theYinYeti

I started my system in "bootchart mode", with auto-generation of the PNG file at the end. Now I have my image. What are "bad signs" in this image? Where does it tell me what I can improve? Thanks, Yves.

Does it work ? I'm very curious

Apparently, your video card was not configured properly, or not at all. You should login as root (never do that in GUI mode), and execute this command: drakx11 This should configure the display. In case it doesn't work at your first attempt, try the VESA driver instead of the recommended one. Yves.

Yet I don't understand why KDE and Gnome can't have a common use of some OS-level VFS, such as FUSE... Yves.

I do have an idea, after all :) I think, I'm not sure, that your problem would be easier to solve if you had a line in /etc/fstab for your drive. The /dev/sda1 device may be many things beside this particular drive BUT nowadays, it is possible to specify a drive UID instead of the device! 1/ Plug your drive in, and execute: find /dev/disk/by-uuid/ -lname "*sda1" -printf "%P\n" This will give you the UUID of your partition. Let it be "D841-58F1" for the example; you'll have to change with the right value because that's the UUID for my USB key. 2/ In a terminal window, login as root (su - root), then execute: cp -a /etc/fstab /etc/fstab.backup; gedit /etc/fstab & BE CAREFUL because this will edit an important system file: be sure not to change the existing lines. In this file, remove the line beginning with "/dev/sda1" if there is one, then add the following line at the end of the file: UUID=D841-58F1 /media/hd ntfs-3g umask=0000,user,nls=utf8,noatime,allow_other,noauto 0 0 3/ Create the mount point (as root), unless it already exists: mkdir /media/hd 4/ Make sure the /etc/fuse.conf file exists, and if necessary create it. Ensure there's the following line in this file: user_allow_other Just this word alone on its line; this will tell fuse that users can mount fuse partitions. 5/ In Mandriva Control Center (aka "mcc"), go into System > Users setup, and add your user to the "fuse" group; this will tell fuse that you are a user allowed to mount fuse partitions. Now you should be able to mount the drive when it is connected (or maybe it will auto-mount?), and read and write to it. Yves.

You did nothing wrong. Unfortunately, to my own dismay, I've recently discovered that HAL settings are broadly overrided by a number of applications, thus rendering those kinds of tweakings partly unpredictable and often useless. I have no more ideas. I hope you'll get help from someone else. I'll keep an eye on this because I'm interested too. Yves.

You can append it to each append="..." line (inside the double-quotes of course). Yves.

Oh sorry, I realize my previous post is not usable for a hot-plugged drive. You can try what I suggested in this post but replace the contents of the file with: <deviceinfo version="0.2"> <device> <match key="volume.partition.type" string="0x07"> <match key="block.is_volume" bool="true"> <append key="volume.mount.valid_options" type="strlist">allow_other</append> <merge key="volume.policy.mount_option.umask=0000" type="bool">true</merge> <merge key="volume.policy.mount_option.allow_other" type="bool">true</merge> </match> </match> </device> </deviceinfo> Yves.

I had the same problem with a ntfs3g-mounted drive. I solved this by changing /etc/fstab, on the line referencing this drive. I set those options: umask=0000,allow_other Yves.

Yes. For the drive that is already mounted for which you want to apply changes, use the "remount" option. For your hda1 drive for example, you'd execute something like that (as root): # mount -o remount /mnt/win_c Yves.

The change by itself is harmless. So you can try. You are aware, though, that this change makes the drive read-write for everyone, aren't you? Actually, were I you, I'd rather change the "umask=0022" bit to "umask=0007,gid=NNN" where NNN is the group id where all users that must have read-write belong; others will not even have read access. To know your groups, run the "id" command in a terminal. Yves.

But to be honnest, this is a guess only, because Mandriva should handle this drive by default, and I'm afraid the above file just does what Mandriva already does, and does better...

From your "lshal" post, and the above "mount" post, we know that: - Your drive is detected OK by the system, although the blank "storage.model" HAL property is strange. - Your partition can be mounted by ntfs. Now you'll indeed probably prefer ntfs-3g over ntfs because the former allows writing on the drive. I don't know NTFS-related issues very well, but it seems to me you could create a /etc/hal/fdi/policy/something.fdi, containing: <deviceinfo version="0.2"> <device> <match key="volume.partition.type" string="0x07"> <match key="block.is_volume" bool="true"> <merge key="volume.fstype" type="string">ntfs-3g</merge> <merge key="volume.policy.desired_mount_point" type="copy_property">volume.label</merge> </match> </match> </device> </deviceinfo> Then as root: # service messagebus restart # service haldaemon restart and try plugging your drive again. Yves.

EasyUrpmi is indeed something you are advised to use. It is not something to be installed though. You have to go the EasyUrpmi site (see link on top of this page), then select "2008.0", your architecture, "urpmi", and go to step 2. There I advise you to select the top 6 checkboxes from the "Official" group (all except "*_backports"), and all checkboxes from the PLF group, with mirrors near you preferably using the "rsync" protocol. I usually enter "2008.0-" or something like that as the prefix, but you're free to differ. Step three gives you a list of commands you have to copy and paste in a console window (Konsole, or Gnome-terminal, or XTerm...) as root. To become root in a terminal window, type "su - root", and ENTER, then type root's password. Be carefull with line breaks when copying-pasting. Yves. [edit:]Ixthusdan was faster. And his advice is more complete than mine :)[/edit]

This all means that your disk is correctly seen, with sensible values. All's fine. The only line I have doubts about is "storage.model = '' (string)", because the storage model is often used as the mount point name, and it is empty here... Could you try the following commands as root, just to be sure: # cd # mkdir testmount # mount -o ro -t ntfs /dev/sda1 testmount # mount | grep sda1 # umount testmount Please give the output of the 2 mount commands. This will check that the filesystem is not corrupted, or any other problem that might prevent the partition to be mounted in any way. Yves.

Could it be a limitation of ISO9660, or UDF, whatever fs is used?

Hi ian, To my knowledge, plugdev doesn't exist in Mandriva, and I'm not aware of an equivalent group (see my thread here: https://mandrivausers.org/index.php?s=&...t&p=351384). kilimanjaro, what does this command say once the drive is connected and powered? lshal | sed -n '/sda1/,/^$/ p' Yves.

Hello, As far as I know, DVD vendors are lying: When they write 4.7GB, it is actually 4700000000B (bytes), which is 4589843KB, which is 4482MB, which is really a bit less than 4.4GB. So this would explain K3B behaviour with your 4.4GB directory, all the more considering the filesystem overhead. This shouldn't explain the single 4.0GB file, however... Yves. BTW: same goes for CDs: 700MB CDs would actually be 700000000B, that is 683593KB, which gives 667MB.

I may be wrong, but here is how I understand the original post: he would want to have two login names for the same UID and the same home directory and so on; one with the usual session (eg: KDE), the other with Compiz. This would ease a lot switching from one to another configuration: - without having to fetch the session change in the right menu of the login screen, - without having to deal with users rights and files' location caused by two different UIDs. If I understood well, then sorry I think it is impossible because as far as I know, the session configuration is stored in the home directory, hence having two session configurations for the same user would be impossible. Yves.

Yesterday was the first day I ever saw this discussion thread; I'm probably not the only one. That's the power of free software; it will survive as long as someone is interested in maintaining it; this someone may be you :) English is not my mother-tongue, so I may be wrong, but my feeling is that this sentence is somewhat aggressive... Please don't do that; you can't generalize on such things. I can't comment on bootsplash stability. However, whatever this stability is, the simple fact that it runs inside the kernel is a security risk, and many many Linux users (especially in enterprises) care a lot more about security than about a nice jpeg at boot. Quite the contrary, see my previous answer. It's just that different users have different priorities, and different needs. Hence different software may coexist for those different goals. Yves.

I've looked at the above source file. I don't see anything wrong with it. uid= is treated in a special way so that it works as expected, and umask is given somewhere as an example mount option to use, so it must be supported. Hence at the very least, files should be owned by the requested UID. As for the GID, while gid= is not treated in a special way as uid= is, at least I don't see anything that would prevent gid= to work. For now, I'll consider HAL source code sane, and not at fault. In fact, I think I found the culprits in my case: those are gnome-mount and DevTray; the former because it seems not to honnor the mount options from the lshal output, only those from the gconf tree!, the latter because it uses option '--mount-options' instead of '--extra-mount-options' when calling gnome-mount, and because there's a faulty "elif". For those using DevTray, here's my diff on DevTray-0.3.2/Handlers/Volume/__init__.py: --- __init__.ORIG.py 2008-01-21 13:54:34.000000000 +0100 +++ __init__.py 2008-01-21 13:47:15.000000000 +0100 @@ -106,10 +106,12 @@ opts += 'flush,' elif 'sync' in valid_opts and volume.o_sync.int_value: opts += 'sync,' - elif 'uid=' in valid_opts: + if 'uid=' in valid_opts: opts += 'uid=%d,' % os.getuid() + if 'gid=' in valid_opts: + opts += 'gid=%d,' % os.getgid() if opts: - cmd.append('-o') + cmd.append('--extra-mount-options') cmd.append(opts) processes.PipeThroughCommand.__init__(self, cmd, None, None) self.__error = '' @@ -700,6 +702,8 @@ mount_opts.append('flush') if 'uid=' in valid_opts: mount_opts.append('uid=%d' % os.getuid()) + if 'gid=' in valid_opts: + mount_opts.append('gid=%d' % os.getgid()) self.__volume.get_device().Mount('', fstype, mount_opts, reply_handler = self._mount_reply, error_handler = self._mount_error) From the changes I made, the change from '-o' to '--extra-mount-options' has no effect because gnome-mount doesn't get usefull default values from the HAL tree. The second set of changes (line 700) is actually useless so far because the mount method used is gnome-mount. So the changes mainly come down to the change from 'elif' to 'if', and the addition of the equivalent lines for 'gid='. Now: $ ls -lad /media/USB2FlashStorage/ drwxr-xr-x 4 yves users 16384 1970-01-01 01:00 /media/USB2FlashStorage// As far as I'm concerned, the main error is corrected. However, I'd still like to get the right umask, and generally all options specified in the HAL tree (including the right group instead of my own main group). Unfortunately, I don't know python at all, and I don't have time to look for gnome-mount source code. Yves.

Bump... Nothing, anyone? I've done a lot of search and I have no ideas left. I'm willing to do a lot more search based on the faintest trail anyone could give. Please help!

I think this DBus policy idea is a dead-end, as far as my problem is concerned, after all. I tried creating the file /etc/dbus-1/system.d/yg.usb-storage.conf: <!DOCTYPE busconfig PUBLIC "-//freedesktop//DTD D-BUS Bus Configuration 1.0//EN" "http://www.freedesktop.org/standards/dbus/1.0/busconfig.dtd"> <busconfig> <!-- HAL: allow mounted for group "disk". --> <policy group="disk"> <allow send_interface="org.freedesktop.Hal.Device.SystemPowerManagement"/> <allow send_interface="org.freedesktop.Hal.Device.VideoAdapterPM"/> <allow send_interface="org.freedesktop.Hal.Device.LaptopPanel"/> <allow send_interface="org.freedesktop.Hal.Device.Volume"/> <allow send_interface="org.freedesktop.Hal.Device.Volume.Crypto"/> </policy> </busconfig> It doesn't work. Looking at the actual contents, it is in fact no surprise. This configuration deals with rights on the bus: users from the "disk" group have the right to send data on the bus to the given interfaces. There's nothing about mounting, actually. So back to HAL. How can I configure HAL so that the mount path is owned by a given group, and with the umask I want? Yves.

Well... As I don't achieve anything, I took a new look at your link, scarecrow. I supposed it was done, since [root@localhost ~]# rpm -qa | grep -Ei 'dbus|hal' hal-info-0.0-5.20070925.1mdv2008.0 hal-0.5.10-0.rc2.4mdv2008.0 libdbus-qt-1_1-0.70-2mdv2008.0 libdbus-glib-1_2-0.74-1mdv2008.0 ndesk-dbus-glib-0.3-1mdv2008.0 perl-Net-DBus-0.33.5-1mdv2008.0 qt4-qtdbus-4.3.1-12mdv2008.0 dbus-1.0.2-10mdv2008.0 libdbus-1_3-devel-1.0.2-10mdv2008.0 python-dbus-0.82.2-2mdv2008.0 libhal1-0.5.10-0.rc2.4mdv2008.0 ndesk-dbus-0.5.2-2mdv2008.0 libdbus-glib-1_2-devel-0.74-1mdv2008.0 libdbus-1_3-1.0.2-10mdv2008.0 dbus-x11-1.0.2-10mdv2008.0 libqtdbus4-4.3.1-12mdv2008.0 I supposed it was done, since [root@localhost ~]# service messagebus status dbus-daemon (pid 8074 3282) est en cours d'exécution... [root@localhost ~]# service haldaemon status hald (pid 3733) est en cours d'exécution... I didn't do anything related to this: Mandriva doesn't have those groups, and I don't know the Mandriva equivalents, although I suspect "cdrom", "cdwriter", and "disk" are, but I'm already part of those groups. I must be deprecated for me, because: [root@localhost ~]# rpm -q hal hal-0.5.10-0.rc2.4mdv2008.0 I tried nonetheless... A closer look on the iPod example led me to some fixes on my file, which is now like that: <deviceinfo version="0.2"> <device> <match key="@block.storage_device:@storage.originating_device:info.bus" string="usb"> <match key="block.is_volume" bool="true"> <append key="volume.mount.valid_options" type="strlist">gid=</append> <merge key="volume.policy.mount_option.umask=002" type="bool">true</merge> <merge key="volume.policy.mount_option.gid=6" type="bool">true</merge> <merge key="volume.policy.desired_mount_point" type="copy_property">@block.storage_device:storage.model</merge> </match> </match> </device> </deviceinfo> There is improvement with this file. Now the key is seen, the mount point is auto-created with the device name :) and it is auto-mounted (by me=yves), and I (me: yves) can unmount it. BUT it still has the wrong rights, despite the good lshal output: [yves@localhost ~]$ ls -lad /media/USB2FlashStorage/ drwxr-xr-x 4 root root 16384 1970-01-01 01:00 /media/USB2FlashStorage// [yves@localhost ~]$ lshal | sed -n '/sda1/,/^$/ p' block.device = '/dev/sda1' (string) block.is_volume = true (bool) block.major = 8 (0x8) (int) block.minor = 1 (0x1) (int) block.storage_device = '/org/freedesktop/Hal/devices/storage_serial_Ut163_USB2FlashStorage_070801e07f725b_0_0' (string) info.capabilities = {'volume', 'block'} (string list) info.category = 'volume' (string) info.interfaces = {'org.freedesktop.Hal.Device.Volume'} (string list) info.parent = '/org/freedesktop/Hal/devices/storage_serial_Ut163_USB2FlashStorage_070801e07f725b_0_0' (string) info.product = 'Volume (vfat)' (string) info.udi = '/org/freedesktop/Hal/devices/volume_uuid_D841_58F1' (string) linux.hotplug_type = 3 (0x3) (int) linux.sysfs_path = '/sys/block/sda/sda1' (string) org.freedesktop.Hal.Device.Volume.method_argnames = {'mount_point fstype extra_options', 'extra_options', 'extra_options'} (string list) org.freedesktop.Hal.Device.Volume.method_execpaths = {'hal-storage-mount', 'hal-storage-unmount', 'hal-storage-eject'} (string list) org.freedesktop.Hal.Device.Volume.method_names = {'Mount', 'Unmount', 'Eject'} (string list) org.freedesktop.Hal.Device.Volume.method_signatures = {'ssas', 'as', 'as'} (string list) storage.model = '' (string) volume.block_size = 512 (0x200) (int) volume.fstype = 'vfat' (string) volume.fsusage = 'filesystem' (string) volume.fsversion = 'FAT16' (string) volume.ignore = false (bool) volume.is_disc = false (bool) volume.is_mounted = true (bool) volume.is_mounted_read_only = false (bool) volume.is_partition = true (bool) volume.label = '' (string) volume.linux.is_device_mapper = false (bool) volume.mount.valid_options = {'ro', 'sync', 'dirsync', 'noatime', 'nodiratime', 'noexec', 'quiet', 'remount', 'exec', 'utf8', 'shortname=', 'codepage=', 'iocharset=', 'umask=', 'dmask=', 'fmask=', 'uid=', 'flush', 'gid='} (string list) volume.mount_point = '/media/USB2FlashStorage' (string) volume.num_blocks = 1974208 (0x1e1fc0) (int) volume.partition.flags = {'boot'} (string list) volume.partition.label = '' (string) volume.partition.media_size = 1010826752 (0x3c3ffe00) (uint64) volume.partition.number = 1 (0x1) (int) volume.partition.scheme = 'mbr' (string) volume.partition.start = 32256 (0x7e00) (uint64) volume.partition.type = '0x06' (string) volume.partition.uuid = '' (string) volume.policy.desired_mount_point = 'USB2FlashStorage' (string) volume.policy.mount_option.gid=6 = true (bool) volume.policy.mount_option.iocharset=utf8 = true (bool) volume.policy.mount_option.umask=002 = true (bool) volume.size = 1010794496 (0x3c3f8000) (uint64) volume.unmount.valid_options = {'lazy'} (string list) volume.uuid = 'D841-58F1' (string) The wanted result was (but IS NOT): [yves@localhost ~]$ ls -lad /media/USB2FlashStorage/ drwxrwxr-x 4 root disk 16384 1970-01-01 01:00 /media/USB2FlashStorage// The "troubleshooting" section about DBus policy seems like a good starting point for the remaining problem, but I don't want to impose a user, just a group. I'll let you know how things go. Yves.

I browsed the DevTray source code. I found nothing. For all places I saw that checked for some values in the lshal output, I checked that the values were as expected. All was fine. Yet, the icon doesn't appear, the mount point is not created, and so on... What I don't understand is that the lshal output is really not very different from what it is when it "kind of" works (read-only). I'll keep searching... Maybe it's the group; there's something with groups in this matter... I'll try changing the group back to root, and the umask to 000. Yves.