[expert] ML9.0:file system completely crashed ext3,"could not find mime type",kde3.0.3
Mandrake Linux 9.0 here: I installed alsa-utils from CD1 with MCC:Software Install: in /var/log/syslog these error messages appeared (SEE BELOW): "XT3-fs error (device ide0(3,2)): ext3_new_block: Allocating block in system zone - block = 294914" Then I rebooted (because I wanted to boot in runlevel 3 to see whether I could play sound..., not relevant here). Error bootmessages appeared (SEE BELOW FOR MORE DETAILS): "fsck: /dev/hda2 contains a file system with errors, check forced"... fsck: /dev/hda2: UNEXPECTED INCONSISTENCY; RUN fsck MANUALLY. failed to check filesytem. Do you want to repair the errors? Y/N (beware, you can lose data) I pressed Yes: a lot of inodes were fixed or cleared ABOUT 22345 MESSAGES APPEARED (SEE BELOW) "/dev/hda2: file system was modified " I was in runlevel 3: tried a few commands but those weren't recognized. Booted into KDE: about 25 windows appeared that said: "could not find mime type application/octet stream" or "could not find mime type application/x-executable" or "could not find mime type application/x-desktop" or "could not find mime type application/x-shellscript" Each window has to be pressed twice to get rid of. There are some Konqueror windows open from the former session, I can use them to browse the internet or my file system. I can not start programs from the menu, in my file system a lot of files have now file type mime. FINALLY I FOUND OUT THAT A LOT OF FILES WERE GONE, FROM DIFFERENT DIRECTORIES. I also have a Konsole window open: except for basic commands no command is recognized. I searched Google, found nothing useful, search function of [EMAIL PROTECTED] mailing list didn't work. Is there an easy way to repair my system or do I have to reinstall everything?? I have a backup from 3 weeks ago, but that won't bring back the lost files from /bin, /user/sbin,etc. UPDATE: I searched Google for "ext3_new_block: Allocating block in system zone": I found the ext3-user mailing-list and read and learned a bit about file system corruption. I tested my memory with memtest86 but no errors were detected, I read that this is not conclusive. I probably will reinstall Mandrake Linux 9.0. But how can I figure out what has caused the corruption: Memory, hard disk, IDE cable, IDE controller, UDMA, CPU, cooling,...? And how can I best recover from this: reinstall complete distribution and then apply backups? How can I figure out which files have been modified or disappeared? I also remember that in the same session KNode newsreader crashed when I tried to read a newsmessage after downloading the headers of alt.os.linux.mandrake. It crashed several times. My system: AMD AthlonXP 1600+ Motherboard MSI K7T266 Pro2 (MS-6380 V2.0): VIA KT266A, VIA VT8233 512 MB memory (tested it with memtest86: no errrors detected) Western Digital Caviar 40GB 7200RPM: UDMA is enabled I think (see below for info from dmesg) Mandrake Linux 9.0 (kernel 2.4.19-16mdk) On this disk I also run: WinXP,Win98,Mandrake8.2(ext2) : no problem with these OSes so far. Very frustrated after this happened, I thought linux file systems were stable. UPDATE: No longer very frustrated, just a little; this is the first time in six years I experience such a problem. vatbier ERROR MESSAGES AFTER INSTALLATION ALSA-UTILS: LOOK AT LINES WITH * IN FRONT OF IT: Jan 31 13:55:22 this rpmdrake[1711]: Extracting header of xine-alsa-0.9.13-3mdk.i586 from /var/lib/urpmi/hdlist.Internationa Jan 31 13:55:22 this rpmdrake[1711]: removed files/directories (recursively) /root/tmp/headers/ Jan 31 13:55:26 this rpmdrake[1711]: removed files/directories (recursively) /root/tmp/headers/ *Jan 31 13:55:53 this kernel: EXT3-fs error (device ide0(3,2)): *ext3_new_block: Allocating block in system zone - block = 294914 *Jan 31 13:55:53 this kernel: EXT3-fs error (device ide0(3,2)): *ext3_new_block: Allocating block in system zone - block = 294915 Jan 31 14:01:00 this CROND[1754]: (root) CMD (nice -n 19 run-parts /etc/cron.hourly) Jan 31 14:03:40 this rpmdrake[1711]: Installing package removable://mnt/cdrom/Mandrake/RPMS//alsa-utils-0.9.0-0.6rc2mdk Jan 31 14:03:46 this kernel: ISO 9660 Extensions: Microsoft Joliet Level 3 Jan 31 14:03:46 this kernel: ISO 9660 Extensions: RRIP_1991A *Jan 31 14:03:49 this kernel: EXT3-fs error (device ide0(3,2)): *ext3_new_block: Allocating block in system zone - block = 294916 *Jan 31 14:03:49 this kernel: EXT3-fs error (device ide0(3,2)): *ext3_new_block: Allocating block in system zone - block = 294917 ABOUT 190 OF THESE ERROR MESSAGES APPEARED. BOOTING: ABOUT 22345 MESSAGES APPEARED: Jan 31 14:29:22 this fsck: /dev/hda2 contains a file system with errors, check forced Jan 31 14:29:25 this /etc/hotplug/usb.agent: ... no modules for USB product 0/0/0 Jan 31 14:29:36 this fsck: Inode 139427 is in use, but has dtime set. FIXED. Jan 31 14:29:36 this fsck: /de
Re: [expert] ML9.0:file system completely crashed ext3,"could not find mime type",kde3.0.3
civileme <[EMAIL PROTECTED]> wrote: > > WD drives of the generation you are using do not calculate that same > CRC for comparison--they store it in their big 600-byte sectors along > with the 512 bytes of data. They may calculate the first in a series > of writes, but not all. > >The drive is safe enough run at udma2 (33Mhz and 32 byte CRCs) or > > If you cannot put a better drive in that box, a 40-pin cable could > make a big difference for the safe operation, to force 33MHz or lower > transfer speed. Thanks! I had already read that WD and UDMA5 give problems but my hard drive WD400BB-32CXA was bought in april 2002. Were can I find the generation info that my drive has no good CRC checking at UDMA3,4,5? I searched the Western Digital website (wdc.com) but they don't give that kind of information. Do the new WD400BB drives do correct CRC checking? You say "They may calculate the first in a series of writes, but not all.": where can I find that kind of insight info, I searched the net but apart from some of your postings haven't found much else. I don't have a 40-pin cable. If I use the hdparm command hdparm -k1 -d1 -X66 /dev/hda (66:UDMA2), would this give the same effect? Can changing the UDMA setting with hdparm give problems/errors? I also downloaded from the WDC website Data Lifeguard Tools that can change the udma setting on the hard drive itself. If I used this to change the UDMA to 2, would my linux have no problems with this (I'm a a little scared to use hdparm or WDC tools, in "man hdparm" they state "Use with extreme caution! This feature includes zero protection for the unwary, and an unsuccessful outcome may result in severe filesystem corruption!") ? How can I prevent CRC errors? Can an IDE-cable suddenly become bad, or experience interference, I read about twisted cables, ... "The HDD should NOT be mounted near the power supply due to fan issues. The fan in the supply causes the HDD to misinterpret CRC-checksums and produces CRC-errors. Also, high-speed CD drives should not be near the HDD" : fan causing CRC-errors? CD drives ? How much slower is UDMA2 than UDMA5? (hdparm -t), You said somewhere that the performance does not differ much? Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
Re: [expert] ML9.0:file system completely crashed ext3,"could not find mime type",kde3.0.3
civileme <[EMAIL PROTECTED]> wrote: > The X66 is safe to use as you have described it. What is dangerous is > overdriving. hdparm -k1 -d1 -X66 /dev/hda (66:UDMA2) I guess you mean if I change the hard drive setting with DataLifeGuard and then do "hdparm -d1 X69" (69:UDMA5) that this is dangerous, or if I try to set it to UDMA6 (:133Mhz)? Mandrake 9.0 uses UDMA5 automatically (without program hdparm), is this configured on booting are is it written at installation of ML9.0 in some configuration file? What program in Mandrake 9.0 is responsible for this, or is it just the kernel that figures out what the highest UDMA setting is? > Also, IIRC, DataLifeGuard phones home and gives you warning if your > drive is about to fail (and it works on non-WDs too). DataLifeGuard is just an utility on a floppy disk, or does it install a phone home program in Windows XP? > As for the variance of speed, consider this... Ok, then I'll probably set it with hdparm to UDMA2. Hm, where is the -k1 option for hdparm written to? man hdparm doesn't talk about a configuration file. > yes check out man elvtune man elvtune: I/O elevator tuner: don't know enough about it to understand this. > a secondary cosmic ray is a significant noise source. Not much charge > moves on those cables at the interface voltage in that time span. "secondary cosmic ray": didn't know it was that delicate, how are IDE-cables protected on the Space Station and what speed would they have? > You can download an ATA66 disable utility (windows compatible) for > your WD. I think this utility is also on the DataLifeGuard floppy disk dlgudma.exe - Data Lifeguard Ultra ATA Management Something else I wonder: A lot of system files (e.g. /usr/bin/play) disappeared, they were cleared because of corruption (bad mode,deleted/unused inode,illegal character device,...). Did the corruption occur with these messages in syslog: "EXT3-fs error (device ide0(3,2)): ext3_new_block: Allocating block in system zone - block = 294914" ? What do these messages mean? I have no idea how on low-level files are written to hard drives. I guess that over the IDE-cable a command must be sent (write/read) with an address (where to store/read on the hard drive) and the data that must be written ( how is this done? Do 512 bytes of data have to be written even if you only want one bit changed? I guess from Civilemes postings that sectors of data on a hard drive are 512 bytes, is this correct? Or do you have to write a full cluster (: group of sectors) even if you want one bit changed?) How could the corruption on my file system have occurred? Was a "read" command changed to a "write" command, was an address changed into another address so that an important part of the file system got overwritten that resulted in massive file corruption? In syslog there were about 22345 messages indicating inodes, files as corrupt. How many writes are needed to corrupt that many files? Did it just occur with one command or was my system producing random data corruption during several seconds? Or is it possible that the data on my hard drive was OK, but that it got corrupted when it got read by my system? Does Linux check, reread when it encounters CRC-errors? Is it possible that data corruption occurs without being noticed immediately? Is it only noticed when a damaged block is read and compared to its CRC? On another partition I have Windows XP: could corruption just as well have occurred during a Windows XP session? Would Windows XP also have noticed it at the next booting? I read that Western Digital supports only Windows and Solaris, does this mean that its drivers for these OSes protect better against CRC-errors in UDMA 3,4,5? Can Linux protect against these possible errors by rereading the hard drive after a write command and thus checking if it was written correctly? Would a double check like this seriously affect performance? Last question: what hard drives do you recommend: Maxtor, Seagate? I read through some hard drive forums, isn't clear which hard drives are the best. vatbier Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
Re: [expert] ML9.0:file system completely crashed ext3,"could not find mime type",kde3.0.3
J. Grant wrote: > > Mandrake 9.0 uses UDMA5 automatically (without program hdparm), > > is this configured on booting are is it written at installation of > > ML9.0 in some configuration file? > mdk9 does not use UDMA5 by default. Your HD/BIOS initalise themselves > to what they consider to be the fastest stable speed and setttings etc > less /etc/sysconfig/harddisks > You can override what mdk9 has in that file to change stuff, default > is often very slow. Everything in that file is commented out, ah now I understand Todd's reply ("Default is everything commented out, so it's whatever the kernel natively recognizes. hdparm is only necessary if the kernel doesn't quite recognize the ide controller chipset and you want to try to wring every bit of performance you can out of it. Todd") > > What program in Mandrake 9.0 is responsible for this, or is it just > > the kernel that figures out what the highest UDMA setting is? > man hdparm No, hdparm is not installed in my freshly installed Mandrake 9.0 yet dmesg shows: "Uniform Multi-Platform E-IDE driver Revision: 7.00alpha2 ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx VP_IDE: IDE controller on PCI bus 00 dev 89 VP_IDE: chipset revision 6 VP_IDE: not 100% native mode: will probe irqs later VP_IDE: VIA vt8233 (rev 00) IDE UDMA100 controller on pci00:11.1 ide0: BM-DMA at 0xfc00-0xfc07, BIOS settings: hda:DMA, hdb:pio ide1: BM-DMA at 0xfc08-0xfc0f, BIOS settings: hdc:DMA, hdd:pio hda: WDC WD400BB-32CXA0, ATA DISK drive hdc: LITE-ON LTR-24102B, ATAPI CD/DVD-ROM drive ide0 at 0x1f0-0x1f7,0x3f6 on irq 14 ide1 at 0x170-0x177,0x376 on irq 15 hda: 78165360 sectors (40021 MB) w/2048KiB Cache, CHS=4865/255/63, UDMA(100)" for hda it reports UDMA(100), the ide controller chipset is recognized as VIA vt8233 (rev 00) IDE UDMA100 controller. I think it is this part of the kernel that is responsible for it: "usr/src/linux-2.4.19-16mdk/drivers/ide/pci/via82cxxx.c" > > I think this utility is also on the DataLifeGuard floppy disk > > dlgudma.exe - Data Lifeguard Ultra ATA Management > I'm not sure why are considering this program as it will never work > on GNU/Linux or GNU/Wine. No, it is a booting floppy disk, the utility changes the UDMA setting on the hard disk itself. > > Did the corruption occur with these messages in syslog: > > "EXT3-fs error (device ide0(3,2)): ext3_new_block: Allocating block > > in system zone - block = 294914" ? What do these messages mean? > Did you install with a search for badblocks? No, but on my new install of ML9.0 I did, install didn't tell if it found any badblocks. > e2fsck -f -c /dev/hdX can do it for you now, make sure you are in > single user mode though and your HD is mounted ro. I already reinstalled ML9.0, but I tried on this new install "e2fsck -f -c /dev/hda2": and it returned a bunch of errors. And this is after a fresh install. I'll complain about this in another thread. vatbier Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
[expert] ML9.0: patch VIA to enable UDMA100 not installed but yet reporting UDMA100, Cycle time 120 ns???
I WROTE THE Western Digital Data Lifeguard Utilities to a floppy disk: tested my WD400BB-32CXA0 No errors. Changed with the Ultra Management utility the UDMA setting on the disk itself from UDMA5 (100) TO UDMA2 (33): ML9.0 and ML8.2 report in dmesg: UDMA(33) Win XP reports also UDMA2. I tested in ML8.2: hdparm /dev/hda: using_dma=1(on) hdparm -Tt /dev/hda: 27,23 MB/s hdparm -I /dev/hda: DMA ... *udma2 cycle time: min=120ns recommended=120ns Changed back from UDMA2 (33) to UDMA5 (100) Now I got these results: hdparm -Tt /dev/hda: 27 MB/s HUH?SAME AS UDMA2 ?? hdparm -I /dev/hda: DMA ... *udma5 cycle time: min=120ns recommended=120ns HUH? 120ns?? SEE http://www.pcguide.com/ref/hdd/if/ide/modes.htm UDMA: " This table shows all of the current Ultra DMA modes, along with their cycle times and maximum transfer rates: Ultra DMA Cycle Time Maximum Transfer Mode (nanoseconds)Rate (MB/s) Mode 0 240 16.7 Mode 1 160 25.0 Mode 2 120 33.3 Mode 3 90 44.4 Mode 4 60 66.7 Mode 5 40 100.0 The cycle time shows the speed of the interface clock; the clock's frequency is the reciprocal of this number (see here for more on clocking.) The maximum transfer rate is four times the reciprocal of the cycle time--double transition clocking means each cycle has two transfers, and each transfer moves two bytes (16 bits). " UDMA5 should have cycle time: 40ns Why does hdparm -I /dev/hda report "cycle time: min=120ns recommended=120ns" ? Is this a bug of hdparm? Or does the cycle time in hdparm means something else? Or could it be that the UDMA5 of my hard drive gives the same speed as UDMA2? Then I found on viaarena.com this patch for ML8.2: VIA ML8.2 ATA133-100 Patch 686B-8231-8233x-8235 ver 0.91A.gz contains a patch file mdk8.2-patch-2.4.18-vpide.gz date: 20/9/02 and a readme.pdf: VIA ML8.2 ATA133-100 Patch Readme.pdf: "hdparm -t /dev/hda: sustained transfer rate of the disk reads: VT8233 Original Kernel 23,7 MB/sec rebuilt kernel with 39,75 MB/sec Patch IDE Success of kernel patching: dmesg: "VP_IDE: VIA vt8233 (rev 00) IDE UDMA100 controller on pci00:11.1 Note the data transfer for VT8233 has changed from 33 to 100" HUH MY dmesg already states UDMA100 but hdparm -t /dev/hda on my ML8.2 is only 26,67 MB/sec Hm, I remember that my ML8.2 has no original kernel, I installed a new one from cooker during the summer. So either my kernel is already patched and for some other reason only manages 26,67 MB/sec for sustained transfer rate or it isn't patched yet and incorrectly states that I have a UDMA100 driver. hdparm on my ML9.0 shows the same speeds. Does ML9.0 has this patch from VIA already included? How in Linux can I test the burst speed which should come close to the UDMA speed: 100 MB/sec (in article below: +/- 87 MB/sec) ? In this article they tested speeds in Windows "http://www.tecchannel.de/hardware/817/index.html: Client / Komponenten / VIA Chipsets slow down PCI cards: Vom 21.12.2001, Update am 04.02.2002 VIA Chipsets slow down PCI cards: a driver patch is available to boost burst speed closer to 100 MB/sec ... The speed while doing sequential transfers from the hard drive media itself is actually much lower than that. High end IDE drives max out at about 40 MBytes/s currently. >>>>AHA BUT I ONLY GET >>>> 26,67 MB/sec If you consider the overhead, these drives can only push older Ultra-ATA/66-controllers to their limits." What is a normal sustained transfer rate for UDMA5 in Linux ( what does "hdparm -t /dev/hda" return on your computer)? How in Win XP can I test the sustained transfer rate like in Linux with the command "hdparm -t /dev/hda" ? Would it be the same speed as in Linux ? vatbier Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
[expert] ML9.0: Fresh install, found one file 2 GB big, "e2fsck -f -c /dev/hda2" returns errors:
I reinstalled ML9.0 (for the third time) on 9/2/03 I found just by coincidence in /lib/modules/2.4.19-16mdk/kernel/drivers/scsi: 53c7,8xx.o.gz 2GB big !! I repeat 2GB big (date was normal 20/9/02) and the partition is only 2GB big. What happened ??? I haven't done much configuration yet to that install, last thing I did with it was make a backup. I guess at install something must have gone wrong. >From my ML8.2 I did a "e2fsck -f -c /dev/hda2" (suggested by J.Grant) to search for badblocks. I first changed in /etc/fstab hda2 to ro (read-only). "e2fsck -f -c /dev/hda2" gave this: "[root@mycom root]# e2fsck -f -c /dev/hda2 e2fsck 1.26 (3-Feb-2002) /dev/hda2 is mounted. WARNING!!! Running e2fsck on a mounted filesystem may cause SEVERE filesystem damage. Do you really want to continue (y/n)? yes Checking for bad blocks (read-only test): done Pass 1: Checking inodes, blocks, and sizes Pass 2: Checking directory structure Pass 3: Checking directory connectivity /lost+found not found. Create? yes Pass 4: Checking reference counts Pass 5: Checking group summary information Free blocks count wrong for group #3 (13860, counted=13859). Fix? no Free blocks count wrong (132161, counted=132160). Fix? no Free inodes count wrong for group #0 (8528, counted=8527). Fix? no Directories count wrong for group #0 (527, counted=528). Fix? no Free inodes count wrong (145295, counted=145294). Fix? no /dev/hda2: * FILE SYSTEM WAS MODIFIED * /dev/hda2: ** WARNING: Filesystem still has errors ** /dev/hda2: 118001/263296 files (0.2% non-contiguous), 393967/526128 blocks" I pressed no when asked to fix things, when it asked to create a /lost+found DIRECTORY I had to say yes: it created a /lost+found on /dev/hda2 allthough in /etc/fstab it says ro. How come that a directory could be created by e2fsck in a partition that has read-only access? I'm not sure what e2fsck does, I think it didn't find badblocks ("Checking for bad blocks (read-only test): done"), is this correct? I'll have to reinstall ML9.0 ( for the 4th time, bah). How can I be certain that it'll be installed correctly? How many other files have been compromised? "rpm -Va" isn't sufficient to check a complete linux installation, is it? I don't know what the errors above mean. And what has caused this corruption of files? Is it memory, hard disk, IDE cable, IDE controller, UDMA, CPU, cooling, CD-Writer,...? Did it happen at installation of ML9.0, afterwards? In thread "ML9.0:file system completely crashed ext3,"could not find mime type",kde3.0.3" I began to think it might have been the bad UDMA5 implementation of my Western Digital hard drive but now I'm not sure wether it might be because of something else. I tested my 512 MB memory a couple of times with memtest86 (not enough I know), I tested with a utility from WD my hard drive: didn't found any errors. J.Grant replied: "If they were badblock errors your HD has failed, replace it and your problems might go away." vatbier Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
Re: [expert] ML9.0: patch VIA to enable UDMA100 not installed but yet reporting UDMA100, Cycle time 120 ns???
--- vatbier <[EMAIL PROTECTED]> wrote: > Then I found on viaarena.com this patch for ML8.2: > VIA ML8.2 ATA133-100 Patch 686B-8231-8233x-8235 ver 0.91A.gz > contains a patch file mdk8.2-patch-2.4.18-vpide.gz date: 20/9/02 > and a readme.pdf: > VIA ML8.2 ATA133-100 Patch Readme.pdf: > > "hdparm -t /dev/hda: sustained transfer rate of the disk reads: > ... > VT8233 Original Kernel 23,7 MB/sec >rebuilt kernel with 39,75 MB/sec > Patch IDE > > So either my kernel is already patched and for some other reason only > manages 26,67 MB/sec for sustained transfer rate or it isn't patched > yet and incorrectly states that I have a UDMA100 driver. > > hdparm on my ML9.0 shows the same speeds. Does ML9.0 has this patch > from > VIA already included? > I looked at this patch mdk8.2-patch-2.4.18-vpide and compared it to ML9.0: usr/src/linux-2.4.19-16mdk/drivers/ide/pci/via82cxxx.c /usr/src/linux-2.4.19-16mdk/include/linux/pci_ids.h The via82cxxx.c of ML9.0 seems updated with this patch (but not entirely), but pci_ids.h does not (e.g. "-#define PCI_DEVICE_ID_VIA_8233_00x3074 +#define PCI_DEVICE_ID_VIA_8233 0x3074" in pci_ids.h it still reads "VIA_8233_0"). I'm not familiar with looking at patches but I get the feeling it mostly adds udma133 support. I have VIA 8233 and ML9.0:via82cxxx.c looks like I should get UDMA100 speeds for hdparm -t /dev/hda. But I'm getting just 26,67 MB/sec instead of the near 40 MB/sec speed they promise. Or there other parameters that influence hdparm -t /dev/hda? (Somewhere I read that "I/O support = 1(32-bit)" should be set to 3, or set "unmaskirq = 1 (on)" to 0 for better performances). So again what does hdparm -t /dev/hda return on your computer (with or without VIA 8233 and/or Western Digital hard drives) ? vatbier __ Do you Yahoo!? Yahoo! Shopping - Send Flowers for Valentine's Day http://shopping.yahoo.com Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
[expert] ML9.0:are services fam, portmap and xinetd necessary on a stand-alone desktop computer
Mandrake Linux 9.0 here. Are services fam, portmap and xinetd necessary on a stand-alone desktop computer connected to the internet just for surfing, reading mails, newsgroups? No printer is connected (cups is not installed). Does Konqueror (KDE 3.0.3) really need fam (and portmap) or is it just non-essential, just for the fun to see a file grow as it is being downloaded? What programs use and/or need fam? (I don't have NFS mounteddirectories) What programs use and/or need portmap? (I don't have RPC servers (NFS, NIS,... ) running I think) Is xinetd necessary? Are only servers listed in /etc/xinetd.d/ used/started by xinetd? If I use none of these servers is it safe to stop xinetd? (/etc/xinetd.d/cvs is to setup a cvs server, correct? it isn't needed for the command "cvs checkout" to download a source repository) vatbier After I wrote the above I found this: lists.kde.org :kde-devel From Josef Weidendorfer 2002-11-29: > Oh, and FAM causes millions of updates on a local open directory with > Konqueror, just when I retag an MP3/Ogg file with EasyTAG. After it's > finished tagging, Konqueror keeps flickering like mad for 10 seconds > > and I can't use any other Konqueror windows. Disabling service >sgi-fam in xinetd does the trick, and Konqueror updates directories >again, at a sane speed. But isn't fam supposed to help performance by >reducing reads to find out whether an element has changed? Hi, the reasoning behind FAM is to keep CPU resources low for applications that want reaction on file changes. The need for this is not given when only one application is looking for changes in a local directory: it can do polling at reasonable intervals. This is the case with Konqui when you disable FAM and dnotify. As you say, you get perfect and reasonable behaviour. The reason for FAM is given when * there is a large number of applications looking for changes in the same directory. FAM can centralize the actions needed, but notify each application on a change. This would be needed if KDE applications keep an eye on their config files: Each KDE app would need to listen for changes to the file with global KDE configuration, too. Here, the centralized approach of FAM is helpful. You can specify FAM to only do STAT polling instead of using kernel support (e.g. DNOTIFY). * the directory to watch for changes is remotely mounted. As STATs over NFS are slow, you create unneeded network loading. When the FAM daemon is running on the server machine, the local FAM on the file server looks for local changes (perhaps again using only STAT polling, but that's fast when local). When a change is detected, the server FAM notifies all watching client FAMs and these demultiplex the notification for each application on their machine which are interested in changes. The centralized approach of FAM is needed here, too: Suppose 1000 clients doing STAT polling over NFS instead of using FAM... How does FAM know about changes on local files? There is either - regular polling with STAT - using some kernel support: imon (IRIX), dnotify (LINUX), kqueue (BSD?) STAT polling every 0,5 seconds is mostly enough (configured when compiling FAM). Disadvantage is a maximal delay of 0,5 s for events to be delivered. STAT does not open the watched files are make a watched dir its cwd, so there will never be any problems with "device busy". Umounting devices is no problem with STAT polling. If there are huge numbers of files to be watched, or you want immediatly notified about changes, use kernel support: * imon uses the /dev/imon pseudo device to get notifications from the kernel. There's also a implementation for LINUX, but because of bad shape of this code, it's not in any legacy kernel and AFAIK no distribution has it build in. BTW, this is the IRIX way on SGI machines. * Don't know about kqueue (is this the right name for this service on BSD?) * dnotify on LINUX: This is in the legacy kernel. And that's the reason all distributions seem to have FAM built to use it (SUSE, Red Hat, Mandrake...). Drawbacks of DNOTIFY: - DNOTIFY forces FAM to open the directory which contains files to be watched. This means problems with unmounting devices. Simple advice: Kill the FAM daemon before unmounting ;-) - DNOTIFY has no event delaying. For every write syscall to a watched file, DNOTIFY produces immediatly a signal (for FAM). Worse, all DNOTIFY applications (FAM, KDirWatch with dnotify support) use realtime signals. These aren't merged by the kernel but queued and could lead to overrunning this queue (thus, the additional need for SIGIO handler). But this is needed as an attribute for the signal is the changed directory file descriptor, and thus the application knows the directory where the change happened. When using a normal signal for dnotify, events
Re: [expert] ML9.0:are services fam, portmap and xinetd necessary on a stand-alone desktop computer
On Saturday January 18 2003 04:52 am, Tom Brinkman wrote: > http://www.mandrakeuser.org/docs/admin/index.html#ss has lists of > all system services, even a deprecated services list. For each, they > are rated optional, common, or essential. Many of the listed > services have links to more documentation for that service. IOW's, > much more info than you get from MCC. > About fam service there is no documentation, the links to more documentation I have already tried. Again: What programs (in Mandrake Linux) use fam, use portmap ? Is it save to not use xinetd? Pfff I can't think right now on this matter, I've to do something else. vatbier Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
[expert] ML9.1: Why difference between screen position of ML8.2 and ML9.1 with same settings in "xvidtune -show" ?
I have two linux OSes on my computer: ML8.2 and ML9.1. I notice that the screen position of [EMAIL PROTECTED] is different although "xvidtune -show" show the same settings and the same monitor has been choosen in Monitor setup (I don't have a modeline in XF86Config-4, it's just one automatically made by X with the DDC Monitor info). In ML9.1 (XFree4.3) the screen is horizontally a bit too large such that maximal opened windows loose part of their sides. I can correct this in ML9.1 by using "xvidtune" but I don't like changing the default modeline and if I use the On Screen Display of my monitor to adjust the screen ML8.2 (XFree4.2) looks too small. Why don't the screens of ML8.2 and ML9.1 use the same monitor space as the settings as shown in "xvidtune -show" are the same? What other parameters play a role in here and is it possible to set the screens exactly the same? Thank you, vatbier Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
[expert] Why doesn't "cat" print content of a file to standard output if the content is only a number ?
ML9.1 here. When I type in a Konsole "cat /var/lock/subsys/dm" no output is generated (dm containing the pid number of process prefdm) to the Konsole. But when I type "kill `cat /var/lock/subsys/dm" the process is effectively killed. I found this command in /etc/rc.d/init.d/dm: " stop) gprintf "Stopping display manager: " if [ -r /var/lock/subsys/dm ]; then kill `cat /var/lock/subsys/dm` > /dev/null 2>&1 " It took me a while to figure out that "cat" wouldn't print the content of a file to the Konsole(standard output) if that content consist only of a number. If that number is followed by a space in that file "cat" shows "number " on the Konsole. But if I type "cat /var/lock/subsys/dm > test.txt" the number is copied to test.txt. And in the command "kill `cat /var/lock/subsys/dm" the number is correctly presented to the "kill" command. Why this behaviour for "cat" ? "info cat" gives me no clue. Is this a bug of "cat" or some sort of feature ? vatbier Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
[expert] ML9.1:Kwrite:unicode encoding breaks reading of old iso8859-15 text files
It was really annoying to find out that ML9.1:KDE3.1:Kwrite has as default unicode encoding (utf8) whereas in ML8.2 Kwrite saved files in iso8859-15 encoding. I had a iso8859-15 encoded file that I write in with Kwrite in ML9.1 and in ML8.2. The normal English characters are represented the same in iso8859-15 and in utf8, but my "European" characters like "é,è,ï,ë,ç" all became crippled by the utf8 encoding of Kwrite in ML9.1. Took me a while to correct that problem as the file is rather big. I read somewhere that KDE3.1 now uses unicode encoding. That's good info, but I'd like to learn about this kind of changes at the time of installing Mandrake 9.1. So please put a big warning sticker at Install time to warn (non-English) people that certain word processors now use unicode encoding without being clear about it. The only wordprocessor that works correctly is KWord that asks when you open a plain text file what kind of encoding it has. You can view iso8859-15 text files correctly by opening them in Kwrite and changing the display encoding with View:Set Encoding:iso8859-15 Also Kwrite has some annoying bugs: When you open a text file, by clicking in Konqueror it is opened in utf8 encoding and when you save it with the "Save" command it is saved in utf8 encoding. But if you use the "Save as" command, the saving defaults to koi8-r encoding (Russian) as it is the first entry in the character encoding listing in the "Save as" dialog. So remember to change this to utf8. Also when you have a text file opened by Kwrite and you want to open another iso8859-15 text file, using File:Open and selecting the iso8859-15 encoding in the "Open File" dialog, it is displayed in utf8 encoding. Only when you open a blank KWrite window from the menu you can use the "Open File" dialog and select the iso8859-15 encoding and it is correctly shown in iso8859-15. Otherwise you have to use View:Set Encoding:iso8859-15. vatbier __ Do you Yahoo!? Yahoo! SiteBuilder - Free, easy-to-use web site design software http://sitebuilder.yahoo.com Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
[expert] Re: Why doesn't "cat" print content of a file to standard output if the content is only a number ?
NO, the output of "cat /var/lock/subsys/dm" is written over by the prompt of my shell (dm has a number like 1114) [EMAIL PROTECTED] home]$ cat /var/lock/subsys/dm [EMAIL PROTECTED] home]$ My system is Mandrake Linux 9.1 with bash v2.05b Is it possible to change this bash behaviour so that the prompt doesn't overwrite output of a command? If I type this: [EMAIL PROTECTED] home]$ echo -n "This is an example of what happens" >me [EMAIL PROTECTED] home]$ cat me [EMAIL PROTECTED] home]$ of what happens then I have to press Enter to be able to write at my prompt. Also the reason why I thought this only happened with a number was that I tested it with text files created by KWrite. I just found out that (sometimes) if I replace a number with some other characters KWrite adds a newline character by itself. I then had the impression that it didn't happen with text characters. Thank you all for the quick responses, vatbier __ Do you Yahoo!? Yahoo! SiteBuilder - Free, easy-to-use web site design software http://sitebuilder.yahoo.com Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
[expert] Solved: Why doesn't "cat" print content of a file to standard output if the content is only a number ?
Solved: this has already been rapported in september 2002: http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=74383 : Bug 74383 - start of line is discarded when writing a prompt Red Hat applied in december a patch from Chet Ramey to bash-2.05b/lib/readline/display.c Mandrake currently has this patch applied only in their cooker: bash-2.05b-14mdk: Bug-Description: When running in a locale with multibyte characters, the readline display updater will use carriage returns when drawing the line, overwriting any partial output already on the screen and not terminated by a newline. Just found this info by accident. vatbier vatbier wrote: > NO, the output of "cat /var/lock/subsys/dm" is written over by the > prompt of my shell (dm has a number like 1114) > [EMAIL PROTECTED] home]$ cat /var/lock/subsys/dm > [EMAIL PROTECTED] home]$ > My system is Mandrake Linux 9.1 with bash v2.05b > Is it possible to change this bash behaviour so that the prompt > doesn't overwrite output of a command? > > If I type this: > [EMAIL PROTECTED] home]$ echo -n "This is an example of what happens" >me > [EMAIL PROTECTED] home]$ cat me > [EMAIL PROTECTED] home]$ of what happens > then I have to press Enter to be able to write at my prompt. > > Also the reason why I thought this only happened with a number > was that I tested it with text files created by KWrite. I just > found out that (sometimes) if I replace a number with some other > characters KWrite adds a newline character by itself. I then had > the impression that it didn't happen with text characters. > > Thank you all for the quick responses, > > vatbier Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
[expert] ML9.1: Why does MandrakeUpdate show a new kernel-source but not a new kernel?
MandrakeUpdate shows a new kernel-source and kernel-doc but to install the new kernel you have to use RpmDrake:Install Software. MandrakeUpdate replaces the old kernel-source and -doc with the new ones, RpmDrake must be used for the kernel itself to install it next to the old one. I thought that these three (kernel,-source,-doc) must be installed together. If you want to update everything with MandrakeUpdate, you have to know that if you also select the new kernel-source and -doc you must run rpmdrake to install the new kernel. Is this info documented somewhere for newbies? How would they know that they have to install the new kernel with rpmdrake or rpm? Because I only find these old links: http://www.mandrakesecure.net/en/kernelupdate.php and http://www.mandrakesecure.net/en/docs/magic.php vatbier __ Do you Yahoo!? The New Yahoo! Shopping - with improved product search http://shopping.yahoo.com Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
[expert] ML9.1: SysV-Init Editor crashes with KUniqueApplication: DCOP communication error
Mandrake Linux 9.1: KDE3.1: kdeadmin-3.1-8mdk: ksysv An editor for System V startup schemes: SysV-Init Editor SysV-Init Editor crashes everytime when I want to start it up. When I start it in Konsole i get this error message ERROR: KUniqueApplication: DCOP communication error! I remember having this problem also in ML8.2, but then some day SysV-Init Editor in ML8.2 started working and still works without problem in ML8.2. But in ML9.1 the problem is persistent. I searched with Google but have no solution found, this bug has also been reported at http://bugs.kde.org/show_bug.cgi?id=37123 but no solution yet. In Mandrake Linux drakxservices is used to set the services, but I liked the program SysV-Init Editor to compare the different runlevels. vatbier Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
[expert] ML9.1: Why did Mandrake put restrictions on using KDE as root? (Konqueror timeout errors)
Sometimes I log in into KDE as root. When I, being logged in as root, use Konqueror to surf on the internet, I notice a lot of timeout errors (trying do to a query search in Mandrakesoft Bugzilla always timed out). In KCC:Network:Preferences I saw that timeout settings of 2 seconds (instead of the default 15,10,20,600 seconds) were saved into /root/.kde/share/config/kioslaverc That kioslaverc file for root came out of directory /usr/share/mdk/kde/root-interface/ . I guess that root-interface directory is full of KDE settings specifically for root. Why are those timeout settings set to 2 seconds for root in KDE, for security reasons or to hinder people at staying too long logged in as root in KDE? Are there any other of those annoying settings that prevent easy use of logging in as root into KDE? And to refresh my memory, why is it a bad idea to login as root in KDE? These days I do a lot of root changes,it's easier for me to stay for a while in KDE as root than to type my root password everytime I want to do some root job as a regular user. vatbier __ Do you Yahoo!? The New Yahoo! Shopping - with improved product search http://shopping.yahoo.com Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com