server. and i shared it via samba and mapped that shared folders inside 5 windows machine. now the problem i am facing is, if any one create a file in that mapped drive the other user cant edit that same file. but he can read it. only for files not folders.
i gave full permission to that folders and subfolders and in smb file i gave readable writable browsable permissions. and i disabled se linux and firewall
I have just recently setup our network with an Ubuntu server which is being accessed by one Ubuntu workstation with nfs, & one XP workstation with Samba. Following instructions from the web, I set permissions like this:
sudo chmod 777 /media/Data sudo chown -R curley:curley /media/Data The problem is that files created in Samba are locked from nfs & vice/versa. I am assuming I need to add the nfs network share to the Curley group or use something more global than curley to make this work?
Samba is working ok from Ubuntu when browsing through Nautilus, but there is no access to shares when doing things like uploading or downloading files to the web (the "save file" window only shows local folders). So I guess I wouldn't mind just using Samba if I could map a folder in Samba to solve that problem.
I have a fileserver running openSUSE 11.2 and samba services for file access from MS Windows based workstations. My question relates to changing default permissions on files and directories created from the windows clients.
Following are extracts of the /etc/samba/smb.conf file :
Even with the above entries, sometimes there are files and directories created by the windows clients having permission
CentOS 5.4 install, likewise open standard install (For active directory authentication).I have a license service which requires a license.txt be in the users home directory.The group owner for license.txt must be the same as the license service. Whenever a new domain user logs in, it creates the all the appropriate files but the group owner for license.txt is the users domain group. My current workaround seems like more effort than it's worth, is there another way to get this process solved easier/more secure?
- copy the license.txt into /etc/skel
- created a script to check for the presence of license.txt, check it's permissions and change them if necessary
- gave the domain's group sudo [nopasswd] access to the script (the script is not writable)
I use a CentOS 5.4 machine as a Samba file server for a WinXP SP3 box. I noted that when I deleted files under the Samba server from WinXP, there is no trash can for them. What I delete is lost for good. Is there a Samba config setting that puts deleted files into a trash can?
Is there any working commandline alternative to # find /some/dir -group xxx -user yyy | chown xxxxx:yyyyyThe main purpose is to replace ownership and goup of certain files in subdirectories. Or nevertheless I need to write shell script for that simple operation ?
The file permissions on the folder are RW for user,group and world.(umask=0000) My main problem is with SELinux, I've tried to audit2allow and that seemed to work, all I had to do then was chcon the directory and files to type samba_share_t but the tool fails with Operation Not Supported. Am I to assume you simply cannot share files from a mounted ntfs drive under SELinux? Because I've just spent 2 hours trying and I've just about ready to just give up and just go back to windows when I need to share those folders. There's no way i can copy the folder contents to my Linux partition, far too big for that. Has anyone EVER been able to do this? Do I have to disable SELinux to do it?
I have an intermittent issue with with Samba. I can access my samba share with windows xp and vista using windows networking and even by mapping the share to a drive. The problem is that the files and folders disappear inadvertently and I can only access them again if I open the share again from the start with windows explorer. Selinux is disabled and the firewall ports for samba are open. The following software is installed:
i wanted to make a fresh installation of samba so i uninstalled and then reinstalled samba and the configuration files remained with the old configuration so i deleted the /etc/samba/ folder and now after reinstalling samba i don't have the configuration files i thought the installation process will create them ,but it did not do that
Installed 5.6 and all is well except my window pc doesn't see the centos laptop on the network, but it is on the network. I have access to the Internet via firefox. Obviously need to set some ipconfig settings, but when I try to edit ipcfg file in sbin, I get a msg, "not the owner, can't save". Is sbin the right directory? Is ipcfg the right file to edit?
After burning files to DVD+RW, the owner is changed to root, and all permissions are read only. I want to periodically open these files, update them, and save to the DVD again, but I no longer have permission and cannot change the permissions since I am no longer the owner. I tried sudo commands, but get responses "Read only file system". I have erased and reformatted the DVD and started over but get the same results. I have Ubuntu 9.04, and have tried Brasero and Nautilus and get the same problem. Am I using the wrong kind of DVD/CD?
I have a Samba share set up on a SUSE server and users connect to the share via Windows XP workstations. On SUSE, if I create a file and grant ownership to "administrator" and give it 770 permissions for example, when someone goes in to modify that file, they become the owner as soon as they save it, and the permissions change to 470 (r--rwx---+) with an access control list. I want to maintain ownership of the file myself and I don't understand why someone changing the file is changing the permissions on it...This is driving me insane because every time someone saves something I have to go in and chmod 770 it before they can save it again.
i have inherited a mixed bag of sorts: several xp users updating an access mdb with the BE on a lamp stack shared via samba. i have a backup device which gets mounted at: /media/disk... each client record (has) a folder by the companyname on the samba share, and all relative documents are placed there. when the backup script runs, it just copies newer or missing files.
someone has been renaming folders, and not matching the folder name to the related companyname from the mdb. so...the backup script captures and duplicates the data in the renamed folders. some client records also have periods in the name (not required from a data pov), such as 'Company Ltd.' instead of 'Company Ltd'. i can produce a list of company names as the folders should be found easily enough, but get a little stuck with the linux scripting.
i can easily remove and further prevent any unwanted punctuation in the company name on the client record, and create the correct folder name on the samba share with vba, but would also like to:
-for each 'client activity' folder on the backup device -rename the folder by removing punctuation marks or -delete the folder if is a dupe
i tried: ls -al | grep '&' - it properly returns only those lines with an ampersand in the folder name, but returns all folders when i try that with a '.'.
what would be the easiest method to do the renaming? i thought if there was a way to change ownership of the mounted device, then the vba code (easy to write) would be simple.
OK - i just ran chown -R on the external device, changing ownership to (me) instead of root. didn't want to because it took too long, but can now use the MoveFolder method of the filesystemobject from my app to do the renaming instead of some sort of bash script (which i was dreading).
Being new to Linux, i've just about got used to the Debian setup procedure now, but had a quick question on the default ownership of files and folders. On my default Debian installation, almost all the folders and files are owned by root:root. Is this the correct advised configuration or should the folders and files be owned by a user without root permissions - eg user:user?
I have a vary unique problem with file and directory ownership. I need to change the ownership of multiple files and directories under a specific subdirectory.Under this directory structure there are files and directories owned my different users and groups. I need to change all files and directories owned by "user1" to "user2". but if any are owned by "user3" I need those left alone.Is there a simple way to do this or will I need to traverse the structure and change things one at a time.
I have to rename a group of machines in my little samba domain (tbd backend) but there is an ugly bug that makes this impossible. have set 'rename user script' variable corectly, also checked all configurations.When i change computer name in my windows box, it shows an error saying something like "Error calling remote procedure"Looking on server side, username for the machine gets correctly changed in /usr/passwd, and also in samba database.But samba log says:
=============================================================== [2009/10/08 11:10:32, 0] lib/fault.c:fault_report(42) INTERNAL ERROR: Signal 11 in pid 11052 (3.0.33-3.7.el5_3.1)
We are having issues with LVM on CentOS 5.4 both 32bit and 64bit on one particular machine. I just can't get my head around it. Usually a device in /dev/ is created when a LVM Volume Group is created. e.g. if you add a volume group called foo, you would have /dev/foo to work with to access LVs inside the volume group.
For some reason it's not being created. The udev service is running, the volume group is created ok, I cant see anything strange in logs or strace.
I installed CentOS 5.2 and it was fine. Later I installed some third party applications. Is this possible to create again ISO image from this install? Or alternatively how one rpm package can be added to the CentOS iso cds.
I guess in most cases when extracting a tar achive ,we will get a directory with the same name as the archive file but different suffix. but in some unlucky case, as I met today, after extract a tar bar I find lots of files spread in the working directory, which is really nuisance.so what I want to learn from you is that how can I move thoes newly created files ? I know it should be some "find plus rm" fancy approch there, but I don't know exactly how.
With the find command it is easy to find files that have been modified or accessed within a given period. When a file is created, the acesss time is the same as the modify time. But as soon it is accessed (read), the access time changes, but the modify time does not. I need to find files that been accessed at all, ie. files which have access time newer than modify time. How do I do that?
I've created a script to FTP some files from a RHEL box to a EMC NAS and the script works because if I run it script manually it runs fine and transfers the files to the NAS but when I schedule it in the crontab the script runs but it doesn't transfer anything and I pipe the output of the cron job and I see message about 'passive mode refused' and lib: not a plain file, sys: not a plain file, etc...I did a vi ~/.netrc which contains the a single line of "machine xxx.xxx.xxx.xxx login ftpuser password xxx"
My script looks like: /usr/bin/ftp -i xxx.xxx.xxx.xxx <<EOF cd ftpdir
Kernel 22.214.171.124, GNU (Slackware 12.0). Bash 3.1.17.
I want to search an entire subtree of /, in the file system, for all files, with extension html, created on the hard disk. In addition, these have to be the last five created. I think I could split the problem into two parts: (a) Forget about the last condition. Then this is a job for the find command. (b) Sort the output of find using the date as the key, then use 'head' to print the desired output. But even two such simple steps are enough to justify the writing of a shell script. And here lies my weakness.
My script writing knowledge is rudimentary. What's the final purpose? Well, I lately saved four or five LQ pages onto disk containing information I consider valuable to me. But I don't exactly remember where on the disk. Then: either the problem posed is really of a very simple nature or it is not, in the latter case a script being mandatory. One of the algorithm drawbacks (the one described above) is that find may be running a great deal of time. My machine resources (RAM and CPU speed are low) are scarce and there possible are a large number of HTML files on the disk.