Re: your own filesystem
On Sun, Sep 02, 2012 at 08:27:44PM +0400, s...@bestmx.ru wrote: is there any detailed manual on nullfs and/or stackable filesystems... with function specifications, flowcharts etc... This paper could be useful: Vnodes: An Architecture for Multiple File System Types in Sun UNIX S.R. Kleiman Sun Microsystems Published around 1986 if I'm not mistaken. -- Francois Tigeot
Re: Errors on SSD
Hi, On Sat, Aug 25, 2012 at 11:03:42AM +0200, Sven Gaerner wrote: I updated my system to the latest release version. The uname says: 3.0-RELEASE DragonFly v3.0.3.1.g2c987-RELEASE. I guess it is not related to the kernel version, but I'm getting these messages on the console. I didn't see them before. ahci0.2: TFES slot 21 ci_saved = 0020 ahci0.2: disk_rw: error [...] That device is an SSD. ahci0.2: Found DISK OCZ-VERTEX2 1.29 serial=OCZ-CU25VMZ6117F3NFM These messages report disk errors, your drive is dead or dying. OCZ has probably the worst failure rate of all SSD brands; I'll avoid their products like the plague. -- Francois Tigeot
Re: Errors on SSD
On Sun, Aug 26, 2012 at 02:47:54PM +0200, Sven Gaerner wrote: Thanks. Then my assumption was ok. The disk is dying, so I'll have some time left to move my data on a magnetic disk and will see how long the SSD will work as swapcache. Do you have a recommendation for SSDs? Intel and Crucial models are fine; some dragonflybsd.org servers use Crucial M4 SSDs for swapcache. Some other brands may be ok too; having a look at reviews on e.g. newegg.com should give you a good idea wrt current models. This web page has some return rate statistics: http://www.behardware.com/articles/862-7/components-returns-rates-6.html (OCZ numbers are exceptional *ahem*) -- Francois Tigeot
Re: Fwd: df64 pkgsrc 2012Q2 DragonFly 3.0/x86_64 2012-07-24 11:35
Hi John, On Fri, Aug 10, 2012 at 10:47:00AM +0200, John Marino wrote: It shouldn't be a surprise to you. I informed you a month ago that libreoffice *will* fail all bulk build attempts on every platform, without a doubt. It simply will not build in a clean environment. I remember, and I started investigating it; unfortunately the font handling code is convoluted mess which should probably be entirely rewritten. I ended up disabling the unit test after this last report. Cheers, -- Francois Tigeot
Re: Fwd: df64 pkgsrc 2012Q2 DragonFly 3.0/x86_64 2012-07-24 11:35
On Fri, Aug 10, 2012 at 12:53:17PM +0200, John Marino wrote: Francois, do you have a patch for misc/libreoffice that I can commit that disables font handling? Disabling font handling is unthinkable! That would render the whole program useless... The error message is bogus, and caused by an unit test run at the end of the build. I've pushed one patch to wip/libreoffice to disable this particular test, patch-sw_Module_sw.mk I haven't been able to reproduce the No fonts could be found on the system error myself, I'm not sure if you'll be able to get a complete build with it. -- Francois Tigeot
Re: Fwd: df64 pkgsrc 2012Q2 DragonFly 3.0/x86_64 2012-07-24 11:35
On Fri, Aug 10, 2012 at 01:21:05PM +0200, John Marino wrote: On 8/10/2012 13:11, Francois Tigeot wrote: On Fri, Aug 10, 2012 at 12:53:17PM +0200, John Marino wrote: Disabling font handling is unthinkable! That would render the whole program useless... The error message is bogus, and caused by an unit test run at the end of the build. I've pushed one patch to wip/libreoffice to disable this particular test, patch-sw_Module_sw.mk It was a typo, I mean the font unit testing. Can I use patch-sw_Module_sw.mk in misc/openoffice without modification? Fixing wip doesn't fix libreoffice that is already in pkgsrc. To only way to reproduce this is build libreoffice in Tinderbox-DragonFly or in pbulk... Go ahead, there is nothing version-specific in it. I'm filling a bug report in the freedesktop bugzilla. Strangely, there is no registered bug even though many people complained on various mailing-lists about this No fonts could be found on the system issue. -- Francois Tigeot
Re: Fwd: df64 pkgsrc 2012Q2 DragonFly 3.0/x86_64 2012-07-24 11:35
On Fri, Aug 10, 2012 at 01:30:34PM +0200, Francois Tigeot wrote: I'm filling a bug report in the freedesktop bugzilla. Strangely, there is no registered bug even though many people complained on various mailing-lists about this No fonts could be found on the system issue. Link to the bug report: https://bugs.freedesktop.org/show_bug.cgi?id=53338 -- Francois Tigeot
Re: Fwd: df64 pkgsrc 2012Q2 DragonFly 3.0/x86_64 2012-07-24 11:35
On Fri, Aug 10, 2012 at 01:49:20PM +0200, John Marino wrote: On 8/7/2012 22:04, Francois Tigeot wrote: On Tue, Aug 07, 2012 at 10:08:37AM -0400, Justin Sherrill wrote: Package Breaks Maintainer - misc/libreoffice ftig...@wolfpond.org I just checked, LibreOffice from pkgsrc-2012Q2 builds perfectly. I'm curious as to what the issue was. By the way, there's already a patch with that name in libreoffice -- it was added by NetBSD to disable that test and two more, but only for NetBSD. Yes, I did disable these few tests permanently upstream. NetBSD is a fragile platform from LO's point of view. It might be worthwhile just pulling these three tests for all platforms. Anyway, FYI. I'll replace the existing page with the wip patch for the my testing purposes. I prefer to have as little differences with what upstream do as possible; even if the situation is better now, LO is still prone to unintentional breakage and these unit tests are a great help for detecting issues quickly. Wiz@ will probably update the rest of misc/libreoffice to the pkgsrc version in a few days if you don't do it first. -- Francois Tigeot
Re: Fwd: df64 pkgsrc 2012Q2 DragonFly 3.0/x86_64 2012-07-24 11:35
On Tue, Aug 07, 2012 at 10:08:37AM -0400, Justin Sherrill wrote: pkgsrc bulk build report Build failures Package Breaks Maintainer - misc/libreoffice ftig...@wolfpond.org I just checked, LibreOffice from pkgsrc-2012Q2 builds perfectly. I'm curious as to what the issue was. -- Francois Tigeot
Re: Motherboard advice
Hi, On Mon, Jul 30, 2012 at 05:37:41AM +, Jelle Hermsen wrote: I'm going to build a new workstation and since I'll only be running Dragonfly I want to take this opportunity to get it right. I've been using laptops for years and I must admit I can't quite see the forest for the trees here. I know that I want to have ECC memory and I would love to have some kind of graphics chipset that just works without too much hassle (since I'll mostly be running a terminal in X). Have a look at the supported hardware page: http://www.dragonflybsd.org/docs/supportedhardware/ For ECC support, you'll probably need to use a server board and add discrete sound and graphics cards. These two Supermicro boards are the only ones with ECC support + sound onboard I know of: http://www.supermicro.nl/products/motherboard/Xeon/C600/X9SRA.cfm http://www.supermicro.nl/products/motherboard/Xeon/C600/X9DAi.cfm There is also one Asus workstation board which *may* support ECC. -- Francois Tigeot
Re: seamonkey 2.10 from 2012Q2 pkgsrc build fails
On Fri, Jul 20, 2012 at 08:49:44AM +0200, John Marino wrote: On 7/20/2012 06:23, Edward M wrote: trying to build seamonkey 2.10 from pkgsrc2012Q2 however, it fails with the following error: pkg_create: lstat failed for file lib/seamonkey/extensions/inpec...@mozilla.org/chrome/ icons/default/winInspectorMain.xpm: no such file or directory Error code 2 [...] anyway, in general, Seamonkey builds on DragonFly, it just happened to be in a bad state for Q2. You can always switch to the trunk branch. Seamonkey-2.11 builds fine on DragonFly, this is purely a packaging error. It's possible to use touch(1) to create the missing files and restart the installation without rebuilding from scratch. Alternatively, removing winInspectorMain.xpm and the other problematic file (winInspectorMain16.xpm ?) from PLIST and recreating the checksums with bmake mdi will fix the packaging list and allow seamonkey to be built and installed from scratch. For all I know this mistake has been fixed in pkgsrc -current. -- Francois Tigeot
Re: pkgsrc current DragonFly 3.1/x86_64 2012-05-06 01:25
Hi, On Tue, May 08, 2012 at 11:24:18PM -0400, Justin Sherrill wrote: Build failures Package Breaks Maintainer - devel/opengrok pkgsrc-us...@netbsd.org Hopelessly out of date. The current version of OpenGrok works fine on DragonFly. net/wminet c...@core.de I have opened a PR (with fix) for this one: http://gnats.netbsd.org/cgi-bin/query-pr-single.pl?number=46318 It was one month ago and it has been assigned to the non-responsive dfly-pkg-people@ The root cause is this change in net/if_var.h: http://gitweb.dragonflybsd.org/dragonfly.git/commitdiff/1afbcbacb38bccc3bac01ae24420658b4b8346c7 Previously public structures are now only shown to userland if _KERNEL_STRUCTURES is defined I wouldn't be surprised if more packages were broken by this commit. -- Francois Tigeot
Re: pppoe
On Sat, Mar 24, 2012 at 10:03:19AM +0400, Andrey N. Oktyabrski wrote: On 24.03.12 09:06, Andrey N. Oktyabrski wrote: PF NAT do not work with tun0 This is not correct: it works but only day or two. This is normal if your ppp connection does not come with a static IP address. You should use a rule only specifying the interface name (tun0) in that case; this one should do the trick: nat on $ext_if inet from !($ext_if) - ($ext_if:0) The 'inet' keyword prevents pf from nating IPv6 traffic. -- Francois Tigeot
Re: Install DragonFlyBSD on 48 MB RAM
On Sat, Mar 03, 2012 at 01:33:35PM -0500, Justin Sherrill wrote: On Sat, Mar 3, 2012 at 9:07 AM, v...@ukr.net wrote: And also, somebody pointed out in this thread that the .tar archive with the pkgsrc tree may not align well with my current DragonFlyBSD version, or have I misunderstood something? Anyway, where can I download the archive with the pkgsrc tree? ftp://ftp.netbsd.org/pub/pkgsrc/pkgsrc-2011Q4/pkgsrc.tar.gz [...] There is no pkgsrc tarball that goes with 2011Q4, that I know of, which would solve this problem. Justin, the tarball you have linked to _is_ 2011Q4. It may also be downloaded from this url to get a more explicit name: ftp://ftp.netbsd.org/pub/pkgsrc/pkgsrc-2011Q4/pkgsrc-2011Q4.tar.gz -- Francois Tigeot
Re: Is there a size limit of natacontrol?
Hi, On Mon, Feb 27, 2012 at 12:28:00PM +, Alex Hornung wrote: On 27/02/12 12:20, Zenny wrote: I tried to create a RAID10 with natacontrol with 4 2TB HDDs, but it only shows 2TB (1718306MB) size of ar0 created instead of 4TB. In principle there is no such limit that I'm aware of, apart from the MBR partition size limit. It all depends where you are seeing those 2 TB. If it's a partition size, try using GPT. Nataraid itself doesn't seem to be limited to 2TB RAID volumes but some of the on-disk metadata format it uses are. By default, when no BIOS-created metadata is recognized, natacontrol creates a volume using the Promise metadata format -- and this one has an inherent limit of 32-bit disk sectors. The Intel MatrixRAID format moves this limit to 64-bit sectors, way beyond the 2TB barrier. The attached patch changes nataraid to use it by default instead of the Promise format. It _could_ allow the creation of 2TB RAID volumes on unrecognized controllers. -- Francois Tigeot diff --git a/sys/dev/disk/nata/ata-raid.c b/sys/dev/disk/nata/ata-raid.c index 8438614..505a912 100644 --- a/sys/dev/disk/nata/ata-raid.c +++ b/sys/dev/disk/nata/ata-raid.c @@ -1105,11 +1105,11 @@ ata_raid_create(struct ata_ioc_raid_config *config) * metadata format from the disks (if we support it). */ kprintf(WARNING!! - not able to determine metadata format\n - WARNING!! - Using FreeBSD PseudoRAID metadata\n + WARNING!! - Using Intel PseudoRAID metadata\n If that is not what you want, use the BIOS to create the array\n); - ctlr = AR_F_FREEBSD_RAID; - rdp-disks[disk].sectors = PROMISE_LBA(rdp-disks[disk].dev); + ctlr = AR_F_INTEL_RAID; + rdp-disks[disk].sectors = INTEL_LBA(rdp-disks[disk].dev); break; }
Re: Hammer Core Dumped on DragonFly v2.13.0.154.g481b38-DEVELOPMENT
Hi, On Mon, Nov 14, 2011 at 01:17:24PM +0530, Siju George wrote: After praying in Tongues I rebooted again :-) Data is back safe! Is there something special about the re boot just after a core dump? The dumped core is usually read from the swap area and put in a file; besides that boot proceeds normally. Just wondering why this didn't come up straight the first time [...] earlier it was [...] # undo -i 2011-11-11-all.sql.bz2 2011-11-11-all.sql.bz2: ITERATE ENTIRE HISTORY: Inappropriate ioctl for device [...] Could some one give the explanation? The first problem was caused by a hardware failure, the WRITE_DMA48 lines in your first mail: ad8: FAILURE - WRITE_DMA48 status=51READY,DSC,ERROR error=4ABORTED LBA=890156672 HAMMER(): Critical error inode=-1 error=5 while flushing meta-data HAMMER(): Forcing read-only mode ad8: FAILURE - WRITE_DMA48 status=51READY,DSC,ERROR error=4ABORTED LBA=890935168 There's not much Hammer can do in this situation; the strange behavior you saw is consistent with a drive failing to respond and then becoming available again. I would check the power supply and the drive itself, and be sure to have the data backuped elsewhere if I was in your shoes. -- Francois Tigeot
Re: eval: fsck: Exec format error
Siju, On Tue, Nov 15, 2011 at 03:59:55PM +0530, Siju George wrote: I was upgrading the server to new src and got this error. [...] : write failed, filesystem is full and the system paniced i did a 'call dumpsys' and rebooted but it gave this error. eval:fsck:Exec format error and I dropped into the shell prompt. None of the errors you posted recently make sense by themselves but there's a pattern -- your hardware is dying I had a look at the core dumps you put on leaf. In both cases, the root cause of the panics was a failure of the ad8 hard drive to execute a command. Sadly, the last two errors are consistent with data corruption. If you value your data, you should get it off this machine now. With luck, only the hard drive is failing but a bad power supply could also cause these sorts of errors and slowly kill most of the components. -- Francois Tigeot
Re: Small Hammer presentation
On Wed, Oct 26, 2011 at 04:59:45PM -0400, Pierre Abbat wrote: On Wednesday 26 October 2011 12:29:19 Francois Tigeot wrote: I recently gave a small talk on Hammer to a sysadmin audience. The slides can be downloaded from here: http://www.wolfpond.org/ Besides the original version (in French), I have provided an English translation. Looks good! Though English words in French like crashs look funny. I'd say écraser for a head crash, but what's the proper way to say that a program crashes? Thanks :-) Crash may come from English but I've always heard it used and it feels natural, be it for planes, programs or disks. Planter can also be used for programs. I've never heard écraser used as a synonym of crash; in a computer context it would mean to overwrite something. Apparently, it is used by Québécois to avoid saying crasher. So says Wikipedia: http://fr.wikipedia.org/wiki/Crash_(aviation) -- Francois Tigeot
Re: Streamline pkgsrc issues: DragonFly developer gained NetBSD commit privilege
On Sun, Sep 11, 2011 at 09:21:58PM -0400, Justin Sherrill wrote: From what I've seen in the Problem Report system, there's a 'dfly-pkg-people' alias that DragonFly issues get placed with; if you are in that group, you'll probably catch things directly. The pkgsrc people are usually very quick to act. The only exception I've found is this dfly-pkg-people@ alias Once a PR is assigned to it, you can abandon all hope it will be fixed. It would be best to remove this alias IMHO. -- Francois Tigeot
Re: radvd
On Wed, Aug 17, 2011 at 05:01:17PM -0400, Justin Sherrill wrote: Would it be worth setting this by default, since (someday, somehow) IPv6 is becoming more common? On Wed, Aug 17, 2011 at 4:07 PM, Matthias Rampke matth...@rampke.de wrote: DragonFly doesn't accept router advertisements by default. You have to set something like ip6mode=autohost in /etc/rc.conf and some sysctl (ends in accept_rtadv). I'd hope not; automatic host configuration is not enabled by default with IPv4 either. I'm actively using IPv6 but my machines are configured with static addresses. And then, there's DHCPv6; if some sort of automatic autoconfiguration is to be enabled, it would have my preference. Router advertisements only provide the bare minimum IP+router information; DNS server data may not even be present. Do not even thing of using them to boot or configure hosts beyond that... Oh, and since RA technology is stateless, that means there's no log to consult if you want to know which IP was assigned to which ethernet port later... -- Francois Tigeot
Re: rebuilding pkg_install fails
Hi Pierre, On Thu, Aug 11, 2011 at 01:15:56AM -0400, Pierre Abbat wrote: On Thursday 11 August 2011 00:06:22 Justin Sherrill wrote: I've built the pkgsrc bootstrap for 2011Q2 several times over recently, and I didn't encounter this. I assume there's something out of place on your machine that's confusing the build. Maybe try doing a 'bmake clean'? That suggestion sounds kinda weak now that I said it. Tried that. Didn't fix it. I don't know what exactly is wrong on your system, but so far I've reliably managed to cure all my pkgsrc issues by nuking the existing installation and rebuilding it from scratch: 0. update your pkgsrc sources be sure to use a quaterly release to avoid unnecessary trouble 1. move away your old pkgsrc installation mv /usr/pkg/usr/pkg.old mv /var/db/pkg /var/db/pkg.old mv /var/db/pkg.refcount /var/db/pkg.refcount.old 2. reuse your existing configuration files mkdir /usr/pkg mv /usr/pkg.old/etc /usr/pkg 3. bootstrap pkgsrc from scratch cd /usr/pkgsrc/bootstrap ./bootstrap ./cleanup 4. reinstall your programs you may want to start with your favorite shell and add basic utilities such as devel/scmgit-base and such before trying to tackle more complicated software You can always go back to the old pkgsrc installation preserved in /usr/pkg.old -- Francois Tigeot
Re: arpresolve: can't allocate llinfo for 127.0.0.1 rt
On Thu, Aug 11, 2011 at 03:54:45PM +0200, Ralf Schmitt wrote: while removing the network cable from the machine, I now even got a kernel panic and debugger prompt. Are there any docs on how to report those? The error is something like (hand transcripted): , | panic: rtrequest1_msghandler: rtrequest table error was not on cpy #0 | cpuid = 4 What is the exact version of your system (uname -a) ? This is suspiciously close to a problem I had in the last few months. -- Francois Tigeot
Running OpenGrok on DragonFly
Hi, My motivation for making the JDK 1.6 work on DragonFly was to run OpenGrok. OpenGrok is a really nice source code browser. Want to see the definition of a function or where it is used ? Just type its name in a search box and click on the results. It is initially the work of a Sun employee, and was heavily used with the OpenSolaris project. Home page: http://hub.opensolaris.org/bin/view/Project+opengrok/WebHome Unfortunately, OpenGrok is written in Java, and is not as easy to run as more traditional programs. I have compiled a few instructions based on what I did to get it to browse the DragonFly source tree: 0. Use DragonFly-2.10 I have had some reports Java crashes on DragonFly 2.11 at the moment. 2.11 is under heavy development and this is not so surprising, but I better say it loud and clear before someone loses too much time trying 1. build lang/kaffe with gcc-4.1 and install it cd /usr/pkgsrc/lang/kaffe CCVER=gcc41 bmake install lang/kaffe is miscompiled by gcc-4.4 and cannot bootstrap the needed JDKs 2. install wip/jdk16 pkgsrc-wip is an additional set of packages which are not in pkgsrc. Get a tarball here and extract it in your pkgsrc tree: http://pkgsrc-wip.sourceforge.net/ You'll then get a new wip/ subdirectory 3. Install Tomcat from www/apache-tomcat6 4. Download the OpenGrok binary tarball I used the 0.10 version available here: http://hub.opensolaris.org/bin/view/Project+opengrok/files Extract it somewhere safe, like in /usr/local 5. Edit the file named bin/OpenGrok Add this at the top: OPENGROK_TOMCAT_BASE=/usr/pkg/share/tomcat And change some other variables: OPENGROK_INSTANCE_BASE=/var/opengrok SRC_ROOT=/usr/src DATA_ROOT=/var/tmp/opengrok_data EXUBERANT_CTAGS=/usr/pkg/bin/exctags JAVA_HOME=/usr/pkg/java/jdk-1.6.0 GIT=/usr/pkg/bin/git 6. Deploy the web application in Tomcat ./bin/OpenGrok deploy 7. Create the index ./bin/OpenGrok index This will take a long time and eat much memory. A x86_64 system with 2GB memory kept swapping during the operation and the web interface was nearly unusable At this stage, OpenGrok is operational; it just won't find results for the parts of the code which are not yet indexed. 8. Point your web browser to http://the.opengrok.machine.address:8080/source That's all folks! -- Francois Tigeot
Java 1.6 now working on DragonFly
Hi, I have fixed the wip/jdk16 package in pkgsrc-wip It is now possible to build it on DragonFly, for both i386 and x86_64 architectures. The pkgsrc-wip package tree can be downloaded from this url: http://pkgsrc-wip.sourceforge.net/ Like all JDKs, wip/jdk16 requires an existing installation of Java to be built. This ultimately fails down to lang/kaffe, which has a known issue with gcc-4.4. Building it with this version of gcc will result in a broken package There is no problem with gcc-4.1, so setting CCVER=gcc41 to build Kaffe should be enough to get things started. -- Francois Tigeot
Re: SATA drive problem
On Wed, Jun 29, 2011 at 10:22:24AM -0700, Tim Darby wrote: I tried to move a big folder from one SATA drive to another and the machine stopped responding to the point that it lost network connectivity (DF 2.10.1). After reboot, I found the following errors: [...] kernel: (da4:ahci0:4:0:0): READ(10). CDB: 28 0 80 3a d6 48 0 0 80 0 kernel: (da4:ahci0:4:0:0): CAM Status: SCSI Status Error kernel: (da4:ahci0:4:0:0): SCSI Status: Check Condition kernel: (da4:ahci0:4:0:0): MEDIUM ERROR asc:0,0 [...] I'm guessing this is a drive problem, not a system problem, but just wanted to check with the experts first. I haven't had a chance to run a drive diagnostic yet. I did try a repeat of that same mv command and it happened again. The drive is just a month old, FWIW. I'll say some specific parts of the magnetic platters of your drive are damaged and that's why you get this error every time trying to access the same files. A drive has not to be old to fail; failure during the first weeks of activity is relatively common. It may have been badly packaged or handled during transport or may have been badly manufactured ... -- Francois Tigeot
Re: DragonflyBSD on Areca w/ HAMMER
Hi, On Tue, May 31, 2011 at 03:07:25PM +1000, Dean Hamstead wrote: Im thinking about reloading my home file server with dragonflybsd. Mainly so i can take advantage of the rather awesome looking HAMMER filesystem. :-) My home server has an Areca 1260 SAS raid card, which is reported as being supported by the arcmsr man page. It should work fine, like most Areca boards. I have to point out it is not a SAS but only a SATA model, tho. Do not expect to be able to plug it into a SAS expander, only direct-attachment is possible AFAIK. Im looking for confirmation from other users that areca cards work well, I've done some extensive testing of different RAID cards one month ago. All Areca models worked out of the box. and that the admin tools are supported from freebsd or otherwise implemented independently. This is something I have yet to check. Howewer, the ARC-1260 has an ethernet port and you can directly manage it from the network, without the need to install tools on your main operating system. Also id like to get some feedback that HAMMER is up to the task of being belted really hard over sustained periods. HAMMER has been production-ready for years now, there shouldn't be any reliability issues. Im also not able to see find a conclusive answer to if HAMMER can be expanded (as im relatively frequently adding more disks, and expanding my raid array - currently im using UFS tool growfs) Someone else will have to answer this question, I've not tried to expand a HAMMER volume yet. Best, -- Francois Tigeot
Re: ad8: FAILURE - READ_DMA48 status=51READY,DSC,ERROR error=10NID_NOT_FOUND LBA=809594688
On Mon, May 30, 2011 at 10:54:26AM +0530, Siju George wrote: On Sun, May 29, 2011 at 12:54 PM, Francois Tigeot ftig...@wolfpond.org wrote: Thanks a lot Francois for the reply :-) You're welcome. This error appears only some times. I refitted the cables of all hard disks once again and this error does not seem to occor for some time. It may be purely a cable issue then. Do the cables feel somewhat loose in the connectors ? The first SATA cable were too easy to pull out; nowadays the plastic part of the connectors has been modified and one really needs to apply some force to pull out a cable. Some modern cable models also use metal latches for additional security. What should I do to find out if the disk is working right? Is a # smartctl -t long /dev/ad8 enough? Time will time. Even though smart should give you a good indication on the state of the disk, it is not 100% perfect. In your place, I would start to become paranoid and do some backups. Also will this type of error affect the Integrity of hammer File System? These errors meant some part of the disk could not be read. If you get them in the wrong places, your hammer filesystem could well become unusable, yes. Should I do some thing to Check the integrity of the file system ? It would be best. Some hammer(8) commands seem to be usable for that purpose. I didn't see a fsck equivalent in the man page, tho. I plan to upgrade to hammer v6 soon. It's stable and doesn't even take 1 minute to do. No reason to delay here ;) -- Francois Tigeot
Re: ad8: FAILURE - READ_DMA48 status=51READY,DSC,ERROR error=10NID_NOT_FOUND LBA=809594688
Hi, On Thu, May 26, 2011 at 03:45:59PM +0530, Siju George wrote: I am getting plenty of the following error. My Kernel Version is DragonFly v2.11.0.247.gda17d9-DEVELOPMENT #36: Mon May 23 12:57:37 IST 2011 GENERIC What could be the trouble? ad8: TIMEOUT - READ_DMA48 retrying (1 retry left) LBA=809594688 ad8: FAILURE - READ_DMA48 status=51READY,DSC,ERROR error=10NID_NOT_FOUND LBA=809594688 Your hard-disk may be dying or you may have a problem with the cable/the controller. What sort of disk is this ? If it's an old IDE model, is your ribbon cable a 80-wire one ? It may be worth it to change it. You may also want to reduce DMA speed as a workaround (you should be able to do it from the BIOS). -- Francois Tigeot
Re: LibreOffice package created
On Tue, May 24, 2011 at 06:30:57AM -0400, Pierre Abbat wrote: The anonymous checkout succeeded. Why is it weird that I did an anonymous checkout? The weird thing is that it failed, not that it was anonymous. Some time-limited connectivity error perhaps ? -- Francois Tigeot
Re: LibreOffice package created
On Tue, May 24, 2011 at 08:43:41AM -0400, Pierre Abbat wrote: On Tuesday 24 May 2011 07:55:48 Francois Tigeot wrote: On Tue, May 24, 2011 at 06:30:57AM -0400, Pierre Abbat wrote: The anonymous checkout succeeded. Why is it weird that I did an anonymous checkout? The weird thing is that it failed, not that it was anonymous. The *onymous* checkout failed, not the *anonymous* checkout. It gave me a permission error or something like that. If you tried to checkout pkgsrc-wip with a Sourceforge account, this isn't too surprising. Accounts have to be approved to be allowed to commit to pkgsrc-wip. I guess the project master has not taken the checkout case into account and denied everything by default. -- Francois Tigeot
LibreOffice package created
Hi all, During the last few months, I have been working on porting LibreOffice to DragonFly. It now works quite well, and I have just created a package in pkgsrc-wip [1] It builds and installs fine on DragonFly 2.10/x86_64; it has not yet been tested on i386. The software prerequisites are pkgsrc-2011Q1 and math/lp_solve from pkgsrc -current. The package takes a long time to build, between 3 hours and 3 hours and a half on a fast Core 2 Duo 3GHz machine. 1: pkgsrc-wip is a repository for work in progress packages. The instructions to install it on your machine are here: http://pkgsrc-wip.sourceforge.net/ -- Francois Tigeot
Re: LibreOffice package created
On Thu, May 19, 2011 at 08:40:49AM -0400, Pierre Abbat wrote: On Thursday 19 May 2011 02:56:16 Francois Tigeot wrote: 1: pkgsrc-wip is a repository for work in progress packages. The instructions to install it on your machine are here: http://pkgsrc-wip.sourceforge.net/ I tried to check it out with my SourceForge account but got Permission denied. What's wrong? Strange. There's a list of SourceForge accounts allowed to commit to pkgsrc-wip. If yours is not on it, maybe you can't even checkout the files... Can you try a tarball snapshot or an anonymous checkout ? -- Francois Tigeot
Re: LibreOffice package created
On Thu, May 19, 2011 at 09:49:22AM -0400, Pierre Abbat wrote: On Thursday 19 May 2011 09:10:51 Francois Tigeot wrote: On Thu, May 19, 2011 at 08:40:49AM -0400, Pierre Abbat wrote: On Thursday 19 May 2011 02:56:16 Francois Tigeot wrote: http://pkgsrc-wip.sourceforge.net/ I tried to check it out with my SourceForge account but got Permission denied. What's wrong? I did an anonymous checkout. Weird. I've put up a snapshot of pkgsrc-wip with my latest changes here: http://dl.wolfpond.org/pkgsrc-wip-20110519-snapshot.tar.bz2 and there's a binary package of LibreOffice for i386 here: http://dl.wolfpond.org/packages.dfly-i386/libreoffice-20110517.tgz A 64-bit package is beeing built at the moment. If there's enough interest, I will also put it up for download. -- Francois Tigeot
Re: pkgsrc reports
On Wed, May 11, 2011 at 07:56:29AM +0100, Alex Hornung wrote: It seems that to get mono building again we just need a simple patch (upstream and pkgsrc ideally :)) to mono/utils/mono-sigcontext.h, as well as undefining the defined(UCONTEXT_GREGS) (config? Makefile?) option. I'd appreciate it if someone could give it a shot, it should be fairly straight forward. I've opened a PR with patch a month ago: http://gnats.netbsd.org/cgi-bin/query-pr-single.pl?number=44846 It has been assigned to dfly-pkg-people@ and never been committed afaik. -- Francois Tigeot
Re: Tests of RAID adapters
On Mon, Apr 25, 2011 at 07:38:57PM +0200, Sascha Wildner wrote: On 4/25/2011 9:12, Francois Tigeot wrote: LSI SAS 3081E-R --- The RAID1 volume created in the BIOS of the card was visible but there were some timeout error messages from the start: I couldn't install DragonFly, newfs_hammer hung at 90% completion. I've put an update for mpt(4) here: http://leaf.dragonflybsd.org/~swildner/0001-mpt-4-Sync-with-FreeBSD.patch Thanks for updating the driver. With this new version, things are much better :) I've used a similar setup as before, with the following changes: - host system is running DragonFly 2.10 + updated mps driver - host system is a 4-core Xeon E5506 - the drives are 2x WD5003ABYX (500 GB, 7200 RPM, 64MB cache) Observations: 1. no timeouts on startup mpt0: LSILogic SAS/SATA Adapter port 0xe000-0xe0ff mem 0xfbbf-0xfbbf,0xfbbec000-0xfbbe irq 16 at device 0.0 on pci3 mpt0: MPI Version=1.5.20.0 mpt0: Capabilities: ( RAID-0 RAID-1E RAID-1 ) mpt0: 1 Active Volume (2 Max) mpt0: 2 Hidden Drive Members (14 Max) mpt0:vol0(mpt0:0:0): Settings ( Hot-Plug-Spares ) mpt0:vol0(mpt0:0:0): Using Spare Pool: 0 mpt0:vol0(mpt0:0:0): 2 Members: (mpt0:1:2:0): Primary Online (mpt0:1:1:0): Secondary Online mpt0:vol0(mpt0:0:0): RAID-1 - Optimal mpt0:vol0(mpt0:0:0): Status ( Enabled ) (mpt0:vol0:1): Physical (mpt0:0:1:0), Pass-thru (mpt0:1:0:0) (mpt0:vol0:1): Online (mpt0:vol0:0): Physical (mpt0:0:2:0), Pass-thru (mpt0:1:1:0) (mpt0:vol0:0): Online da0 at mpt0 bus 0 target 0 lun 0 da0: LSILOGIC Logical Volume 3000 Fixed Direct Access SCSI-2 device da0: 300.000MB/s transfers da0: 476837MB (976562176 512 byte sectors: 255H 63S/T 60788C) no B_DEVMAGIC (bootdev=0) 2. newfs_hammer completes succesfully, although performance is bad $ iostat da0 2 tty da0 cpu tin tout KB/t tps MB/s us ni sy in id 1 245 0.000 0.00 0 0 2 0 98 0 40 16.00 118 1.85 0 0 0 0 100 0 643 16.00 116 1.81 0 0 0 0 100 3. blogbench also runs without crashing $ blogbench -d /mnt/bench [...] Final score for writes: 1407 Final score for reads :244892 $ iostat da0 2 tty da0 cpu tin tout KB/t tps MB/s us ni sy in id 0 22 15.99 3045 47.55 11 0 89 0 0 0 11 15.86 865 13.39 11 0 89 0 0 0 11 15.98 1886 29.44 11 0 89 0 0 0 23 16.00 2974 46.46 10 0 90 0 1 For information, this is the blogbench score of a disk of the same model plugged on one of the AHCI SATA ports: Final score for writes: 1307 Final score for reads :215165 -- Francois Tigeot
Re: Tests of RAID adapters
On Wed, Apr 27, 2011 at 02:18:49PM +0200, Francois Tigeot wrote: On Mon, Apr 25, 2011 at 07:38:57PM +0200, Sascha Wildner wrote: On 4/25/2011 9:12, Francois Tigeot wrote: LSI SAS 3081E-R --- [...] I've put an update for mpt(4) here: http://leaf.dragonflybsd.org/~swildner/0001-mpt-4-Sync-with-FreeBSD.patch [...] 2. newfs_hammer completes succesfully, although performance is bad The bad performance was caused by disabled write-cache on the disks. MPT SAS controllers disable disk cache for writes by default and do not allow the user to enable it in their BIOS setup. The disks I used can read/write linearly at about 135 MB/s. On startup, linear read speed on the array is 133 MB/s, and write speed is an abysmal 13.50 MB/s (no, it's not a typo) Fortunately, it's possible to enable write-cache again with a sysctl: # sysctl -w hw.mpt0.vol_member_wce=On After typing that command, linear speeds became 136 MB/s for reads and 132 MB/s for writes. -- Francois Tigeot
Tests of RAID adapters
constantly seeking LSI SAS 9211-8i --- http://www.lsi.com/channel/products/hba/sas_sata_hbas/internal/lsisas92118i/ The card was not recognized at all. It seems the mps(4) driver needs to be updated to support this model. LSI is known to be developing a new version of the driver, which should be committed to the FreeBSD tree soon. 3Ware 9690SA-4i --- http://www.lsi.com/channel/products/raid_controllers/sata_sas/3ware_9690sa4i/ 3Ware has been recently bought by LSI, but the 3Ware branded cards are still different from the LSI ones and use different drivers. The card was recognized instantly and DragonFly could be installed without any problem dmesg extract: 3ware device driver for 9000 series storage controllers, version: 3.80.06.003 twa0: 3ware 9000 series Storage Controller twa0: INFO: (0x15: 0x1300): Controller details: Model 9690SA-4I, 128 ports, Firmware FH9X 4.08.00.006, BIOS BE9X 4.08.00.001 ... da0: AMCC 9690SA-4I DISK 4.08 Fixed Direct Access SCSI-5 device da0: 100 MB/s transfers da0: 152577MB (312477696 512 byte sectors: 255H 63S/T 19450C) A little while after boot, some informational messages appeared on console: twa0: WARNING: (0x04: 0x002F): verify not started; unit never initialized: unit=0 twa0: INFO: (0x04: 0x000C): Initialize started: unit=0 Adaptec AAR-1120SA -- http://www.adaptec.com/en-us/products/controllers/hardware/sata/entry/aar-1220sa/ This is a SATA card During startup, I could not enter directly the BIOS setup. After pressing CTRL+A, I got an error message saying not enough free memory to load the utility! At the end of BIOS initialization, the setup started correctly nevertheless Neither the card nor a disk volume was detected by DragonFly. This model may not have a driver. Adaptec 3405 http://www.adaptec.com/en-us/products/controllers/hardware/sas/value/sas-3405/ During startup, I pressed CTRL+A to enter its BIOS setup but at the end of the boot sequence all I got was an error message saying to hit Enter to force the Config Utility to load. It didn't load, and the machine froze. The controller was constantly beeping and I had to do a hard reset to get the mainboard to start normally again. I then plugged this card into a Supermicro X7SBL-LN2 mainboard. This time, I could enter the BIOS setup and create a RAID1 volume The controller was recognized by the aac(4) driver but was unfortunately unusable: aacd0: RAID 1 (Mirror) on aac0 aacd0: 152490MB (312299520 sectors) disk scheduler: set policy of aacd0 to noop CAM: Configuring 8 busses intr18 at 40001/4 hz, livelocked limit engaged! **WARNING** waiting for the following device to finish configuring: wpt: func=0x80277157 arg=0 [last two lignes repeated 4 times] Giving up, interrupt routing is probably hosed no B_DEVMAGIC (bootdev=0) aac0: COMMAND 0xffe035d902a0 (TYPE 601) TIMEOUT AFTER 137 SECONDS [three almost identical lines with slight command# and type# variations] At this point, the machine was deeply frozen and I had to do a hard reset to get it to reboot. It seems I'm not the only one to have trouble with Adaptec and Supermicro mainboards: http://www.webhostingtalk.com/archive/index.php/t-926239.html Conclusions --- I'm very impressed by Areca. All their cards worked flawlessly out of the box. The Areca people were very cooperative, they tested and reviewed the initial port of the arcmsr driver (from FreeBSD). Many of their adapters include an ethernet port which can be used for supervision, removing the need to install a special low-level utility in the host OS. I only had access to one 3Ware adapter, but it also worked out of the box and I have reports of different recent models also working flawlessly. There are some LSI models known to work flawlessly. Unfortunately, they seem to be old products which are not sold anymore. With all the troubles I had getting Adaptec cards to work reliably before DragonFly was even booted, I cannot seriously consider to purchase products of this brand. If you want a RAID adapter to use with DragonFly, Areca and 3Ware are the two best choices of the moment, my first choice beeing Areca. -- Francois Tigeot
Re: Fwd: pkgbox64 pkgsrc 2011Q1 DragonFly 2.10.0/x86_64 2011-04-12 03:54
On Sun, Apr 24, 2011 at 01:20:01AM -0400, Brian Gianforcaro wrote: I'm looking into lang/mono now, I know this has been a sort of on going project to get it to build on DragonFly. FWIW, I've recently opened a PR with some patches: http://gnats.netbsd.org/cgi-bin/query-pr-single.pl?number=44846 I don't think they have been committed yet. -- Francois Tigeot
Re: Hardware.
Hi, On Sat, Apr 23, 2011 at 07:00:33AM +1000, David Crosswell wrote: I've been checking out your hardware page here: http://tinyurl.com/3qbp9ck and wanting to know which of these supermicro opteron server boards work best with Dragonfly off the shelf. They all do. I've never had trouble with Supermicro boards. The only parts not working out-of-the box I know of are the included SAS adapters on certain models. You'll be better of buying a real RAID card if you want to go that route. I'm looking at building a small server to familiarise myself with all the BSDs, for study purposes, and although I like to play, I don't have a lot of leeway as far as cash goes to make any mistakes. The cheapest board/cpu combination is the X7SPA-H: http://www.supermicro.com/products/motherboard/ATOM/ICH9/X7SPA.cfm?typ=H The included Atom CPU is not very fast for compilation, so you may want to with this combo instead: X7SBL-LN2 + Core 2 Duo http://www.supermicro.com/products/motherboard/Xeon3000/3200/X7SBL-LN2.cfm I use both models in small servers, they're great. -- Francois Tigeot
Re: Fwd: pkgbox32 pkgsrc 2011Q1 DragonFly 2.10.0/i386 2011-04-20 02:49
On Wed, Apr 20, 2011 at 10:59:15PM -0400, Justin Sherrill wrote: DragonFly 2.10, i386, pkgsrc-2011Q1. I think there's a fix for rpm2pkg that is newer than the version of 2011Q1 I used. That's right, a fix has been committed on April 16. -- Francois Tigeot
Re: 2.10 Release schedule - Release will be April 23rd 2011
On Thu, Apr 21, 2011 at 11:28:58AM +0530, Siju George wrote: On Thu, Apr 7, 2011 at 7:39 AM, Matthew Dillon dil...@apollo.backplane.com wrote: I am following development branch should I wait till 23rd to try live dedup on my backup servers? The dedup code is solid and already used everyday on some machines. You can enable it already, it won't change in two days. I ask this because I have been terribly busy since December and have not updated the DragonFLY since Dec 6 I am on .9-DEVELOPMENT DragonFly v2.9.1.176.gc94587-DEVELOPMENT #30: Mon Dec 6 13:34:38 IST 2010 I don't think hammer deduplication was operational back then. Why not update to todays version of the 2.10 branch ? -- Francois Tigeot
Re: Dragonfly network changes - U-Verse almost a complete failure
On Sun, Feb 20, 2011 at 01:02:00PM -0800, Matthew Dillon wrote: Hahaha... ok, well, I spoke too soon. U-Verse is a piece of crap. [...] Sigh. You'd think ATT would be smart enough to do this properly, but after 5 years of trying they are still clueless about IP networks. Maybe in another year or two they will fix their stuff. Or not. It seems they're busy breaking it more instead: http://arstechnica.com/tech-policy/news/2011/03/is-atts-new-150gb-dsl-data-cap-justified.ars You now have a 250 GB data cap per month... -- Francois Tigeot
Re: who use dragonflybsd.
On Mon, Mar 07, 2011 at 02:18:55PM +0800, kevdmx wrote: who use dragonflybsd for product? Not sure about the for product part, but you can have a look at this thread from september: http://leaf.dragonflybsd.org/mailarchive/users/2010-09/index.html#00018 -- Francois Tigeot
Re: Dragonfly network changes
On Fri, Feb 18, 2011 at 02:29:05AM -0600, Chris Turner wrote: On 02/18/11 00:53, Francois Tigeot wrote: Do they offer IPv6 ? man gif(4) MUHUAHAHAHAA I don't know about the US, but I've add native v6 connectivity for about 8 years (forgot exactly when), so there's really no excuse these days. -- Francois Tigeot
Re: Dragonfly network changes
On Thu, Feb 17, 2011 at 07:18:18PM -0800, Matthew Dillon wrote: Various DragonFly machines are now running on a much faster network thanks to ATT u-verse, Great :-) For those interested this is ATT Business U-Verse. Downlink speed is around 16 MBits and uplink speed is around 2 MBits with their highest-grade service. My comcast cable internet (which I will be getting rid of soon), also the highest grade service, has a faster downlink speed of around 30 MBits, but around the same uplink speed of 2 MBits. Do they offer IPv6 ? The global IPv4 pool is now gone and APNIC, RIPE and ARIN pools will almost certainly be depleted in a few months too. Some people are deploying IPv6-only networks right now, and without IPv6 connectivity I'm afraid access to DragonFly resources could become troublesome in the future. -- Francois Tigeot
Re: Comments on pkgsrc and DragonFly
On Wed, Jan 12, 2011 at 08:19:47PM +0530, Siju George wrote: It would be great if there can be a porter's handbook. i would atleast like to make a try porting some stuff The pkgsrc developper's guide is your friend: http://www.netbsd.org/docs/pkgsrc/developers-guide.html -- Francois Tigeot
Re: Comments on pkgsrc and DragonFly
Hi, On Fri, Jan 07, 2011 at 11:39:54PM -0500, Stephane Russell wrote: Hi, thanks for your reply, which I read carefully. You're welcome. Logically, I think that if a system is not defining BSD, than it's simply not a BSD, it's a BSD fork at most. BSD has been dead since the nineties. Nothing to see here, move along. Yes, not that a problem, but some common tag would useful here too, when possible. Also, like you say, Darwin is considered a BSD, but this doesn't say much. It tells that BSD like code might compile. There is no BSD like code; it varies on a case-by-case basis. Part of the problem with porting is software checking for platform names whereas it should be looking for features. NetBSD's pkgsrc, FreeBSD's popularity and Linux's widespread innovations, might be the only things keeping some similarities between the BSD forks. Code/feature sharing is happening. I feel there is more similarity between recent *BSD and Linux systems than between old proprietary Unices from the 80s. So at most, BSD forks can only be used seriously as strong servers. That's how I'm using dfly. FUD. All my desktop systems have been running on FreeBSD or DragonFly for more than 10 years. Sometimes the lack of a good Microsoft Word alternative is a bit painful, but with LibreOffice now unleashed, there's a good chance this matter will be resolved relatively quickly. -- Francois Tigeot
Re: Comments on pkgsrc and DragonFly
On Fri, Jan 07, 2011 at 03:03:58PM -0500, Stephane Russell wrote: While porting programs to DragonFly, I had these issues (which are not bugs): - BSD is undefined in DragonFly, this isn't working: #if (defined(BSD) BSD = 199306) Never saw this one. All tests I encountered in third-party software were looking for full OS names or OS-specific defines like __FreeBSD__ or irix - In many autoconf scripts, BSD variants are grouped this way: case ${OSARCH} in *BSD) case $uname in *BSD*) both are excluding DragonFly, since uname -s return DragonFly and These are a pain, but I don't think they are so pervasive; Darwin is also considered as a BSD system and has the same problem here. In all cases I've seen, it was fixed with a new test checking for the full OS name: case ${OSARCH} in *BSD) blah + Darwin) + blah But it does mean that choosing FreeBSD as build type won't necessarly mean no changes required on the autotool scripts. - Some programs are compiling successfully by defining FreeBSD as build type. Is DragonFly kept close as possible to FreeBSD on purpose or should we expect this to be a vanishing legacy? This is a vanishing legacy. I had to distinguish between FreeBSD and DragonFly when porting the jdk In some cases, FreeBSD oriented code was failing and I had to use the same #define directives as NetBSD or Linux. Kde4 packages are also troublesome; if I remember correctly, they consider DragonFly as FreeBSD and fail at some stage during the compilation. Official positions here will me help me knowing what to expect while porting. When it's on a small scale, one-liner patches fit the bill and they can be safely sent upstream: - #ifdef __FreeBSD__ + #if defined(__FreeBSD__) || defined(__DragonFly__) In other cases, adding an abstraction function such as isLinuxOrBSD may help replace a bunch of #defines. Sometimes, the original code is really weird/non-portable and can be safely replaced by an equivalent construct which works equally well on all known operating systems. This is no official position, simply my experience so far. -- Francois Tigeot
Re: Info on clustering and HAMMER
On Thu, Dec 30, 2010 at 08:54:26AM -0500, Francisco Reyes wrote: I believe right now Hammer is able to replicate to a second machine in a master slave mode. However the only docs I have seen for data replication is http://www.dragonflybsd.org/docs/how_to_implement_hammer_pseudo_file_system_ _40___pfs___41___slave_mirroring_from_pfs_master/ Hopefully, others can add more info. hammer's manual page is more complete: http://leaf.dragonflybsd.org/cgi/web-man?command=hammersection=8 I'm using the mirror-stream feature to do a pseudo-real-time backup of a remote imap server. hammer uses ssh for the transport. You have to do some initial key setup to allow the backup machine to connect without trying to use passwords and then it's as simple as that: hammer -b 100k mirror-stream r...@bigserver:/home backup.bigserv-home This command synchronizes a local slave filesystem from the /home fs of bigserver. Since the backup machine is behind a small ADSL line, the bandwidth used is capped to 100 KB per second. The remote machine may crash and burn, I'll still have a good copy of its important data, up to the last few in-flight transactions. This stuff beats periodic backup schemes hands-down. -- Francois Tigeot
Re: Compiling with gcc -march ix86
On Fri, Dec 24, 2010 at 05:34:38PM -0500, Stephane Russell wrote: I'm actually trying to compile asterisk on DFBSD. It needs to compile with the compiler option -march=ix86, with x3 (gcc spec). But DFBSD uname -m is returning systematically i386. Is their a workaround for that, that would allow me to compile without hacking the autotools scripts? Why don't you try asterisk18 from pkgsrc -head ? It builds fine out-of the box on DragonFly/i386. -- Francois Tigeot
wip/jdk15 status
For information, I have found the causes of the crashes of lang/kaffe on i386. There's a fix, and I have reported it in pkg/44249: http://gnats.netbsd.org/cgi-bin/query-pr-single.pl?number=44249 With this, wip/jdk15 can now be built on all recent DragonFly systems. -- Francois Tigeot
Re: Native jdk15 build
On Mon, Nov 15, 2010 at 08:23:57AM -0700, Tim Darby wrote: Thanks, I'd like to try this. On Sun, Nov 14, 2010 at 9:10 AM, Francois Tigeot ftig...@wolfpond.orgwrote: I have just succeeded in building a native jdk: $ /usr/pkg/java/jdk-1.5.0/bin/java -version java version 1.5.0_16-p9 I have been given access to pkgsrc-wip and will integrate my changes there piece by piece : http://pkgsrc-wip.cvs.sourceforge.net/viewvc/pkgsrc-wip/wip/jdk15/ It still is not possible to complete a build without copying some .class files from another machine, but I have good hope for the future. The OpenBSD guys have a jdk15 port which can be fully bootstrapped by kaffe; I try to reuse their work when it makes sense. -- Francois Tigeot
Re: 2.8 release schedule - tentitively Wednesday 27 October.
On Fri, Oct 22, 2010 at 11:08:42PM -0400, Justin C. Sherrill wrote: I'm restarting the x86_64 build on 2.9 with the suggested patches for pango/gstreamer, and the other builds are progressing well, so we should be done Wednesday, barring surprises. (crossing my fingers...) Fixes for the devel/gettext-lib bug were just committed to pkgsrc. If you could also rebuild this package from pkgsrc -head, this would definitely help 64-bit users. -- Francois Tigeot
pkgsrc dfly-pkg-people
Hi, I have a pkgsrc bug report with fixes currently sitting in gnats. The responsible field has been changed to dfly-pkg-people. Is there any real person behind this name ? What should I do to have the patches committed ? I figured I should ask first here before going to one of the pkgsrc lists... -- Francois Tigeot
Re: pkgsrc dfly-pkg-people
On Sun, Oct 17, 2010 at 10:23:18PM +0300, Hasso Tepper wrote: On 17.10.10 18:15, Francois Tigeot wrote: The responsible field has been changed to dfly-pkg-people. It's a standard procedure and makes sense of course. Unfortunately no people behind dfly-pkg-people (including me) is active at the moment. As far as I know it will change hopefully. That's a shame :-( I only need someone to commit the patches in pkg/43879. They are really short and to the point (about 4 lines of changes). -- Francois Tigeot
Re: Intel CPU question
On Tue, Oct 12, 2010 at 12:09:51AM +0200, Antonio Huete Jimenez wrote: I recently bought a Phenom X6 1055T in a Gigabyte 880GA-UD3H motherboard, and besides ACPI, everything I've used _seems_ to work fine. I haven't done much testing yet though. For example, I have NOT tested the audio card or run Xorg on it :) 2010/10/11 Tim Darby t+df...@timdarby.net: Does DragonFly support Core I7 or is AMD a safer choice? I'm ready to build a high end multi-core box and was wondering what my options are. Known-to-work motherboard suggestions would also be greatly appreciated. AMD is a safer bet, there's too much complexity on the Intel side nowadays. This Gigabyte board is a good choice; you should try to get an AMD880 chipset based board, everything just works including Xorg. -- Francois Tigeot
Re: Wget Coredumps
On Tue, Oct 12, 2010 at 02:58:26PM -0500, Tyler Mills wrote: [r...@rtgbox] % gdb wget www.google.com [...] Program received signal SIGSEGV, Segmentation fault. 0x000800ab27e4 in libintl_dcigettext () from /usr/pkg/lib/libintl.so.3 (gdb) On latest build from master: DragonFly rtgbox.pavlovmedia.corp 2.7-DEVELOPMENT DragonFly v2.7.3.1279.gf2e93-DEVELOPMENT #1: Tue Oct 12 13:39:29 CDT 2010 r...@rtgbox.pavlovmedia.corp:/usr/obj/usr/src/sys/X64_SMP x86_64 Any ideas what would cause this? There's a bug in gettext-lib, it truncates some pointers to 32-bit on DragonFly/x86-64 if you use non-default locales. More details (and the fix) can be found in this PR: http://gnats.netbsd.org/cgi-bin/query-pr-single.pl?number=43879 This is starting to become a FAQ. Any idea how to get the patches in this problem report committed to pkgsrc ? -- Francois Tigeot
Re: KDE 3 and 4 in recent pkgsrc
On Sat, Oct 09, 2010 at 11:42:17PM -0400, Pierre Abbat wrote: I'm running KDE 3 in pkgsrc 2010Q1. If I upgrade to Q2 or Q3, can I still run KDE 3 or will I be automatically upgraded to KDE 4? Both are in -head; you can choose one or the other. I'm not sure kde4 is fully complete though. -- Francois Tigeot
Re: Qemu on amd64 does not boot installation CDs
On Thu, Oct 07, 2010 at 03:23:24PM -0400, Venkatesh Srinivas wrote: - Some programs dump core from time to time (uptime dependant) This evening, this is the turn of dosbox. It fails with this message: assertion: z-z_Magic == ZALLOC_SLAB_MAGIC in _slabfree Are you running the latest version of world, in addition to a recent kernel? If so, could you file a bug with the core of DOSBox? There were a number of bugs in nmalloc that have been addressed recently, I'd like to know if there are more... I have just upgraded everything: kernel, world and pkgsrc apps (now using -head). Dosbox doesn't crash anymore and most of the strange little issues have disappeared as well. -- Francois Tigeot
Re: Qemu on amd64 does not boot installation CDs
On Thu, Oct 07, 2010 at 06:06:22PM +0530, Siju George wrote: Also is anybody using amd64 as desktop? ;-) I am. The latest 2.7/x86-64 DragonFly versions are now generally quite stable. There are still some quirks left, howewer. Generally, the longer the system stays running, the longer it becomes unstable. Short list: - Seamonkey fails to load web sites on slow servers and spins with 100% CPU. State in top is shown as 'kqread'. Using my ISP proxy makes this behaviour disapear. - Some programs dump core from time to time (uptime dependant) This evening, this is the turn of dosbox. It fails with this message: assertion: z-z_Magic == ZALLOC_SLAB_MAGIC in _slabfree - After a few hours of uptime, keyboard leds always stay off under Xorg even though caps lock/num lock/etc can still change state. The most troublesome stuff is due to a bug in pkgsrc devel/gettext-lib. There's a patch to fix it in my bug report: http://gnats.netbsd.org/cgi-bin/query-pr-single.pl?number=43879 Apart from the gettexti-lib related crashes, 2.6/x86-64 does not have these issues. -- Francois Tigeot
Re: Qemu on amd64 does not boot installation CDs
On Thu, Oct 07, 2010 at 03:23:24PM -0400, Venkatesh Srinivas wrote: - Some programs dump core from time to time (uptime dependant) This evening, this is the turn of dosbox. It fails with this message: assertion: z-z_Magic == ZALLOC_SLAB_MAGIC in _slabfree Are you running the latest version of world, in addition to a recent kernel? If so, could you file a bug with the core of DOSBox? There were a number of bugs in nmalloc that have been addressed recently, I'd like to know if there are more... My system is currently a few days old; I'l upgrade and see if this still happens. The strange thing is, the same dosbox binary ran without trouble a few hours ago... -- Francois Tigeot
Re: example of dfbsd deployment or product that based on dfbsd
On Wed, Sep 29, 2010 at 09:17:29AM +0700, Iwan Budi Kusnanto wrote: I'm looking for production server that used by some company for their mission critical application. My company uses DragonFly on its mail servers and for running its internal ERP application. I'm not sure that really is mission critical tough; we can afford some downtime. -- Francois Tigeot
Re: chlamydia inconsistency?
On Fri, Sep 24, 2010 at 08:49:12PM -0400, Justin C. Sherrill wrote: On Fri, September 24, 2010 6:38 pm, PrzemysÅaw PaweÅczyk wrote: BTW.1. pkg_radd prints segmentation fault, see the picture: http://pp.blast.pl/www.png/dfbsd/df_01.png This I've never seen before. I have. Przemysław, you're running a 64-bit DragonFly system, right ? Did you by chance customize any locale value ? I'm seeing the same crashes with LANG set to non-default/non-english locales on 64-bit systems. The culprit is devel/gettext-lib Try setting LANG to en_US.UTF-8; hopefully this should make the crashes stop. Alternatively, you could try the patches in this bug report: http://gnats.netbsd.org/cgi-bin/query-pr-single.pl?number=43879 Hopefully it will soon be integrated in pkgsrc. -- Francois Tigeot
Re: chlamydia inconsistency?
On Sat, Sep 25, 2010 at 02:10:48PM +0200, Przemysław Pawełczyk wrote: Francois Tigeot Actually, it's François with a cedilla on the C but I drop it for mail to keep everything ascii compliant. I beg your pardon for changing your name to Francis, I do not know how it was possible. Probably I didn't read your name (Fransuaz) but noticed it with my eyes (saved picture Francois in my mind), then I lost o. I'm realy sorry. No problem, don't worry about it. Francis is the english version of my first name. You'll find all sorts of details on Wikipedia ;-) http://en.wikipedia.org/wiki/Francis http://en.wikipedia.org/wiki/François_(disambiguation) -- Francois Tigeot
Re: chlamydia inconsistency? part II.
On Sat, Sep 25, 2010 at 02:01:20PM +0200, Przemysław Pawełczyk wrote: On Sat, 25 Sep 2010 12:28:09 +0200 My main workhorse is Scientific Linux (SL) 5.5- one of the Red Hat Enterprise Linux (RHEL) clone (it has all RHEL marks removed but it is pure RHEL just like PC-BSD which runs on pure FreeBSD). VirtualBox 3.2.8 r64453 (rather latest). I used DragonFly-x86_64-LATEST-ISO.iso from September 23. [...] b) Creating bigger root and home partition were to remove the sig-faults. Alas, it didn't happen. http://pp.blast.pl/www.png/dfbsd/df_06.png - mc compilation http://pp.blast.pl/www.png/dfbsd/df_07.png - kbdmap http://pp.blast.pl/www.png/dfbsd/df_08.png - kbdmap http://pp.blast.pl/www.png/dfbsd/df_09.png - kbdmap http://pp.blast.pl/www.png/dfbsd/df_10.png - kbdmap It seems to me that Francois Tigeot http://leaf.dragonflybsd.org/mailarchive/users/2010-09/msg00125.html was 100% correct. Francis, you wrote: Did you by chance customize any locale value ? I misinterpreted the word any relating it to keyboard language but probably it concerns **ANY** language settings - be it keyboard or console fonts, etc. I'm not sure which settings give the sig-faults as I changed all the three language values but your claims is valid. I hope the above screenshots will help to nail the culprit. Francis Tigeot wrote about devel/gettext-lib, here it was kbdmap. gettext-lib provides a library which is used by all sort of third-party programs for internationalization. sysutils/mc is one of them. I was not aware about the kbdmap issue but I have just run a test and the crash also happens on my system. Since kbdmap does not use gettext-lib from pkgsrc, this must be an other bug. -- Francois Tigeot
Re: Why did you choose DragonFly?
On Mon, Sep 20, 2010 at 01:33:28PM -0600, Samuel J. Greear wrote: What has drawn you to use the DragonFly BSD operating system and/or participate in its development by following this list? Technical features, methodologies, something about the community? I suspect the HAMMER filesystem to be the popular choice, but what other features affect or do you see affecting your day to day life as an administrator, developer, or [insert use case here], now or in the future? I was a FreeBSD user back in 2003; I have seen the announce of DragonFly in one of the mailing lists and watched its progress from afar. I liked some of the early ideas. A new SMP implementation based on the real hardware topology was full of performance promises and since I have a small historical interest in VAXclusters and VMS, I found the SSI goal really appealing. I never really used DragonFly before 1.6, when one of my FreeBSD machines kept crashing (known kernel bug) and I had to find a lasting solution. Since then, I have gradually put DragonFly on production machines, one at a time. I now use it on customer-facing servers and do not regret it. You hear very fast when a mail server is down. What really impressed me is the stability of the system and Hammer. I've never lost any data with DragonFly; even in the worst cases it is possible to synchronize your work with a known good historical state of the filesystem and carry on. The good atmosphere of the community is also a big plus imho. I have not had to refrain from reporting bugs for fear of beeing chastised and most of them are fixed rapidly. I really can see constant progress. So far so good :) -- Francois Tigeot
Re: How to start CUPS?
On Thu, Sep 09, 2010 at 03:02:38PM -0400, Ed Berger wrote: It appears that your shell search path prefers the bsd lpr commands instead of the ones from CUPS that are most likely installed in /usr/pkg/bin instead of /usr/bin by pkgsrc. You can rename the bsd files to something else, or copy them over, but its likely a future upgrade will some time replace them again You can add a line NO_LPR= true in /etc/make.conf and the system will stop building and installing lpr related commands during each make build / installworld. -- Francois Tigeot
pkgsrc patches for www/seamonkey
Hi, The attached patches fix the build problems of www/seamonkey. The resulting binary works fine on DragonFly/i386 I'll send them to the maintainer (tnn@) as soon as I have build a 64-bit binary. -- Francois Tigeot --- directory/c-sdk/ldap/libraries/libldap/Makefile.in.orig 2008-12-24 10:39:55 +0100 +++ directory/c-sdk/ldap/libraries/libldap/Makefile.in 2010-07-05 20:25:40 +0200 @@ -262,6 +262,10 @@ EXTRA_LIBS = -L$(dist_libdir) -l$(LBER_LIBNAME) -pthread endif +ifeq ($(OS_ARCH), DragonFly) +EXTRA_LIBS = -L$(dist_libdir) -l$(LBER_LIBNAME) +endif + ifeq ($(HAVE_SASL), 1) EXTRA_LIBS += $(SASL_LINK) endif --- config/static-config.mk.orig2010-05-04 21:14:54 +0200 +++ config/static-config.mk 2010-07-06 11:07:04 +0200 @@ -104,6 +104,10 @@ STATIC_EXTRA_LIBS += $(MOZ_ALSA_LIBS) endif +ifeq ($(OS_ARCH),DragonFly) +STATIC_EXTRA_LIBS += -lcompat +endif + # Component Makefile always brings in this. # STATIC_EXTRA_LIBS+= $(TK_LIBS) --- directory/c-sdk/configure.in.orig 2009-10-06 23:43:08 +0200 +++ directory/c-sdk/configure.in2010-07-07 11:14:02 +0200 @@ -1233,6 +1233,16 @@ _DEBUG_FLAGS= ;; +*-dragonfly*) +if test -z $USE_NSPR_THREADS; then +USE_PTHREADS=1 +fi +AC_DEFINE(XP_UNIX) +AC_DEFINE(DRAGONFLY) +DSO_CFLAGS=-fPIC +DSO_LDOPTS='-shared' + ;; + *-freebsd*) if test -z $USE_NSPR_THREADS; then USE_PTHREADS=1 --- ./directory/c-sdk/ldap/include/portable.h.orig 2006-10-03 22:43:40 +0200 +++ ./directory/c-sdk/ldap/include/portable.h 2010-07-05 23:15:43 +0200 @@ -295,7 +297,7 @@ #elif defined(HPUX10) #define GETHOSTBYNAME_BUF_T struct hostent_data #define GETHOSTBYNAME( n, r, b, l, e ) nsldapi_compat_gethostbyname_r( n, r, (char *)b, l, e ) -#elif defined(LINUX) +#elif defined(LINUX) || defined(DRAGONFLY) typedef char GETHOSTBYNAME_buf_t [NSLDAPI_NETDB_BUF_SIZE]; #define GETHOSTBYNAME_BUF_T GETHOSTBYNAME_buf_t #define GETHOSTBYNAME( n, r, b, l, rp, e ) gethostbyname_r( n, r, b, l, rp, e ) @@ -317,7 +319,7 @@ || defined(OSF1V4) || defined(AIX) || defined(UnixWare) \ || defined(hpux) || defined(HPUX11) || defined(NETBSD) \ || defined(IRIX6) || defined(FREEBSD) || defined(VMS) \ -|| defined(NTO) || defined(OPENBSD) +|| defined(NTO) || defined(OPENBSD) || defined(DRAGONFLY) #define NSLDAPI_CTIME( c, b, l )ctime_r( c, b ) #elif defined( OSF1V3 ) #define NSLDAPI_CTIME( c, b, l ) (ctime_r( c, b, l ) ? NULL : b)
Re: DragonFly 64-bit stability
On Wed, Jun 09, 2010 at 10:52:44AM +0200, Francois Tigeot wrote: I'm thinking of upgrading one server from 2GB to 6GB of memory. Since the regular DragonFly/i386 version will not be able to fully use it, I'm also considering upgrading the OS to Dragonfly/x86-64. The machine is mainly running Postgres, Apache and Ruby (fast-cgi) for use with a Ruby-on-Rails application. What is your experience with the 64-bit version ? Is it now stable enough to be used in a server ? For the archives, I have now upgraded this server to DragonFly/x86-64. It has been running for a little bit less than a week and is rock solid. I just had to be careful to not set LANG in my shell. I have opened a bug entry for this problem: http://bugs.dragonflybsd.org/issue1782 -- Francois Tigeot
Re: DragonFly 64-bit stability
On Fri, Jun 11, 2010 at 12:44:26PM -0400, Justin C. Sherrill wrote: On Fri, June 11, 2010 2:36 am, Francois Tigeot wrote: Yeah, I don't believe Postgres is to blame either. During the pkgsrc build, many make instances were also dying with signal 11. Every time I have tested the amd64/x86-64 DragonFly port, I found out this segfault problem was a constant. I don't see signal 11 errors on any of the failed builds for x86_64 that I've been doing as bulk builds. Has this happened on more than one x86_64 machine? It's strange. I've found the explanation: The crashes are dependants on the value of the LANG environment variable. With LANG=fr_FR.UTF-8 (my default value), I get instant crashes in many applications. After unsetting $LANG, all applications run as intended, including Postgres. I'll try to find more details and open a proper bug report soon. -- Francois Tigeot
Re: DragonFly 64-bit stability
On Thu, Jun 10, 2010 at 06:55:53PM -0600, Samuel J. Greear wrote: On Thu, Jun 10, 2010 at 6:47 PM, Justin C. Sherrill jus...@shiningsilence.com wrote: On Thu, June 10, 2010 4:32 pm, Francois Tigeot wrote: Installing applications from pkgsrc went well. Unfortunately, running Postgres is a different matter: # /usr/pkg/etc/rc.d/pgsql start Starting pgsql. seg-fault accessing address 0x58 rip=0x80077037d pid=20186 p_comm=pg_ctl Segmentation fault Is this from a prebuilt binary or one that you compiled yourself? It may be worth building locally if you did not before. It was built locally. Otherwise: http://www.postgresql.org/support/submitbug If I had to guess I would say it is likely that this is our bug, probably in one of the kernel sysv subsystems. Yeah, I don't believe Postgres is to blame either. During the pkgsrc build, many make instances were also dying with signal 11. Every time I have tested the amd64/x86-64 DragonFly port, I found out this segfault problem was a constant. -- Francois Tigeot
Re: DragonFly 64-bit stability
On Fri, Jun 11, 2010 at 02:23:59AM -0500, Tyler Mills wrote: What version of postgres are you running? I am able to load 8.4 on an x64 build: DragonFly tyler-bsd.local 2.7-DEVELOPMENT DragonFly v2.7.3.132.g6846f-DEVELOPMENT #4: Thu Jun 10 02:46:08 CDT 2010 r...@tyler-bsd.local:/usr/obj/usr/src/sys/X64_SMP x86_64 I also tried with Postgres 8.4. My test was with DragonFly 2.6, howewer. I'll try to upgrade to 2.7 and see if it makes a difference. -- Francois Tigeot
Re: DragonFly 64-bit stability
On Fri, Jun 11, 2010 at 12:44:26PM -0400, Justin C. Sherrill wrote: On Fri, June 11, 2010 2:36 am, Francois Tigeot wrote: Yeah, I don't believe Postgres is to blame either. During the pkgsrc build, many make instances were also dying with signal 11. I don't see signal 11 errors on any of the failed builds for x86_64 that I've been doing as bulk builds. (wandering through here for example) http://avalon.dragonflybsd.org/reports/x86_64/2.7/20100611.1041/meta/report.html Has this happened on more than one x86_64 machine? It's strange. I have only tested the 64-bit build on one machine for now. The hardware is not unusual: Core 2 Duo, Intel D975XBX2 mainboard. DragonFly/i386 is rock solid on the same PC. -- Francois Tigeot
Re: DragonFly 64-bit stability
On Wed, Jun 09, 2010 at 05:00:18PM -0400, Justin C. Sherrill wrote: On Wed, June 9, 2010 4:52 am, Francois Tigeot wrote: I'm thinking of upgrading one server from 2GB to 6GB of memory. What is your experience with the 64-bit version ? Is it now stable enough to be used in a server ? There's rarely some difference in what stuff from pkgsrc compiles on x86_64 vs. i386, though this is usually not because of DragonFly. A way to check would be looking at the reports on avalon: http://avalon.dragonflybsd.org/reports/ - look at the meta/ directory in each report. Postgres, apache, and ruby build fine going on a quick browse... Thanks Justin and Matt. Since you were so enthusiastic, I had to give a try. Installing applications from pkgsrc went well. Unfortunately, running Postgres is a different matter: # /usr/pkg/etc/rc.d/pgsql start Starting pgsql. seg-fault accessing address 0x58 rip=0x80077037d pid=20186 p_comm=pg_ctl Segmentation fault seg-fault accessing address 0x128 rip=0x800a5037d pid=20190 p_comm=postgres Jun 10 21:32:00 test64 kernel: pid 20186 (pg_ctl), uid 1002: exited on signal 11 Jun 10 21:32:00 test64 kernel: pid 20190 (postgres), uid 1002: exited on signal 11 DragonFly version is the latest stable: v2.6.3.17.g58d915-RELEASE -- Francois Tigeot
DragonFly 64-bit stability
I'm thinking of upgrading one server from 2GB to 6GB of memory. Since the regular DragonFly/i386 version will not be able to fully use it, I'm also considering upgrading the OS to Dragonfly/x86-64. The machine is mainly running Postgres, Apache and Ruby (fast-cgi) for use with a Ruby-on-Rails application. What is your experience with the 64-bit version ? Is it now stable enough to be used in a server ? All answers are welcome. -- Francois Tigeot
Re: HEADS UP: BIND Removal. Short instructions for migration to pkgsrc-BIND
On Sun, Jun 06, 2010 at 12:45:39PM -0400, Justin C. Sherrill wrote: On Sun, June 6, 2010 5:12 am, Jan Lentfer wrote: Jan Lentfer schrieb: After another discussion I have decided to do the following: I will only remove BIND from base, no ldns and drill import. So anyone wanting to have either of the both will have to install them from pkgsrc (before updateing their world, I would recommend). We will see until next release how we will proceed with this. I prefer to just leave it this way and add pkgsrc-BIND to the Live-CD. Due to public demand I have now also committed ldns and drill. Does this mean that we now have a live CD that contains BIND from pkgsrc (and so has host, dig, nslookup, etc) _and_ the lnds/drill tools? The whole point of Jan's work was making it so we have less third-party code to update in base, I thought. My understanding is that nobody should be without the normal BIND tools - they'll just be from the pkgsrc package. It would come on the CD/DVD. If you happened to have an older (2.5 - 2.7) system that was upgraded to remove BIND, it's fixable with 'pkg_radd bind96', rather than needing these additional tools. Someone correct me if I'm not describing reality. +1 The first thing I did on my 2.7 test machine was to install the bind96 package. I just cannot live without host and dig. Of course, the server part of the package is completely useless to me. A minimal bind9-client pkgsrc package would be ideal, IMHO. -- Francois Tigeot
Updating a pkgsrc package
Hi, I have locally updated mail/prayer to version 1.3.2. The pkgsrc package is more than two years old. It now compiles and installs cleanly on DragonFly. Is there any thing I should be aware before submitting my work to the pkgsrc guys ? My main concern is that I do not have any NetBSD machine to test my changes... -- Francois Tigeot
Re: Updating a pkgsrc package
On Sun, May 02, 2010 at 11:09:22AM -0400, Justin C. Sherrill wrote: On Sun, May 2, 2010 7:25 am, Francois Tigeot wrote: I have locally updated mail/prayer to version 1.3.2. Is there any thing I should be aware before submitting my work to the pkgsrc guys ? My main concern is that I do not have any NetBSD machine to test my changes... You also don't have a FreeBSD, Linux, Haiku, or Solaris machine to test on, though pkgsrc runs there... I don't know what your changes look like, but it may be clear whether it has a negative effect or not just from looking at it. A unified diff should work. Send a PR through the support section on the NetBSD website: Right, I have just done so. The PR can be viewed via this URL: http://www.netbsd.org/cgi-bin/query-pr-single.pl?number=43238 -- Francois Tigeot
Re: Snapshots ordering on slave and pfs according to freequency for snapshot management
Hi George, On Wed, Apr 28, 2010 at 03:40:08PM +0530, Siju George wrote: On Tue, Apr 27, 2010 at 1:15 PM, Francois Tigeot ftig...@wolfpond.org wrote: I think sysutils/rsnapshot does what you want. It uses hard links to simulate snapshots on classic filesystems and manages different ranges of snapshots. You can specify how much snapshots you want to keep for each range: [rsnapshot.conf] intervaldaily 7 intervalmonthly 12 The hammer utility would be much better if it implemented some similar mechanism IMHO. I am familiar with rsnapshot and have been using it/backuppc but I guess hammer snapshot is much faster and that rsnapshot when the directory has many files. Also when there are many small changes to many files spread over the day rsnapshot will store all the changed file as they are induvidually and will take more space if I take a snapshot every 5 mins. Yeah, I'm in full agreement with you. I think I was a bit misanderstood: Hammer performance is *much better* than rsnapshot (obviously) but there is no way to easily tell it to keep its snapshot distribution in different intervals for archiving purposes. This is the one thing rsnapshot does extremely well; I was just trying to say this is a good idea we should steal :-) -- Francois Tigeot
Re: Snapshots ordering on slave and pfs according to freequency for snapshot management
On Mon, Apr 26, 2010 at 06:56:50PM -0700, Matthew Dillon wrote: :All I need is to figure out how to remove the 5mins snapshots that :gets mirrored on the slave older than two days with out removing the :daily snapshots. : :But I am a bit confused now since I dont see snapshots actually :removed after a hammer cleanup. :I will send the details with a new subject : :--Siju hammer cleanup only removes snapshots over X days old. It can't distinguish between fine-grained and coarse-grained snapshots that you explicitly tell hammer to make. You would have to remove those yourself (if you want to expire them before the X days) using hammer snaprm. You can script it fairly easily by setting the comment field for each snapshot you take, then filtering out the list based on that. See the manual page. I think sysutils/rsnapshot does what you want. It uses hard links to simulate snapshots on classic filesystems and manages different ranges of snapshots. You can specify how much snapshots you want to keep for each range: [rsnapshot.conf] intervalhourly 6 intervaldaily 7 intervalweekly 4 intervalmonthly 12 The hammer utility would be much better if it implemented some similar mechanism IMHO. -- Francois Tigeot
Re: DragonFly 2.6.2, 2.7.2 tags pushed - fixes for serious HAMMER issue
On Tue, Apr 20, 2010 at 11:03:55AM -0700, Matthew Dillon wrote: :Yah, indeed it does :'( : : sudo hammer -f /dev/serno/QM2.s1a checkmap :Volume header records=0 next_tid=00010841bec0 :bufoffset=4404 :Collecting allocation info from B-Tree: done :BM block=20001000 calc 114688 free, got 1163264 : :Now what? Is really the mirror-copy the only solution? I basically :have no means to do that ... Pretty much, short of modifying the hammer utility to correct the allocation info. From the point of view of a guy having the means to do the copy stuff (two harddrives), it is still a bit of a PITA. 300+ GB take hours to copy :-( Having a fsck.hammer would be a big plus IMHO. -- Francois Tigeot
Re: Monitoring CPU time
On Mon, Apr 05, 2010 at 07:00:06PM +0200, Francois Tigeot wrote: On Mon, Apr 05, 2010 at 09:21:41AM -0700, Matthew Dillon wrote: systat -pv 1 In terms of extracting it in a script... I dunno about that. I'm beginning to think I will need to write a C program for that. top(1) uses a kinfo_proc structure to get its data and AFAIK there's no easy-to-use sysctl. systat and vmstat give only an instant snapshot of the system state with 1 second precision or so. An average on 5 minutes would be much better for my purposes. I have written a small C program to obtain the same sort of results sysctl kern.cp_time returns on Free, Open and possibly NetBSD. The actual values are a sum of the relevant fields on all CPUs. You just have to be careful to adjust N_CPUS. -- Francois Tigeot /* * cp_time.c * returns user,nice,system,intr and idle values from cputime counters * similar to some BSDs kern.cp_time sysctl */ #include stdio.h #include kinfo.h #define N_CPUS 2 int main( void ) { int cpu, len; struct kinfo_cputime cp_t[N_CPUS]; uint64_t user, nice, sys, intr, idle; bzero( cp_t, sizeof(struct kinfo_cputime)*N_CPUS ); len = sizeof( cp_t[0]) * N_CPUS; if (sysctlbyname(kern.cputime, cp_t, len, NULL, 0)) err(1, kern.cputime); user = nice = sys = intr = idle = 0; /* Sum up cp_time for all cpus */ for (cpu = 0; cpu N_CPUS; cpu++) { user += cp_t[cpu].cp_user; nice += cp_t[cpu].cp_nice; sys += cp_t[cpu].cp_sys; intr += cp_t[cpu].cp_intr; idle += cp_t[cpu].cp_idle; } printf(%llu %llu %llu %llu %llu\n, user, nice, sys, intr, idle ); return 0; }
Re: Monitoring CPU time
On Mon, Apr 05, 2010 at 09:21:41AM -0700, Matthew Dillon wrote: systat -pv 1 In terms of extracting it in a script... I dunno about that. I'm beginning to think I will need to write a C program for that. top(1) uses a kinfo_proc structure to get its data and AFAIK there's no easy-to-use sysctl. systat and vmstat give only an instant snapshot of the system state with 1 second precision or so. An average on 5 minutes would be much better for my purposes. -- Francois Tigeot
Monitoring CPU time
Hi, I'm currently playing with munin http://munin-monitoring.org/ and I would like to monitor the CPU usage on DragonFly hosts. The FreeBSD plugin uses the sysctl kern.cp_time, which is not present on DragonFly. Is there any way to easily get the different percentages of nice, idle, user, system, etc... time from a shell ? FreeBSD cpu usage plugin sources: http://munin-monitoring.org/browser/trunk/node/node.d.freebsd/cpu.in?rev=900 -- Francois Tigeot
Re: Monitoring CPU time
On Sun, Apr 04, 2010 at 03:01:47PM +, Chris Turner wrote: Francois Tigeot wrote: The FreeBSD plugin uses the sysctl kern.cp_time, which is not present on DragonFly. Is there any way to easily get the different percentages of nice, idle, user, system, etc... time from a shell ? there might be a better way to do this - but you can snag that particular set of data from vmstat -c 1 .. Thanks for the tip, it's almost what I want. FreeBSD: $ sysctl kern.cp_time kern.cp_time: 1797279 579339 1349174 128959 811149589 DragonFly: $ vmstat -c 1 | tail -n 1 | awk '{ print $14,$15,$16 }' 675875492 705365686 1527141678 Unfortunately, there's no data for nice or interrupt. Time to dig into top sources... -- Francois Tigeot
Re: SAS RAID controllers support
On Thu, Feb 04, 2010 at 06:20:51PM +0100, Francois Tigeot wrote: I'm curious about the state of hardware RAID controllers in DragonFly. I'm now pretty sure the only hardware RAID adapters which *could* be usable are based on the LSI 1078 chipset At least 6 different cards are based on it: - LSI MegaRAID SAS 8704ELP / 8708ELP - LSI MegaRAID SAS 8704EM2 / 8708EM2 - LSI MegaRAID SAS 8880EM2 - LSI MegaRAID SAS ELP I should be able to test a MegaRAID SAS ELP soon. It was not recognized last year but Matt has since updated the mpt(4) driver. The MegaRAID SAS ELP is still not recognized. dmesg : pci3: PCI bus on pcib3 pci3: mass storage, RAID (vendor 0x1000, dev 0x0060) at device 0.0 irq 11 pciconf -lv : non...@pci0:3:0:0: class=0x010400 card=0x10061000 chip=0x00601000 rev=0x04 hdr=0x00 vendor = 'LSI Logic (Was: Symbios Logic, NCR)' device = 'SAS1078 PCI-X Fusion-MPT SAS' class = mass storage subclass = RAID So it seems DragonFly doesn't support any recent hardware RAID controller. -- Francois Tigeot
Re: SAS RAID controllers support
On Wed, Feb 24, 2010 at 12:02:36AM +0100, Michael Neumann wrote: 2010/2/23 Francois Tigeot ftig...@wolfpond.org The MegaRAID SAS ELP is still not recognized. So it seems DragonFly doesn't support any recent hardware RAID controller. The Adaptec RAID (aac) controllers seems to be supported very well. For example the Adaptec RAID 5405 seems to work pretty well on a box I tested Dragonfly. Thanks for the information. Adaptec is the only vendor I didn't review - too many problems with their products. I think I will go for a software raid solution. -- Francois Tigeot
Re: SAS RAID controllers support
On Mon, Feb 01, 2010 at 07:16:46PM +0100, Francois Tigeot wrote: On Mon, Feb 01, 2010 at 08:48:37AM -0800, Freddie Cash wrote: On Mon, Feb 1, 2010 at 4:41 AM, Francois Tigeot ftig...@wolfpond.orgwrote: On Mon, Feb 01, 2010 at 12:02:38PM +0100, Francois Tigeot wrote: I'm curious about the state of hardware RAID controllers in DragonFly. I'm now pretty sure the only hardware RAID adapters which *could* be usable are based on the LSI 1078 chipset At least 6 different cards are based on it: - LSI MegaRAID SAS 8704ELP / 8708ELP - LSI MegaRAID SAS 8704EM2 / 8708EM2 - LSI MegaRAID SAS 8880EM2 - LSI MegaRAID SAS ELP http://www.lsi.com/storage_home/products_home/internal_raid/megaraid_sas/megaraid_sas_8704elp/index.html http://www.lsi.com/storage_home/products_home/internal_raid/megaraid_sas/value_line/megaraid_sas_8704em2/index.html http://www.lsi.com/storage_home/products_home/internal_raid/megaraid_sas/megaraid_sas_8708elp/index.html http://www.lsi.com/storage_home/products_home/internal_raid/megaraid_sas/value_line/megaraid_sas_8708em2/index.html http://www.lsi.com/storage_home/products_home/internal_raid/megaraid_sas/feature_line/megaraid_sas_8880em2/index.html http://www.lsi.com/storage_home/products_home/internal_raid/megaraid_sas/megaraid_sas_elp/index.html I should be able to test a MegaRAID SAS ELP soon. It was not recognized last year but Matt has since updated the mpt(4) driver. -- Francois Tigeot
SAS RAID controllers support
Hi, I'm curious about the state of hardware RAID controllers in DragonFly. I know there is some sort of support for old 3Ware SATA cards. Nowadays I'm more interested in SAS controllers, especially since cheap versions can be found onboard recent entry-level server mainboards. My goal is to use two SATA drives in a mirroring configuration. Would a LSI1068-E controller be usable with DragonFly-2.4 ? Can the status of the RAID volume be probed from the OS ? Is there any way to get an email alert on a disk failure ? Can the volume reconstruction be managed from the OS in the event of a disk replacement ? Inquiring minds want to know ! -- Francois Tigeot
Re: SAS RAID controllers support
On Mon, Feb 01, 2010 at 12:02:38PM +0100, Francois Tigeot wrote: I'm curious about the state of hardware RAID controllers in DragonFly. Would a LSI1068-E controller be usable with DragonFly-2.4 ? This page answers my question for this controller: http://blogaristoo.lqx.net/index.php/2009/01/14/sassy-lip-from-the-lsi-1068e LSI1068e is *not* usable at all. -- Francois Tigeot
Re: SAS RAID controllers support
On Mon, Feb 01, 2010 at 08:48:37AM -0800, Freddie Cash wrote: On Mon, Feb 1, 2010 at 4:41 AM, Francois Tigeot ftig...@wolfpond.orgwrote: On Mon, Feb 01, 2010 at 12:02:38PM +0100, Francois Tigeot wrote: I'm curious about the state of hardware RAID controllers in DragonFly. Would a LSI1068-E controller be usable with DragonFly-2.4 ? This page answers my question for this controller: http://blogaristoo.lqx.net/index.php/2009/01/14/sassy-lip-from-the-lsi-1068e LSI1068e is *not* usable at all. You'd be better off going with separate PCI-X/PCIe controllers. Not sure about DragonFlyBSD, but FreeBSD fully supports these cards (mpt(4), mfi(4), twa(4)): [...] 4-port PCIe: LSI SAS 3041E-R 8-port PCIe: LSI SAS 3081E-R There's also the 3Ware cards (these are SAS and SATA): [...] Thanks for the list. There is no PCI-X support on the server boards I have in mind, so I can only validate PCI-E adapters. There's a twa driver in sys/dev/raid/ but it has not been updated since 2004, so I guess none of the recent 3Ware adapters are supported. There's also a mpt driver in sys/dev/disk/ which has been recently updated. No trace of a mfi driver. I will focus on LSI external adapters for now. -- Francois Tigeot
Re: Anyone tried an Atom 330 with Dragonfly
Hi, On Sat, Jan 30, 2010 at 12:24:12PM +, Steve O'Hara-Smith wrote: One of my workstations died and I'm looking to replace it, since the computing needs are not great I thought it might be nice to use something low powered and the dual core Atom 330 looks like a good option. I'm a little torn between two Asrock boards the A330GC with an Intel chipset and the rather newer A330ION with it's faster memory and all out more of everything and apparently lower power consumption than the lesser Intel chipset. There's a brand new generation of Atom CPUs out there since this month : http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3692 If you can afford to wait a little while, I would recommend to go with an Atom D510 (330 equivalent: two cores, 2 threads per core). I'm thinking myself of getting this board soon: http://www.supermicro.com/products/motherboard/ATOM/ICH9/X7SPA.cfm?typ=H The two gigabit ethernet and 6 sata ports are particularly nice for a small file server. -- Francois Tigeot
Re: OpenOffice
On Wed, Jan 20, 2010 at 10:57:25PM -0500, Justin C. Sherrill wrote: On Wed, January 20, 2010 5:14 pm, Pierre Abbat wrote: I tried building openoffice2 and it said it's not available on DragonFly. FreeBSD and NetBSD, yes, but not DFly. How come? And openoffice3 gave me Could not find ../../devel/xulrunner/buildlink3.mk and quit. Openoffice2 probably just never was tested on DragonFly; looking at the Makefile, it's pretty specific about supported versions. Openoffice3 also fails with a openoffice3-3.1.1nb1 is not available for DragonFly-2.4.1-i386 error on the last 2.4.1 i386 bulk build, so I bet it's a similar issue there. It may just be a matter of someone familiar with the build process needing to poke at it. The build process is a mess. It would need a considerable time investment to be rendered usable. If you want to use OpenOffice on DragonFly, your best bet is the linux binary in misc/openoffice3-bin. And then it manages to crash if you use a hammer filesystem... There are some workarounds in this thread: http://www.mail-archive.com/users@crater.dragonflybsd.org/msg08862.html Nowadays, I use koffice. It's native and fast. -- Francois Tigeot
Re: gcc mm_malloc.h
On Fri, Aug 28, 2009 at 12:52:30AM +0200, Simon 'corecode' Schubert wrote: Francois Tigeot wrote: As far as I know, mm_malloc.h is part of gcc-4.1. Is there any reason it is not installed in a DragonFly-2.3.2 system ? I have a patch for this, will commit in the morning. That's great, thanks ! -- Francois Tigeot
gcc mm_malloc.h
Hi, wip/kdegraphics4 fails to build with a missing reference to mm_malloc.h [ 72%] Building CXX object khtml/CMakeFiles/khtml.dir/misc/borderarcstroker.o In file included from /usr/pkgsrc/wip/kdelibs4/work/kdelibs-4.3.0/khtml/misc/borderarcstroker.cpp:36: /usr/libdata/gcc41/xmmintrin.h:41:23: error: mm_malloc.h: No such file or directory --- khtml/CMakeFiles/khtml.dir/misc/borderarcstroker.o --- *** [khtml/CMakeFiles/khtml.dir/misc/borderarcstroker.o] Error code 1 1 error The error is not in a kde file per se but in a gcc header. As far as I know, mm_malloc.h is part of gcc-4.1. Is there any reason it is not installed in a DragonFly-2.3.2 system ? -- Francois Tigeot
Re: Is the Giant lock completely removed?
On Fri, Aug 07, 2009 at 11:36:40AM +0100, Steve O'Hara-Smith wrote: On Fri, 7 Aug 2009 12:43:19 +0530 Siju George sgeorge...@gmail.com wrote: Tomorrow I will be addressing a meeting hosted by our police department to educate people on the security advantages of using FOSS. I would like to include a section in my talk to speak on dragonfly. One doubt is has the Giant lock been completely removed from dragonfly? And in addition to HammerFS what else can I state as advantages for using dragonfly? Any hints I will be very grateful :-) Distinguishing features of DragonFly include: Downsides No Java I wouldn't be so sure: at one time it was relatively painless to build jdk-1.4 binaries. I still have a tomcat installation running with it in a vkernel: $ pkg_info | grep jdk jdk14-1.4.2.8 Java Development Kit 1.4.2 $ uname -sir DragonFly 2.2.2-RELEASE VKERNEL I remember there were also some patches to get jdk15 to build floating on the mailing-lists. AFAIK, the big problem with getting a native jdk is due to the old linuxulator code: the build has to be bootstrapped with a Linux jdk binary. -- Francois Tigeot
Re: ps2 mouse driver problem -- kernel psmintr error messages
On Wed, Jun 24, 2009 at 11:13:06AM -0400, Ed Berger wrote: With either the liveDVD or installing from CDROM and building X11 from pkgsrc, on a HP m8000n media PC I have problems with the mouse cursor, which flies to the right side of the screen as soon as its touched, and seems to just jump up and down on the far right of the screen when the mouse is moved. The PS2 mouse works fine with linux (centos, opensuse) or windows (xp, vista). It was detected as kernel: psm0: PS2 Mouse irq 12 on atkbdc0 kernel: psm0: model IntelliMouse, device ID 3 In the log there are messages, which don't look encouraging to me. kernel: psmintr: out of sync (0008 != ) kernel: psmintr: discard a byte (1). kernel: psmintr: out of sync (0008 != ) kernel: psmintr: discard a byte (2). kernel: psmintr: out of sync (0008 != ) kernel: psmintr: discard a byte (3). You may have an interrupt problem. I had a similar issue when I enabled the emergency interrupt stuff recently. The log messages where exactly the same. -- Francois Tigeot
Instant crash with Linux OpenOffice
Hi, I have recently upgraded the root fs to Hammer on one of my machines. Since then, I have been unable to run any version of OpenOffice. Previously, misc/openoffice2-bin and misc/openoffice3-bin ran fine. The OS is DragonFly 2.2.2-RELEASE. The splash screen begins to appear and then OpenOffice crashes with some weird errors: $ soffice javaldx: Could not find a Java Runtime Environment! sh: g/,: No such file or directory $ soffice javaldx: Could not find a Java Runtime Environment! sh: -c: line 0: unexpected EOF while looking for matching `'' sh: -c: line 1: syntax error: unexpected end of file I'm not sure how to debug this. -- Francois Tigeot
Re: Instant crash with Linux OpenOffice
On Tue, Jun 02, 2009 at 12:11:35PM -0700, Matthew Dillon wrote: What I did on my laptop to make it work was put /usr/pkg/openoffice.org2.4/share/fonts on a small UFS partition (my /boot) and then null-mount it onto /usr/pkg/openoffice.org2.4/share/fonts. OpenOffice has other issues unrelated to the filesystem... it has issues trying to locate files due to the prioritized linux emulation path lookups. It gets very confused sometimes. I had a quick look at the source distribution; the build system is full of arcane and unneeded stuff. Like configure scripts auto-generating perl scripts, themselves sourcing environment variables from two possible different shell scripts, one bash, one tcsh for good mesure, etc... It also has its own make program, and about a dozen old versions of other programs and libraries thrown in the tarball for fun. Pure madness. I can only imagine what sort of bugs and assumptions lie in the source code proper... Aniway, I tried your trick of null-mounting fonts/ from /boot. It works for reading documents but fails when trying to save them: Error saving the document blah: Write Error. The file could not be written. blah.ods is in my /home pfs, and chmod 0666. The auto-save feature gives a different error message: Openoffice.org could not save important internal information due to insufficient free disk space at the following location: /home/ftigeot/.openoffice.org/3/user/backup You will not be able to continue working with OpenOffice.org without allocating more free disk space at that location. There is 80 GB of free space on this fs. Could it be counting free inodes ? -- Francois Tigeot
Re: Java 2 status
On Mon, Apr 28, 2008 at 10:28:22PM +0200, VOROSKOI Andras wrote: On Mon, Apr 28, 2008 at 01:18:47PM -0700, Matthew Dillon wrote: I haven't tried. I did hear recently that Sun said something about opening up Java for real, hopefully that will result in easier portage. Yep, they say so: http://news.zdnet.co.uk/software/0,100121,39405249,00.htm?r=7 The OpenBSD people seem to have created a usable jdk 1.7 port: http://www.undeadly.org/cgi?action=articlesid=20080321023803 -- Francois Tigeot