[opensuse] kde partition maanger...what is aligned vs. not?
Was told kdepm might be an up-to-date replacement for system-config-lvm (not sure if suse carried it but was a gui for lvm management from redhat). But coming up kdepm gave me a bunch of warning involving my 3-disk SAS RAID5. Namely: QStandardPaths: wrong ownership on runtime directory /var/run/user/5013, 5013 instead of 0 QStandardPaths: wrong ownership on runtime directory /var/run/user/5013, 5013 instead of 0 QStandardPaths: wrong ownership on runtime directory /var/run/user/5013, 5013 instead of 0 Loaded backend plugin: "pmlibpartedbackendplugin" libGL error: No matching fbConfigs or visuals found libGL error: failed to load driver: swrast "Using backend plugin: pmlibpartedbackendplugin (1)" "Scanning devices..." getting smart status failed for "/dev/sda" : Operation not supported "Device found: LSI MR9286CV-8e" getting smart status failed for "/dev/sdb" : Operation not supported "Device found: LSI MR9286CV-8e" getting smart status failed for "/dev/sdc" : Operation not supported "Device found: DELL PERC 6/i" blkid: unknown file system type "" on "/dev/sdc4" blkid: unknown file system type "" on "/dev/sdc10" (*inPA=is not properly aligned) "Part. ‘/dev/sdc1’ inPA (last sect: 25398764, modulo: 1517)." "Part. ‘/dev/sdc2’ inPA ( 1st sect: 25398765, modulo: 1517)." "Part. ‘/dev/sdc2’ inPA (last sect: 42009974, modulo: 1399)." "Part. ‘/dev/sdc3’ inPA ( 1st sect: 42009975, modulo: 1399)." "Part. ‘/dev/sdc3’ inPA (last sect: 44130554, modulo: 251)." "Part. ‘/dev/sdc4’ inPA ( 1st sect: 44130555, modulo: 251)." "Part. ‘/dev/sdc4’ inPA (last sect: 284687864, modulo: 1529)." "Part. ‘/dev/sdc5’ inPA ( 1st sect: 44130618, modulo: 251)." "Part. ‘/dev/sdc5’ inPA (last sect: 60918479, modulo: 720)." "Part. ‘/dev/sdc6’ inPA ( 1st sect: 60918543, modulo: 720)." "Part. ‘/dev/sdc6’ inPA (last sect: 92389814, modulo: 439)." "Part. ‘/dev/sdc7’ inPA ( 1st sect: 92389878, modulo: 439)." "Part. ‘/dev/sdc7’ inPA (last sect: 113370704, modulo: 1617)." "Part. ‘/dev/sdc8’ inPA ( 1st sect: 117186560, modulo: 1985)." "Part. ‘/dev/sdc9’ inPA ( 1st sect: 156252160, modulo: 1985)." "Part. ‘/dev/sdc10’ inPA (1st sect: 175783936, modulo: 1985)." "Scan finished." partition table from fdisk -l: Disk /dev/sdc: 135.8 GiB, 145760452608 bytes, 284688384 sectors Units: sectors of 1 * 512 = 512 bytes Sector size (logical/physical): 512 bytes / 512 bytes I/O size (minimum/optimal): 512 bytes / 512 bytes Disklabel type: dos Disk identifier: 0x85830e80 Device Boot Start End Sectors Size Id Type /dev/sdc1 63 25398764 25398702 12.1G 83 Linux /dev/sdc2 25398765 42009974 16611210 7.9G 83 Linux /dev/sdc3 * 42009975 44130554 2120580 1G 83 Linux /dev/sdc4 44130555 284687864 240557310 114.7G 5 Extended /dev/sdc5 44130618 60918479 16787862 8G 82 Linux swap / Solaris /dev/sdc6 60918543 92389814 31471272 15G 83 Linux /dev/sdc7 92389878 113370704 20980827 10G 83 Linux /dev/sdc8 117186560 156250111 39063552 18.6G 83 Linux /dev/sdc9 156252160 175781887 19529728 9.3G 83 Linux /dev/sdc10 175783936 195311615 19527680 9.3G 83 Linux Near as I can tell, part1 starts at 32K (128 sectors) which seemed right, but the rest, I think I did in DOS compatible mode (pretty old format). At the time tried for alignment, but now sure doesn't look like it. Oh well. Seems like it didn't recognize 4 or 10 -- 4 is the extended partition --odd that it wouldn't recognize that. Not sure why it would have a problem with sdc10's type. Also a bit odd that it tried to read SMART status from SAS disks and RAID controllers, though theoretically it could have red SMART status for the disks in sda+sdb as they are all SATA. Another thing I don't like is that I'm certain I saw linux recognize sdb as having 4k-physical (512B logical) (type 512e - 512 emulated). But I would like the kernel to know about it "again" so it would be more likely to flag a problem if things weren't on a 4k boundary. Anyone have any of those and had luck getting kernel to recognize the physical size? Anyone else use an LVM GUI ? Thanks -linda -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 22/12/2018 05.00, Linda Walsh wrote:
Was told kdepm might be an up-to-date replacement for system-config-lvm (not sure if suse carried it but was a gui for lvm management from redhat).
But coming up kdepm gave me a bunch of warning involving my 3-disk SAS RAID5. Namely:
QStandardPaths: wrong ownership on runtime directory /var/run/user/5013, 5013 instead of 0
To me that means that the tool wants to be run as root, while it was running as plain user.
Loaded backend plugin: "pmlibpartedbackendplugin" libGL error: No matching fbConfigs or visuals found libGL error: failed to load driver: swrast "Using backend plugin: pmlibpartedbackendplugin (1)" "Scanning devices..." getting smart status failed for "/dev/sda" : Operation not supported "Device found: LSI MR9286CV-8e"
To me this means that your sda hard disk is not supported by smartctl. About the rest, I'm confused. -- Cheers / Saludos, Carlos E. R. (from oS Leap 15.0 x86_64 (Minas Tirith))
Hello, On Sat, 22 Dec 2018, Carlos E. R. wrote:
Loaded backend plugin: "pmlibpartedbackendplugin" libGL error: No matching fbConfigs or visuals found libGL error: failed to load driver: swrast "Using backend plugin: pmlibpartedbackendplugin (1)" "Scanning devices..." getting smart status failed for "/dev/sda" : Operation not supported "Device found: LSI MR9286CV-8e"
To me this means that your sda hard disk is not supported by smartctl.
Try on the cmdline, direct, with adding the appropriate type, e.g.: smartctl -d sat ... [-T permissive] or smartctl -d scsi ... [-T permissive] And yes, for smartctl, there is a difference between SATA and real SCSI devices. See 'man smartctl' under the '-d' option (search for '\-d TYPE') Speaking of which: using '-d sat' helps with some USB-SATA-Bridges.
About the rest, I'm confused.
I'm too confused by Linda's description. But the partitions clearly are not aligned to fit to 4k disks. -dnh -- / "You know how cats always land on their feet when they fall \ [ from a sufficient height? Well, so do CPU's, but they don't ] \ make such a good job of landing intact..." -- Chris King / -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 12/22/2018 2:11 PM, David Haller wrote:
Hello,
On Sat, 22 Dec 2018, Carlos E. R. wrote:
QStandardPaths: wrong ownership on runtime directory /var/run/user/5013, 5013 instead of 0
To me that means that the tool wants to be run as root, while it was running as plain user. ===
Sigh...you'll love this -- it was just the opposite. I was running as root, as I ran wireshark with 'sudo wireshark' (isn't that the normal way to run it?). But my 'XDGRUNTIME' directory was still pointing at my user's rundir. That's why it says "user/5013" (root would have "user/0"). And, _of_course_ my user-runtime dir would be owned by the user (UID=5013), and that is the complaint! It's complaining that my UID's runtime dir, "user/5013" is owned by UID 5013 (as it should be!) instead of 0." That's not really the problem I was writing about (which is really something to do with kde-partition-manager). However, I feel a minor need to explain why an environment might generate this error. (side topic)... The problem with the ownership message comes from an earlier fork in the road, where, when someone "su's (sudo's) to root to launch something that requires privilege, they can either keep their session environment variables or start a new session with root's defaults. The user is in the middle of a login session and wants to get a network trace of some program's network activity. What should happen? 1) The desktop environment should ignore the user's current environment and reinitialize things to root's login-defaults, OR 2) the environment stays the same and user gets a privileged start of wireshark and continues their current session. What does happen? By _default_: (1): sudo is configured to suspend the current session and start with a fresh login session. The users wonders why wireshark never opens and never comes back while wireshark has open'd its GUI on the console in a server room nowhere near where the user is sitting. The alternative (2) tell sudo not to reinit all those vars and start a new login session, which leads to things root running with all the ENV vars configured for the user, not root -- thus that first warning (which I was gonna skip, cuz one of the first signs of having a "weird setup" is one that works with a remote desktop (of course when everyone's desktop is from some remote machine, THEN see who's desktop still works. :-)
Loaded backend plugin:� "pmlibpartedbackendplugin" libGL error: No matching fbConfigs or visuals found libGL error: failed to load driver: swrast "Using backend plugin: pmlibpartedbackendplugin (1)" "Scanning devices..." getting smart status failed for� "/dev/sda" :� Operation not supported "Device found: LSI MR9286CV-8e" To me this means that your sda hard disk is not supported by smartctl.
It's a RAID adapter. It isn't a disk, so yeah: not suported.
Try on the cmdline, direct, with adding the appropriate type, e.g.: smartctl -d sat ... [-T permissive] smartctl -d scsi ... [-T permissive]
Hey, I dunno how permissive smartctl can get, but I really don't think it is likely to work with a RAID adapter, but maybe I'm overly conservative in my thinking...? *sigh*
And yes, for smartctl, there is a difference between SATA and real SCSI devices.
...And maybe RAID controllers? though this raid controller handles either SAS or SATA, so don't know what I'd do If I had a mix of those drives.
libGL error: No matching fbConfigs or visuals found libGL error: failed to load driver: swrast
I'm too confused by Linda's description.
FYI, the libGL error means that libGL is accelerated and can run at over 60FPS on a 2560x1600 display...because, um, libGL can't redirect my requests to a sw-rasterizer which would run at about 1-2 FPS, so it had to use an Indirect context talking to the native GLX interface on the client over an 1MB/s ethernet interface <*innocent look*>. Ah, topic!!
But the partitions clearly are not aligned to fit to 4k disks.
Well, apparently as clear for you as for me, since, (only sdb has the 4k disks):
fdisk -l /dev/sdb Disk /dev/sdb: 43.7 TiB, 48002701983744 bytes, 93755277312 sectors Disk model: MR9286CV-8e ... Device Start End Sectors Size Type /dev/sdb1 2048 93755275263 93755273216 43.7T Linux LVM
table starts at 2048 sectors, which should cover the 4k requirement, assuming lvm is internally aligned, which I believe to be the case. Even the listed case for sdc, the 1st partition starts at 32k, which would cover a 4k alignment (though the disks there only need 512), no? One of the weirdities I didn't get, anyone know what the modulus number is? -l -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 24/12/2018 01.05, L A Walsh wrote:
On 12/22/2018 2:11 PM, David Haller wrote:
Hello,
On Sat, 22 Dec 2018, Carlos E. R. wrote:
QStandardPaths: wrong ownership on runtime directory /var/run/user/5013, 5013 instead of 0
To me that means that the tool wants to be run as root, while it was running as plain user. ===
Sigh...you'll love this -- it was just the opposite. I was running as root, as I ran wireshark with 'sudo wireshark' (isn't that the normal way to run it?). But my 'XDGRUNTIME' directory was still pointing at my user's rundir. That's why it says "user/5013" (root would have "user/0"). And, _of_course_ my user-runtime dir would be owned by the user (UID=5013), and that is the complaint!
It's complaining that my UID's runtime dir, "user/5013" is owned by UID 5013 (as it should be!) instead of 0."
Ah!!
That's not really the problem I was writing about (which is really something to do with kde-partition-manager). However, I feel a minor need to explain why an environment might generate this error. (side topic)...
No, it is important, because I thought that it was running as user and that was the cause of the problems. Me, I don't use sudo, but "su -". "su" would have a similar problem. And notice that both sudo and su would write files owned by root on the user owned directories, which is not good.
Loaded backend plugin:� "pmlibpartedbackendplugin" libGL error: No matching fbConfigs or visuals found libGL error: failed to load driver: swrast "Using backend plugin: pmlibpartedbackendplugin (1)" "Scanning devices..." getting smart status failed for� "/dev/sda" :� Operation not supported "Device found: LSI MR9286CV-8e" To me this means that your sda hard disk is not supported by smartctl.
It's a RAID adapter. It isn't a disk, so yeah: not suported.
ah. A hardware raid adapter. Ok, that's different.
libGL error: No matching fbConfigs or visuals found libGL error: failed to load driver: swrast
I'm too confused by Linda's description.
FYI, the libGL error means that libGL is accelerated and can run at over 60FPS on a 2560x1600 display...because, um, libGL can't redirect my requests to a sw-rasterizer which would run at about 1-2 FPS, so it had to use an Indirect context talking to the native GLX interface on the client over an 1MB/s ethernet interface <*innocent look*>.
Ah, topic!!
Yes, please, topic ;-)
But the partitions clearly are not aligned to fit to 4k disks.
Well, apparently as clear for you as for me, since, (only sdb has the 4k disks):
And what is sdb, another raid?
fdisk -l /dev/sdb Disk /dev/sdb: 43.7 TiB, 48002701983744 bytes, 93755277312 sectors Disk model: MR9286CV-8e ... Device Start End Sectors Size Type /dev/sdb1 2048 93755275263 93755273216 43.7T Linux LVM
table starts at 2048 sectors, which should cover the 4k requirement, assuming lvm is internally aligned, which I believe to be the case.
Even the listed case for sdc, the 1st partition starts at 32k, which would cover a 4k alignment (though the disks there only need 512), no?
What says fdisk about the units? You have not pasted the complete output of fdisk.
One of the weirdities I didn't get, anyone know what the modulus number is?
I don't even know what program printed that, which is why I said I was confused... -- Cheers / Saludos, Carlos E. R. (from 42.3 x86_64 "Malachite" at Telcontar)
On 24/12/2018 10:35, L A Walsh wrote:
On 12/22/2018 2:11 PM, David Haller wrote:
Hello,
On Sat, 22 Dec 2018, Carlos E. R. wrote:
QStandardPaths: wrong ownership on runtime directory /var/run/user/5013, 5013 instead of 0
To me that means that the tool wants to be run as root, while it was running as plain user. ===
Sigh...you'll love this -- it was just the opposite. I was running as root, as I ran wireshark with 'sudo wireshark' (isn't that the normal way to run it?). [...]
Actually, for some time now the recommended way to run Wireshark has been as a normal user, but using setcap (set capabilities) to allow dumpcap access to the network data streams. See here: https://wiki.wireshark.org/CaptureSetup/CapturePrivileges The relevant excerpt: ----------------------------------------------------------------------------------------------------------- Most UNIXes Wireshark has implemented Privilege Separation which means that the Wireshark GUI (or the tshark CLI) can run as a normal user while the dumpcap capture utility runs as root. This can be achieved by installing dumpcap setuid root. The advantage of this solution is that while dumpcap is run as root the vast majority of Wireshark's code is run as a normal user (where it can do much less damage). GNU/Linux distributions, Wireshark is installed using a package manager GNU/Linux distributions usually provide package managers which handle installation, configuration and removal of software packages. Wireshark is provided by several distributions and some of them help in configuring dumpcap to allow capturing even for non-root users. Debian, Ubuntu and other Debian derivatives By installing Wireshark packages non-root users won't gain rights automatically to capture packets. To allow non-root users to capture packets follow the procedure described in the Wireshark debian/README.Debian file. Other Linux based systems or other installation methods Other Linux distributions may require that you give dumpcap sufficient privileges by hand. Setting network privileges for dumpcap if your kernel and file system support file capabilities 1. Ensure that you have installed the necessary tools, such as the setcap command. 2. setcap 'CAP_NET_RAW+eip CAP_NET_ADMIN+eip' /usr/sbin/dumpcap (NOTE: Replace /usr/sbin with /usr/bin in case you receive an error that indicates that dumpcap isn't in /usr/sbin) 3. Start Wireshark as non-root and ensure you see the list of interfaces and can do live capture. Setting network privileges for dumpcap if your kernel and file system don't support file capabilities In this case, you will need to make dumpcap set-UID to root. 1. chown root /usr/sbin/dumpcap (NOTE: Replace /usr/sbin with /usr/bin in this command and the next command in case you receive an error that indicates that dumpcap isn't in /usr/sbin) 2. chmod u+s /usr/bin/dumpcap Limiting capture permission to only one group After having set dumpcap's network privileges: 1. Create user "wireshark" in group "wireshark". 2. chgrp wireshark /usr/sbin/dumpcap 3. chmod o-rx /usr/sbin/dumpcap 5. Ensure Wireshark works only from root and from a user in the "wireshark" group" -------------------------------------------------------------------------------------------------- -- ============================================================== Rodney Baker VK5ZTV rodney.baker@iinet.net.au CCNA #CSCO12880208 ============================================================== -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 12/24/2018 09:21 PM, Rodney Baker wrote:
Actually, for some time now the recommended way to run Wireshark has been as a normal user, but using setcap (set capabilities) to allow dumpcap access to the network data streams.
See here: https://wiki.wireshark.org/CaptureSetup/CapturePrivileges
Perhaps this should be the default for OpenSUSE. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On Tuesday, 25 December 2018 13:38:23 ACDT James Knott wrote:
On 12/24/2018 09:21 PM, Rodney Baker wrote:
Actually, for some time now the recommended way to run Wireshark has been as a normal user, but using setcap (set capabilities) to allow dumpcap access to the network data streams.
See here: https://wiki.wireshark.org/CaptureSetup/CapturePrivileges
Perhaps this should be the default for OpenSUSE.
I would agree. I have no idea who the packagers are, but it would be good to see it setup this way by default. -- ============================================================== Rodney Baker VK5ZTV rodney.baker@iinet.net.au CCNA #CSCO12880208 ==============================================================
On 25/12/2018 09.19, Rodney Baker wrote:
On Tuesday, 25 December 2018 13:38:23 ACDT James Knott wrote:
On 12/24/2018 09:21 PM, Rodney Baker wrote:
Actually, for some time now the recommended way to run Wireshark has been as a normal user, but using setcap (set capabilities) to allow dumpcap access to the network data streams.
See here: https://wiki.wireshark.org/CaptureSetup/CapturePrivileges
Perhaps this should be the default for OpenSUSE.
I would agree. I have no idea who the packagers are, but it would be good to see it setup this way by default.
Just write a bugzilla :-) -- Cheers / Saludos, Carlos E. R. (from 42.3 x86_64 "Malachite" at Telcontar)
On 25/12/2018 03.21, Rodney Baker wrote:
Actually, for some time now the recommended way to run Wireshark has been as a normal user, but using setcap (set capabilities) to allow dumpcap access to the network data streams.
See here: https://wiki.wireshark.org/CaptureSetup/CapturePrivileges
Wow! Thank you for telling this tidbit! :-)))
The relevant excerpt:
1. Ensure that you have installed the necessary tools, such as the setcap command.
zypper install libcap-progs
2. setcap 'CAP_NET_RAW+eip CAP_NET_ADMIN+eip' /usr/sbin/dumpcap (NOTE: Replace /usr/sbin with /usr/bin in case you receive an error that indicates that dumpcap isn't in /usr/sbin)
Telcontar:~ # l /usr/bin/dumpcap -rwxr-xr-x 1 root root 98272 Aug 31 08:42 /usr/bin/dumpcap* <-- Green Telcontar:~ # setcap 'CAP_NET_RAW+eip CAP_NET_ADMIN+eip' /usr/bin/dumpcap Telcontar:~ # l /usr/bin/dumpcap -rwxr-xr-x 1 root root 98272 Aug 31 08:42 /usr/bin/dumpcap* <-- Red Telcontar:~ # The listing is the same, but the colour changes from green to red.
3. Start Wireshark as non-root and ensure you see the list of interfaces and can do live capture.
And it works! Thanks! -- Cheers / Saludos, Carlos E. R. (from 42.3 x86_64 "Malachite" at Telcontar)
On 2018-12-22 3:14 p.m., Carlos E. R. wrote:
About the rest, I'm confused.
It may well be an emergent property of, as you say, running an application that NEEDS root privilege as a regular user. -- A: Yes. > Q: Are you sure? >> A: Because it reverses the logical flow of conversation. >>> Q: Why is top posting frowned upon? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 12/23/2018 4:05 AM, Anton Aylward wrote:
On 2018-12-22 3:14 p.m., Carlos E. R. wrote:
About the rest, I'm confused.
It may well be an emergent property of, as you say, running an application that NEEDS root privilege as a regular user.
---- Indeed, I would say so. As I described in my 'side-issue' reply, earlier today, it is a matter of needing privilege, but that requiring root which doesn't like parts of my normal user ENV. -l -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 2018-12-23 7:08 p.m., L A Walsh wrote:
On 12/23/2018 4:05 AM, Anton Aylward wrote:
On 2018-12-22 3:14 p.m., Carlos E. R. wrote:
About the rest, I'm confused.
It may well be an emergent property of, as you say, running an application that NEEDS root privilege as a regular user.
---- Indeed, I would say so. As I described in my 'side-issue' reply, earlier today, it is a matter of needing privilege, but that requiring root which doesn't like parts of my normal user ENV.
Hmm Try running 'env' ater 1) su root 2) su - root Actually you should be able to redirect the output into two files and then run a diff or a sdiff I try this and get many KDE differences as well as a difference in LANG -- POSIT in one, en+CA in the other. It is VERY clear that one is simply running as a subshell of my normal UID and inheriting the environment whereas the other is doing clean-sheet setting for what we expect root to be. Try, for example, in each case echo $USER I'd expect the use of 'sudo' to be even more perplexing. See under "Defaults specification" It seems you can play endless games with the environment. -- STATUS QUO is Latin for "the mess we're in." -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
participants (8)
-
Anton Aylward -
Carlos E. R. -
Carlos E.R. -
David Haller -
James Knott -
L A Walsh -
Linda Walsh
-
Rodney Baker