FS: APC SUA1000RM2U Smart-UPS
It's heavy so selling online/shipping isn't practical. Anyone interested? Available for pickup in Hollis or I might be convinced to bring it somewhere close by. Looking for $225, but make me an offer. APC SUA1000RM2U Smart-UPS 2U Rackmount w/AP9617 Network Card & 4-post rails Testing and working, was in use in my basement rack until no longer needed. Includes Fresh Batteries (about 1 year old) Includes AP9617 Network Managent Card Includes both 2-post and 4-post rails with screws and hardware https://www.apc.com/shop/gy/en/products/APC-Smart-UPS-1000VA-USB-Serial-RM-2U-120V/P-SUA1000RM2U -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Local inexpensive media destruction?
Anyone know of a local/inexpensive media destruction service? As I'm sure many of you do, I have a box with 15+ years of obsolete hard drives, floppy, tapes, CDR, DVDR, flash, etc... in my house. While there has been many trips to e-scrap over the years to get rid of computers, media is always removed before hand. There are plenty of companies that will provide me with destruction services including secure transport or on-site processing, liability insurance, certification, logs of all items, videos of my actual items being shredded, etc.. they all come with equally impressive bills. Simple degausing and/or physical destruction (preferably while I watch because that'd be cool) and the resulting scrap recycled is sufficient. If not, sounds like an opportunity for a data destruction party where someone has/rents a degauser and appropriate shredder. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
free: Supermicro 5012B6 1U server
Supermicro 5012B6 1U server http://www.supermicro.com/products/system/1u/5012/sys-5012b-6.cfm Intel P4 Celeron 1800Mhz Single 512MB ECC DIMM Two U160 SCA 36GB hotswap SCSI drives (plus 1 spare drive) Ran reliabily for 5+ years, multi-year uptimes, and zero unexpected crashes over its lifetime, retired a few years ago. Was fully functional at time of retirement, includes rackmount rails and original boxes/packing materials. It's going to e-scrap Monday morning unless someone wants it. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Widget to manipulate parallel port signals ?
Michael ODonnell writes: Anybody know of a (commandline or GUI) utility that I could use to wiggle/sense the individual data/control lines of a parallel port? I'd prefer that it operate using one of the standard drivers (like parport_pc) via ioctls rather than poking around directly in I/O or memory space at hardcoded addresses as I'd hope it'd be flexible enough to work with either an integral legacy device or an add-in device connected via PCI, USB, etc. I found (but for the moment have misplaced the URL of) one that appears to be somebody's homework project; it has a GUI that presents the control/data lines separated into two clusters of buttons whose color indicates current state and that you can click to change their state, etc, etc, but it looked like I'd have to duplicate his C++ development environment to get it working, so before I go down that road I'm hoping somebody knows of one that'd be more of a turn-key experience... I wrote a test program when I was making my lcd control daemon for my router. http://centerclick.org/temp/lcd.tgz the part you want is lcdraw.c it'll take commands from cmdline and do all basic operations to the parallel port. Usage: ./obj/lcdraw command [command...] command is one of: rr data direction in rw data direction out dr data read dw value data write cr control read cw value control write sr status read kr lcd data ram read kw value lcd data ram write ir lcd instruction ram read iw value lcd instruction ram write -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: How can I retrieve the mount count for an ext3 volume?
Alex Hewitt writes: My Ubuntu 8.10 system uses EXT3 for the root filesystem and will automatically fschk the volume every 35 mounts. I haven't been able to find out where the mount count is stored or how that data can be retrieved. I don't want to change the automated fschk but I'd like to display the count so I can anticipate when the volume will be checked. -Alex $ dumpe2fs -h /dev/sda1 [...] Mount count: 9 Maximum mount count: 24 Last checked: Tue Mar 10 22:06:05 2009 Check interval: 15552000 (6 months) Next check after: Sun Sep 6 22:06:05 2009 [...] -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: How can I retrieve the mount count for an ext3 volume?
Ben Scott writes: On Tue, Oct 6, 2009 at 10:43 AM, Tom Buskey t...@buskey.name wrote: And RAIDs should be scrubbed periodically. Modern RAID controllers usually feature something called patrol read, which reads all the blocks on the physical disks in the background, when otherwise idle. Is there a similar feature in Linux's RAID implementation? Yes, md has a 'check' action that you can issue to the kernel to have it run a RAID consistancy check. mdadm issues this once a week (by default) you can also issue this manually with something like: $ echo check /sys/block/md0/md/sync_action -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: [OT] Generator testing
Ben Scott writes: But it's funny, lots of people have disaster stories... it seems everyone knows what *not* to do, or what can go wrong... but if you ask what you *should* do, and people are less certain. :) umm.. ya. speaking of what not to do, don't keep adding load to your building transformer and not notice this. Worked at a place where after adding more and more servers and lab equipment every month, the transformer outside the building was way overloaded 24/7. I think it was already scheduled for upgrade, but it wasn't soon enough. It started leaking oil into the parking lot one day where someone noticed it and called the power company. They got there really quick and shut it off right now before it could fail spectacularly. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Finding *unfiltered* free WiFi? (was: WAP/Router for use with OpenVPN)
Bill McGonigle writes: We've got the `open database of general knowledge' (Wikipedia), the open database of maps (OpenStreetMap), the open database of speed-limit signs (Wikispeedia), the open database of GSM cell-sites (OpenBmap)..., why not one for WiFi-hotspots? We actually talked about this a bit at the DLSLUG meeting on OpenStreetMap. A WiFi node is just another type of node, with a certain tag. I think somebody said wardrivers have already automated this? It makes more sense to add the data to OpenStreetMap than to create another database. Google has been recording location data of WiFi APs (no surpise there), too bad the data isn't exported in a friendly way. From what I can tell, anywhere that has been Street View'ed has also had all WiFi AP's recorded as the car passed by taking pictures. This was rather obvious when using the iPhone 2G (no GPS). It would contact some server via HTTPS and (presumibily) send nearby WiFi AP data in an attempt to get a more precise location. Worked great when driving down a street that had been Street View'ed. Whenever an AP from someone's house got in range, it would narrow down the location rather well. As soon as you went to a street with no street view pictures, it would revert back to the less accurate cell tower location. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: WAP/Router for use with OpenVPN
Ben Scott writes: And they have to have enough computing-power to run WPA, right? Wireless crypto may be implemented in dedicated hardware in the wireless chipset, not on the general-purpose processor (where Linux and OpenVPN run), so that may not mean anything. Yes, WiFi crypto will definately be done in hardware. If you're going to use openvpn without a hardware assist, (like a HiFn, etc.. ) cpu performance may be of concern. Hardware assist in openssl is usually hard to find in general. Shouldn't be an issue for occasional single user connectivity though, unless you need many mbps. I have my openvpn links use blowfish instead of AES for the data channel because it's less cpu intensive especially for small block sizes. For comparision, I use a Soekris net4801 (266Mhz NSC/AMD Geode) for my router/firewall and openvpn endpoints. It does IPv4 forwarding fine, up only up to about 50mbps for large packets. Definately not good for LAN-LAN, but fine for LAN-WAN. Throw in openvpn, the crypto and compression will drop the vpn data to a few mbps. Good for connectivity, just not performance. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: recommendations on virtualization software
Mark Ellison writes: Hi, I am seeking recommendations and pros/cons of different virtualization software. The physical machine is a Intel T9400 quad core with 8GB ram, 2x500GB sata disks and 1Gb nic. My current plan is to run 64 bit Fedora Core 10 (or 11 as available) as the host OS. The guest OSes will include a mix vista, xp and other UNIX variants. I am aware of the commercially available VMware workstation, VirtualBox and Xen. Any feedback and recommendations are appreciated. I've been impressed with kvm and have it running several guests right now. Pros: * supported by linux kernel proper * active development * emulates guest NICs that work with jumbo frames (Intel 82540EM, e1000) * built-in kvm server for vga console, standards compliant, not some custom viewer * serial console via telnet * designed for headless hosts * faster than needed Cons: * no support for multi-cpu/core guests * virtio isn't fully baked yet, but looks promising * had to write some custom start/stop scripts -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: /proc/$$/fd mystery - {t,}csh vs. bash
Michael ODonnell writes: e521:~ 514--- /bin/csh e521:~ echo My PID is $$, contents of my /proc/$$/fd follow... ; ls -l /proc/$$/fd My PID is 27244, contents of my /proc/27244/fd follow... total 0 lrwx-- 1 mod mod 64 2009-01-08 20:47 15 - /dev/pts/0 lrwx-- 1 mod mod 64 2009-01-08 20:47 16 - /dev/pts/0 lrwx-- 1 mod mod 64 2009-01-08 20:47 17 - /dev/pts/0 lrwx-- 1 mod mod 64 2009-01-08 20:47 18 - /dev/pts/0 lrwx-- 1 mod mod 64 2009-01-08 20:47 19 - /dev/pts/0 Ah the wonders of strace Because on startup /bin/csh dups stdin/out/err to higher fds... execve(/bin/csh, [/bin/csh], [/* 31 vars */]) = 0 [... blah blah ...] dup2(0, 16) = 16 fcntl64(16, F_SETFD, FD_CLOEXEC)= 0 dup2(1, 17) = 17 fcntl64(17, F_SETFD, FD_CLOEXEC)= 0 dup2(2, 18) = 18 fcntl64(18, F_SETFD, FD_CLOEXEC)= 0 dup2(16, 19)= 19 fcntl64(19, F_SETFD, FD_CLOEXEC)= 0 [... blah blah ...] close(0)= 0 close(1)= 0 close(2)= 0 this means /bin/csh uses FDs 16,17,18 as it's stdin/out/err instead of 0/1/2. when running a program: after clone(), it just dup()s back to stdin/out/err before exec(): clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0xb7d266f8) = 11073 [... blah blah ...] [pid 11073] close(0)= -1 EBADF (Bad file descriptor) [pid 11073] dup(19) = 0 [pid 11073] fcntl64(0, F_SETFD, 0) = 0 [pid 11073] close(1)= -1 EBADF (Bad file descriptor) [pid 11073] dup(17) = 1 [pid 11073] fcntl64(1, F_SETFD, 0) = 0 [pid 11073] close(2)= -1 EBADF (Bad file descriptor) [pid 11073] dup(18) = 2 [pid 11073] fcntl64(2, F_SETFD, 0) = 0 [... blah blah ...] [pid 11073] execve(/bin/ls, [ls, -l, /proc/11071/fd], [/* 37 vars */]) = 0 -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: /proc/$$/fd mystery - {t,}csh vs. bash
Michael ODonnell writes: Because on startup /bin/csh dups stdin/out/err to higher fds... [...] this means /bin/csh uses FDs 16,17,18 as it's stdin/out/err instead of 0/1/2. [...] when running a program: after clone(), it just dup()s back to stdin/out/err before exec(): Cool - thanks for the analysis. Why might they be doing this? from csh.h: /* * The shell moves std in/out/diag and the old std input away from units * 0, 1, and 2 so that it is easy to set up these standards for invoked * commands. */ #define FSHTTY 15 /* /dev/tty when manip pgrps */ #define FSHIN 16 /* Preferred desc for shell input */ #define FSHOUT 17 /* ... shell output */ #define FSHERR 18 /* ... shell diagnostics */ #define FOLDSTD 19 /* ... old std input */ It does make it less complicated if you invoke programs with weird redirection arguments. If fd 0/1/2 are all unused just before you exec() then its simple to dup/open these based on whatever redirection you want the program to run under. If they are used then you may have to move stuff out of the way before setting up redirection or accidentially misplace the shell's actual stdin/out/err while setting up redirection. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Is this a postfix problem? Receiving mail from cellphone
Dan Coutu writes: This is a RHEL server running postfix. Sending email to the server from my cell phone is giving an error and I don't understand why. I'm hoping that someone here can shed light on it for me. Here's the mail log entries that show the problem: Dec 1 01:55:27 ec2-75-101-156-55 postfix/smtpd[31695]: connect from 150.sub-69-78-129.myvzw.com[69.78.129.150] Dec 1 01:55:27 ec2-75-101-156-55 postfix/smtpd[31695]: NOQUEUE: reject: RCPT from 150.sub-69-78-129.myvzw.com[69.78.129.150]: 450 4.7.1 njbrspamp5.vtext.com: Helo command rejected: Host not found; from=[EMAIL PROTECTED] to=[EMAIL PROTECTED] proto=ESMTP helo=njbrspamp5.vtext.com Dec 1 01:55:32 ec2-75-101-156-55 postfix/smtpd[31695]: disconnect from 150.sub-69-78-129.myvzw.com[69.78.129.150] Thanks for any help or pointers to resolving this. You phone (IP 69.78.129.150) is using a smtp client (or smtp relay) that is identifying itself in the HELO SMTP message as njbrspamp5.vtext.com which doesn't resolve using DNS. njbrspamp5.vtext.com is likely a smtp relay which simply doesn't have a public dns entry. You can (carefully) loosen the HELO restrictions on your mail server if you want to bypass this check for specific hosts. Since the postfix default is to allow everyting then you've likely already modified this. See: http://www.postfix.org/postconf.5.html#smtpd_helo_restrictions -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Old rackmount equipment?
Drew Van Zandt writes: Hey all, I'm looking for old, not necessarily functional rackmount equipment to fill a rack with for some airflow experiments I'm doing, and I was wondering if anyone had junk lying around that would qualify. Anything I can bolt into a rack will do, from shelves to token-ring switches. Bonus points if I can plug it in and it produces heat (but not fire, preferably.) I've got 3 Bay/Nortel BayStack 450 switches (1.5U) with rack-mount hardware, stacking modules, various uplink modules, etc... Fully functional (though I think one has a single broken fan). Free to you or anyone else who wants to put them to actual use and can pick them up from Dracut MA. They heat left to right instead of front to back, not sure how that'll play into your airflow experiments. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Blackberry webpages
Travis Roy writes: This is a bit off topic, but does anybody know of any programs or webpages that will let me test a website to see how it will look on a blackberry? I'm trying to get a webpage to view properly and not having a blackberry to test it on makes for slow going. Doesn't have blackberry, but still an interesting site: http://browsershots.org/ -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Serial admin console program
Shawn O'Shea writes: Back in my days of managing my Sun boxes, we had a Lantronix and I hated it. My experience with them is *years* old though. I'm partial to the Cyclades (now Avocent) myself. Formerly had the TS2000, now Avocent pushes the ACS48. I've also used Logical Solutions SCS series (liked supporting them when I was in CT because they were local, Milford, CT, got to visit their offices) and some guys here at work also use the Raritan Dominions. They are all Linux based. If you are looking to roll-your-own, basically all of these Linux products do their port management with Portstlave. Linkroll: Avocent Cyclades ACS: http://www.avocent.com/Products/Default.aspx?id=6846 Logical Solutions SCS: http://www.thinklogical.com/product.asp?ID=27 Raritan Dominion: http://raritan.com/products/serial-console-switches/ Portslave: http://sourceforge.net/projects/portslave/ I use both the Cyclades TS2000 and Avocent ACS32 at work. Cyclades boxes have an underpowered CPU (they grind to a halt if you try to actually do 32 interactive, plus 32 sniff sessions all pushing 115k full rate over 64 different ssh sessions), buy the Avocent ones are rock solid. Very expensive solution though, but they run Linux which make them cool. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Linux network behavior wierdness
Paul Lussier writes: Hi all, I have a wierd problem where some hosts respond to a broadcast ping packet and others don't. I have some hosts which, when I do a ping -b 10.95.0.0 everything answers. On other hosts, doing exactly the same thing, I get no response. A reboot resets a broken host, but over time it will re-develop the problem. And I can't seem to figure out how to make the problem occur... I can't figure out if it's something we're doing which is causing this change, or if it's a kernel thing where some threshold is reached and it just stops. Fwiw, we're running Debian/Etch with a 2.6.18 kernel. Most NICs are Intel e1000, though some are broadcom. In general, I've seen it happen across our lab on different hardware platforms with different motherboards and nics. Any insights would be most appreciated. 10.95.0.0 is an unusual broadcast address, how did you end up with that? Watch out for /proc/sys/net/ipv4/icmp_echo_ignore_broadcasts. The default was changed between 2.6.13 and 2.6.14 to ignore by default. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Linux network behavior wierdness
Paul Lussier writes: Paul Lussier [EMAIL PROTECTED] writes: Watch out for /proc/sys/net/ipv4/icmp_echo_ignore_broadcasts. The default was changed between 2.6.13 and 2.6.14 to ignore by default. Ooh, I didn't know that, thanks! Fascinating, so how does this work? $ ssh farm-519 ping -b 10.95.255.255 -c 1 64 bytes from 10.95.34.112: icmp_seq=1 ttl=64 time=0.074 ms 1 packets transmitted, 1 received, 0% packet loss, time 0ms $ ssh farm-519 cat /proc/sys/net/ipv4/icmp_echo_ignore_broadcasts 1 Ping can still send, just farm-519's kernel won't reply. I'd assume 10.95.34.112 is some other host on the network that does reply right? -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: How do you determine the amount of system memory?
Paul Lussier writes: Thomas Charron [EMAIL PROTECTED] writes: On 7/29/08, Paul Lussier [EMAIL PROTECTED] wrote: Thomas Charron [EMAIL PROTECTED] writes: Are you talking about a real bug, or the fact that meminfo only reports non kernel memory? A real bug. A bug in that /proc/meminfo doesn't report the amount of physical memory uder MemTotal? Yes, and that possibly, over time, the amount of memory in /proc/meminfo actually decreases. I'm assuming you mean Total memory not free memory :) If so, which total? Mem, High, Low, Swap, Vmalloc? Are you're saying the *Total lines change while the system is running or simply that MemTotal line isn't what you expect at boot time? I know there's an official, known-to-be-a-problem bug on this somewhere in the kernel bug tracking. I just don't know the exact details. I was given the task of figure out how we can test this. And, just being back from vacation, my mind is rather fuzzy on just about everything, including the time of day and day of week... -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: mythtv and digital tv
Derek Atkins writes: The biggest difference is that cable companies can choose to encrypt their QAM, which means you either need a cablebox or a device with a cablecard. As far as I know no device usable by mythtv directly can use a cablecard without having yet another Digital-Analog-Digital conversion. The 2nd biggest difference is cable companies tend to re-compress the heck out of their digital channels to squeeze as many channels into as little bandwidth as possible. Still, just 1-2 HDTV channels on cable is on the order of what is devoted to Internet bandwidth (10-12mbps). Imagine if the cable companies dropped all TV (both analog and digital) and did only broadband Internet over coax. Throw in some 20mbps+ IPTV steams via multicast and it'd give FIOS some competition. As mentioned before comcast at least does the broadcast channels QAM unencrypted in both SD and HD which my pcHDTV HD3000 grabs bit-for-bit off the wire just fine. Way better quality then having the SD cable box decode the QAM MPEG2 sent over horrible composite video/RCA audio (stupid cable box doesn't even have svideo) into my PVR250 which then re-encodes them back into MPEG2. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: laptop fails to hear replies
Lloyd Kvam writes: When my laptop boots, it is configured to use a dhcp server. For some time now, that processing has failed. I can run tcpdump -n ether host mac address and see the packets going out, but not the replies. At the dhcp server, the same tcpdump command shows the requests and replies. I can get the laptop network stack to function by manually assigning a valid IP address and then repeatedly pinging other computers on the network. The arp table is empty, showing incompletes for the mac addresses. Finally a ping will work and the arp table gets filled in and the networking functions are OK from then on. I can restart the network service and now dhcp works. (This behavior occurs at other sites, so it is not a switch or cable problem.) Since everything eventually works OK, I figure the hardware is good and I've fouled up some configuration item. I've tried enabling and disabling NetworkManager, but it does not seem to have any impact. Stopping iptables also has no effect. I'm hoping someone can suggest a debugging approach or possibly stuff I misconfigured to create this problem. Sounds like your switch(es) are running STP, LACP, and/or EAPoL. Any of these can disable your ethernet port on the switch side temporarially after link up. Those protocols will give up after 30-40 seconds and enable your port. LACP has a state where a newly link-up port can send into the network but not recieve from the network, however I think this only applies when a port is entereing into a LAG (this is a staging state called the 'collecting state' that is used to allow a new port to join into a pre-existing group without causing packet loss). How long is dhcp trying before giving up? You can control this in isc dhcp3 with the 'timeout' configurable in /etc/dhcp3/dhclient.conf you can see if your switches are causing problems by manually configuring the laptop, then bounce the link on the cable and see if you immediately recover or it takes time after each link bounce. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: 1Gb ramdisk in RHEL3
Michael ODonnell writes: OK, the ramdisk story is looking a bit better now. First, the tmpfs trick is definitely cool and worth remembering but isn't what we're looking for because some of what makes it cool also makes it unsuitable for our purposes. In particular, tmpfs has close ties to the kernel's buffer cache so files in tmpfs may actually be on backing store and since we're investigating a ramdisk solution to get some low-latency characteristics that makes tmpfs unsuitable for our purpose. Too bad, because rigging tmpfs is easier than rigging a ramdisk. The key to getting the [EMAIL PROTECTED]@!!! ramdisk to be usable was figuring out that ext2 and the ramdisk device driver had to agree on block size. The default for ramdisk is 1k and the default for ext2 is 4k. At first I tried to specify ramdisk_size=1048576 and ramdisk_blocksize=4096 on the kernel commandline (units are Kb for the former and bytes for the latter) and all would have been well. except that the machine in question boots from an initrd and the ext2 filesystem therein was already specified as using 1k blocks and the kernel panic'd when it failed to mount the initrd due to that mismatch. So, since I don't feel like rebuilding the initrd's filesystem with 4k blocksize I had the ramdisk fall back to the default 1k and told mkfs.ext2 to build the filesystem in the ramdisk using a 1k blocksize, thus: dd if=/dev/zero bs=1024k count=1048576 of=/dev/ram9 mkfs.ext2 -b 1024 /dev/ram9 mount /dev/ram9 /mnt/test If you're using a ramdisk for purposes other than booting, make sure you have backported this fix if you are running a kernel older than 2.6.24: if you don't, your ramdisk blocks will disapear once you use all the ram in the system and the kernel needs to free memory http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=5d0360ee96a5ef953dbea45873c2a8c87e77d59b -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: How do you determine the amount of system memory?
Paul Lussier writes: Hi all, Recent Linux kernels have had a minor bug in that the amount of memory reported in /proc/meminfo is incorrect. I'm trying to find a way to determine whether the amount reported is correct or not. I need some means of reliably knowing whether this value is accurate or not. Does anyone have any ideas? Physically looking is insufficient, given that I a) need to test 400+ systems, and b) I may need to run this test on boxes to which I have no physical access. one more way: $ grep 'System RAM' /proc/iomem basically the same info as from the e820, but in a nicer format: system 1 $ grep 'System RAM' /proc/iomem -0009f3ff : System RAM 0010-7fe4 : System RAM 1-107fffefff : System RAM system 2 $ grep 'System RAM' /proc/iomem -0009efff : System RAM 0010-efed : System RAM 1-10fff : System RAM system 3 $ grep 'System RAM' /proc/iomem -0009fbff : System RAM 0010-7ded : System RAM -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: How do you determine the amount of system memory?
Paul Lussier writes: Hi all, Recent Linux kernels have had a minor bug in that the amount of memory reported in /proc/meminfo is incorrect. I'm trying to find a way to determine whether the amount reported is correct or not. I need some means of reliably knowing whether this value is accurate or not. Does anyone have any ideas? Physically looking is insufficient, given that I a) need to test 400+ systems, and b) I may need to run this test on boxes to which I have no physical access. You can also just go to the source and read the SPD EEPROM off each DIMM. This is hit-or-miss depending on how well the i2c and eeprom kernel modules find your hardware, plus SPD addresses aren't guaranteed to be at the expected addresses (typically 0x50-0x5f) as that's up to the motherboard menufacturer. $ sensors eeprom-i2c-0-52 Adapter: SMBus I801 adapter at e8a0 Memory type:DDR2 SDRAM DIMM Memory size (MB): 2048 eeprom-i2c-0-50 Adapter: SMBus I801 adapter at e8a0 Memory type:DDR2 SDRAM DIMM Memory size (MB): 2048 -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: How do you determine the amount of system memory?
Stephen Ryan writes: On Wed, 2008-07-16 at 15:20 -0400, Paul Lussier wrote: Hi all, Recent Linux kernels have had a minor bug in that the amount of memory reported in /proc/meminfo is incorrect. I'm trying to find a way to determine whether the amount reported is correct or not. I need some means of reliably knowing whether this value is accurate or not. Does anyone have any ideas? Physically looking is insufficient, given that I a) need to test 400+ systems, and b) I may need to run this test on boxes to which I have no physical access. Thanks. dmidecode. Look for sections titled Memory Module Information. The e820 info from the BIOS can help too (provided your BIOS is bug-free), get it from dmesg or /var/log/dmesg You'll need account for memory both below 4GB and above 4GB PA. top of low memory is usually between 2GB and 3.75GB with the remaining memory starting at 4GB PA Below 4GB includes the low 640KB-1MB reserved, plus ACPI, SMM, and other reserved at the top of low memory, but from the 'usable' sections you can determine where the top of low mem boundry is (just round up to the next few MB in the last usable section): system 1 BIOS-provided physical RAM map: BIOS-e820: - 0009f400 (usable) BIOS-e820: 0009f400 - 000a (reserved) BIOS-e820: 000f - 0010 (reserved) BIOS-e820: 0010 - 7fe5 (usable) BIOS-e820: 7fe5 - 7fe58000 (ACPI data) BIOS-e820: 7fe58000 - 8000 (reserved) BIOS-e820: fec0 - fed0 (reserved) BIOS-e820: fee0 - fee1 (reserved) BIOS-e820: ffc0 - 0001 (reserved) BIOS-e820: 0001 - 00107000 (usable) below 4GB PA is from - 8000 ( 2GB) above 4GB PA is from 0001 - 00108000 (62GB) this system has 64GB RAM, split between physical addressses 0GB-2GB and 4GB-66GB system 2 BIOS-provided physical RAM map: BIOS-e820: - 0009f000 (usable) BIOS-e820: 0009f000 - 000a (reserved) BIOS-e820: 000f - 0010 (reserved) BIOS-e820: 0010 - efee (usable) BIOS-e820: efee - efee3000 (ACPI NVS) BIOS-e820: efee3000 - efef (ACPI data) BIOS-e820: efef - eff0 (reserved) BIOS-e820: f000 - f400 (reserved) BIOS-e820: fec0 - 0001 (reserved) BIOS-e820: 0001 - 00011000 (usable) below 4GB PA is from - f000 (3840MB) above 4GB PA is from 0001 - 00011000 ( 256MB) This system has 4GB RAM, split between physical adresses 0MB-3840MB and 4096MB-4352MB. system 3 BIOS-provided physical RAM map: BIOS-e820: - 0009fc00 (usable) BIOS-e820: 0009fc00 - 000a (reserved) BIOS-e820: 000f - 0010 (reserved) BIOS-e820: 0010 - 7dee (usable) BIOS-e820: 7dee - 7dee3000 (ACPI NVS) BIOS-e820: 7dee3000 - 7def (ACPI data) BIOS-e820: 7def - 7df0 (reserved) BIOS-e820: fec0 - 0001 (reserved) below 4GB PA is from - 8000 (2GB) above 4GB PA is no memory This system has 2GB RAM all of which are from 0MB-2GB PA -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: How do you determine the amount of system memory?
Dave Johnson writes: The e820 info from the BIOS can help too (provided your BIOS is bug-free), get it from dmesg or /var/log/dmesg system 3 BIOS-provided physical RAM map: BIOS-e820: - 0009fc00 (usable) BIOS-e820: 0009fc00 - 000a (reserved) BIOS-e820: 000f - 0010 (reserved) BIOS-e820: 0010 - 7dee (usable) BIOS-e820: 7dee - 7dee3000 (ACPI NVS) BIOS-e820: 7dee3000 - 7def (ACPI data) BIOS-e820: 7def - 7df0 (reserved) BIOS-e820: fec0 - 0001 (reserved) below 4GB PA is from - 8000 (2GB) above 4GB PA is no memory This system has 2GB RAM all of which are from 0MB-2GB PA one thing to watch out for is system 3 above has integrated graphics where the bios has stolen 32MB (7e00-8000) -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Alternatives to Comcast
Ben Scott writes: If what you see from your modem is PPPoE, then what happens is: Your PC builds an IP connection over PPP. It puts the PPP frames into Ethernet frames and sends them to the modem. The modem takes the PPP frames off Ethernet and puts them in ATM frames and beams them over the wire. At the DSLAM, the ATM frames go into whatever equipment Verizon has until they get to the IP provider. The IP provider takes the PPP frames out of the ATM frames and does IP routing with them. If what you see from your modem is IP-over-Ethernet (RFC-894), then the modem is either doing PPP termination locally and acting as a router, or it's encapsulating IP datagrams in ATM frames. (Some implementations are worse, and encapsulate the entire Ethernet frame in the ATM frames.) If the former, from the modem onward it's the same as for the PPPoE drill above. For the later, the encapsulated frames get to the IP provider, are dencapsulated, and then handled appropriately. Vitts Networks was one of the providers doing the worse Ethernet-over-ATM scenario I describe. Their DSL modem was basically just an Ethernet bridge. In the CO, they had boxes which basically took DSL on one side and spit Ethernet out the other. Then they'd patch each Ethernet port into a managed switch. So now instead of running PPP over the serial DSL, you're running an emulation of broadcast-based Ethernet, which actually had *higher* overhead than PPP would have. that sounds bad... For both VZ's PPPoE direct customers and those customers that have a different ISP, the dsl modem puts your packets directy into AAL5 ATM and send them over a VC to the CO (VPI/VCI 0/32 if I recall). for Bridged IPoE, that's IP over ETH over LLC/SNAP over AAL5 over ATM For anyone with a westell brand dsl modem on VZ you may be interested in a program I wrote to gather data from the modem. It sends the modem magic packets that cause it to send statistics data periodically to the LAN. The modem uses a multicast group to send these, I gather that it uses multicast packets because in a bridged configuration the device doesnt' have an IP address (at least not one that you are supposed to see from your LAN). http://davej.org/programs/westell.tgz The program can generate human or machine readable output peridically such as: $ westell --monitor --rate 10 Waiting for Packet Got packet on 2008-05-22-09:20:34: Status : online Uptime : 214D 20H 44M (since 2007-10-20-12:35:40) Upstream Signal/Noise Ratio : 12.0 db Power: 12.0 dbm Attenuation : 27.0 db Data Rate (kbps) : 864 atm, 782 aal5, 764 1500byte, 522 64byte Downstream Signal/Noise Ratio : 5.5 db [ -0.5 db ] Power: 16.5 dbm Attenuation : 47.0 db Data Rate (kbps) : 3360 atm, 3043 aal5, 2972 1500byte, 2029 64byte ATM Signal Lost : 0 Frame Lost : 0 FEC Errors : 0 CRC Errors : 5655 HEC Errors : 3068 OAM Cells:527 Loopback Cells : 0 TX Cells : 13172471[ 16616 cells, 705 kbps ] RX Cells : 14095746[ 543 cells, 23 kbps ] ETH Dropped Packets :150 TX Packets : 229991791[ 264 ] RX Packets : 249194133[ 524 ] I've tested the program on both my Bridged non-VZ-ISP connection and a PPPoE VZ-ISP connection of a friend. It seems to work on PPPoE however the data rate calculations for may be off due to the different encapsulation.. It's good to note VZ provisions the raw data rate higher than advertised so that once the LLC/SNAP header, AAL5 trailer and ATM cell header is removed, a 1500 byte IP packet can run at the advertised bandwidth. Unfortunately smaller packets suffer due to the 32 bytes of overhead plus sawtooth effect of ATM SARing. My 3mb downstream rate is only 2mb with 64 byte IP packets. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Alternatives to Comcast
Ben Scott writes: Where are you located? If we know where you're at, we might be able to make specific recommendations. In the Manchester area, give MV a call (http://www.mv.com). Around Haverhill or Amesbury (MA), try USAi (http://www.usai.net). For those in MA, I'd highly recommend AceDSL via Verizon's ATM cloud. They (along with many other ISPs) should be able to get you a line in most of Verizon's COs in eastern MA. http://www.acedsl.com/acedsl_coverage.php?state=MA I've had a dry line DSL (no voice) 3m/768k connection to them for almost 3 years. Pure IP from their network to the dsl bridge, none of this PPPoE garbage you'd get with Verizon as the ISP. The 3 or 4 times I've had to call them, got a real person right away that knew what they were doing. Oh and someone on their generic support email even knows what a PTR is without having to explain it and will change them to whatever you want. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Alternatives to Comcast
Jarod Wilson writes: I'm rather fond of my Verizon FiOS here in Tyngsboro, MA. No PPoE, PPoE, etc., just straight up fiber to the optical network terminal on the outside of the box, ethernet from there to my switch, off which I hang stuff, utilizing the 5 static IP addresses I've got. $99/mo, wide-open EULA, 20Mbps down, 5Mbps up (business-class package, but at my residence). Well of course if that was an option I'd switch :( -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
re: server uptime
Warren Luebkeman writes: I am curious how common it is for peoples servers to go extremely long periods of time without crashing/reboot. Our server, running Debian Sarge, which serves our email/web/backups/dns/etc has been running 733 days (two years) without a reboot. Its in an 4U IBM chassis with dual power supplies, which was old when we fired it up (PIII Server). Does anyone have similar uptime on their mission critical servers? Whats the longest uptime someone has had with Windows? I have a Sharp Zaurus SL-5500 PDA that's been accumulating a rather impressive uptime sitting unused in it's cradle. Just checked and it's now up to 1594 days, however the openzaurus kernel it's running has a 32bit jiffies counter so it's wrapped it's uptime 3 times so far and only shows 103 days, the other 3*497 days are there but hidden :( It's survived many power outages by simply auto-suspending itself if power is lost, just resume back on after the outage and uptime picks up where it left off. I think there's probably another month of missed uptime due to forgetting to resume it after power outages. If only I could find a more useful purpose for it besides accumulating uptime. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: gcc linking / libtools / building stuff
Bruce Labitt writes: So I tried to just build it from the command line and this is what I got. $ gcc main_tb_fftw.cpp -otbfft -I/usr/local/include -L/usr/local/lib -lfftw3 -lm /tmp/ccJMISfw.o: In function `__static_initialization_and_destruction_0(int, int)': main_tb_fftw.cpp:(.text+0x23): undefined reference to `std::ios_base::Init::Init()' /tmp/ccJMISfw.o: In function `__tcf_0': main_tb_fftw.cpp:(.text+0x6c): undefined reference to `std::ios_base::Init::~Init()' /tmp/ccJMISfw.o:(.eh_frame+0x12): undefined reference to `__gxx_personality_v0' collect2: ld returned 1 exit status So what does all that mean??? The program did compile and link using Dev-C++ (gcc) on windoze. It looks like it can't find iostream ? I have that included in main_tb_fftw. Thanks! Add '-lstdc++'. Might also be more correct to invoke 'g++' instaed of 'gcc'. gcc will pick up on the language by the filename extension, but if you did compile and link on two different commands, you'd need to invoke g++ for the link stage instead of gcc. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Motherboard capable of supporting over 32 GB RAM
Dan Jenkins writes: Does anyone have any recommendations? Preferably with DDR-800 support. It'll run both Linux and Windows XP 64 and is used for simulations. Thanks. You're probably in for a Xeon 5xxx or Opteron 2xx or 2xxx board if you want this much RAM. Check Tyan, Supermicro, or for full systems Dell, HP, IBM. You can probably pick up an opteron 2xx system for relatively cheap if you don't need the latest and greatest system. See here: http://www.supermicro.com/products/motherboard/Xeon1333/ http://www.supermicro.com/Aplus/motherboard/Opteron/Op200.cfm http://www.supermicro.com/Aplus/motherboard/Opteron2000/ http://www.tyan.com/product_board_list.aspx?cpuid=1socketid=10chipsetid=9 http://www.tyan.com/product_board_list.aspx?cpuid=4socketid=9chipsetid=9 http://www.tyan.com/product_board_list.aspx?cpuid=4socketid=16chipsetid=9 Most of those are extended ATX or custom sized though. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: sshd config problem.
Steven W. Orr writes: For reasons that are probably very boring, I want to make a change to my sshd config. I have two interfaces, eth0 and eth1. I had previously modified my listening port from 22 to something with a couple of extra digits for the kiddys. Now I have a situation where I want to maintain that setup only on eth0 (a.b.c.d) but I want to make it listen on port 22 for eth1 (192.168.0.2). I tried ListenAddress 192.168.0.2:22 ListenAddress 207.172.210.41:1234 but I must be doing something wrong. Anyone? That should work, but you should also remove any 'Port' config lines in the config file for the ListenAddress to take effect, expecially if Port is below the ListenAddress lines. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Success stories with MythTV and Schedule Direct?
Ted Roche writes: Just checking in to find out if anyone has switched their MythTV setups over to Schedules Direct [1]? (Schedules Direct is a non-profit organization that provides raw U.S./Canadian tv listing data to Free and Open Source Applications. Those applications then use the data to provide things like PVR functionality, search tools, and private channel grids.) Singed up friday, and then went through the painful process of upgrading 1 backend and 2 frontends from 0.19 to 0.20.2.. The prepackaged myth binaries required tons of upgrades as usual (Debian sarge to etch plus a few dozen unstable packages) breaking lots of minor things like squid, cups, nis, nfs (nfs sure breaks in suttle ways when rpc.statd just isn't started for some reason post-upgrade) Schedules Direct seems to work just fine. The upgrade of mythtv itself wasn't bad but now I have to get used to a different set of bugs... Only 3 misterious backend crashes so far :( -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: [OT] Charging UPS batteries outside the UPS
Ben Scott writes: Now, I don't want to have to spend $350 on a replacement battery only to find out that it's the UPS itself that's shot. I'm thinking if I can find some way to charge up the battery to minimum levels, I can at least test the UPS to see if it works. It doesn't have to hold a load. $350 is a bit high.. I've bought replacement batteries from here many times without issues: http://www.powersupersite.com/MZIproducts.cfm?full=1ID=RBK12-f 175.00 + ~25 shipping for the RBC12 The battery consists of eight smaller units, wired together. The wiring is easily disconnected. Each unit is labeled 12 V, 7.2 Ah/20HR. Anyone if I can just hook each unit up to a regular automotive battery charger (one at a time) and charge them that way? I'd say probably so, but dont blame me if they explode. useful info here: http://en.wikipedia.org/wiki/Lead-acid_battery -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: [OT] Charging UPS batteries outside the UPS
[EMAIL PROTECTED] writes: First, confirm that these batteries are wired IN PARALLEL, NOT IN SERIES. (If they are in series, you're looking at a highly non-standard 96V implementation. No problem, though... just disconnect them all and charge them separately.) This ups takes 2 sets of 2-parallel 2-series providing 2 24V plugs with 4 batteries each. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Excessive processor usage
Sean writes: I am beginning to think it is my drive as the problem, it seems to be writing a great deal when I see the problem. Time for backups! try 'vmstat 5' -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: [OT] Network In Flight data sizes
Star writes: Does anyone know of a tool that will help determine Internet-application performance throughput for overall data window size? My company has a client that depends on a hosted application. While only one of their offices used this app, things worked very well for them. Now that they're rolling it globally, they're noticing significant slowdowns in certain areas. We already use Akamai and some fairly extreme caching settings to keep the dead-bits to a minimum, but the dynamic parts are showing some trouble. Essentially, they're telling us that they're seeing choke points in the 8k range for throughput. We've gone through all of our equipment and assured that we're using 64k windows sizes on the send and receive sides. Still they see this. It's one of those it must be on your end discussions and we're working hard to get all the data that they request, but it's hard to quantify this in flight number that they keep touting and showing pretty graphs of. The tool that the client's group is using is Opnet IT Guru/ACE which is a fine tool... but if I can get a req for $50k for software in under 6 months, Hell may have a need for those double hockey-sticks... Any advice is much appreciated. ~ Star FYI: linux should be using window size scaling to get windows larger than 64k (there's a sysctl to turn this off, but I think it's on by default). But more fundamentilly, a tcpdump/ethereal/etc.. will easially tell you the amount of data outstanding, and netstat/etc... will tell you the fullness of socket buffers. If that isn't helping you should get a capture on both sides so you can compare to make sure there isn't any proxy or evil router/firewall mucking with your tcp stream (i.e. reassemble, then add un-QoS, jitter, backpressure, shaping, random drops, etc...) Any retransmitions, SACKs, etc.. are a sign of packet loss, if you see flags or window sizes showing up magically changing from when the same packet was sent then some intermidiate device is changing them. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: How to open a device for exclusive access?
Ben Scott writes: On 7/15/07, Steven W. Orr [EMAIL PROTECTED] wrote: ... my particular problem would be solved if the char device I was trying to open would only honor O_EXCL. My C is somewhat rusty, but doesn't O_EXCL just prevent one from creating/opening the file if it already exists? For regular files yes, but char and block drivers can do whatever they want with the open flags. Some drivers use O_EXCL for exclusive access to a device. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: How to open a device for exclusive access?
Steven W. Orr writes: I have a device (it happens to be a somewhat exotic serial port) which is managed by a server process. I want my server to detect whether another instance of that server already has that device open. I'm guaranteed that no other program other that the one I'm in control of is capable of opening that particular device. I was hoping that I could maybe not have to use either mandatory or advisory file locking. I have tried O_EXCL | O_RDWR | O_NONBLOCK when opening the device node hoping to get back EBUSY or EAGAIN. Does anyone know if I'm SOOL or is there a way to do it? Block devices support O_EXCL for exclusive access (something that fdisk can make use of to prevent mounting of a partition while it has the device open). Since char devices are more generic, it's up to the individual driver to either support or not support O_EXCL on open. The driver can also support exclusive opens all the time if it wants. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: OT: PC Gigabit Throughput Question
Most have already covered the PCI 33/32 bandwidth limits (avoid PCI 33/32 if you can). Integrated into the chipset, PCIX bus (faster but still shared), or PCIe (dedicated) will solve the bandwidth problems. Just about any PCIX/PCIe/Integrated MAC will give you 100% gigabit full duplex line rate provided the driver and memory controller can keep up. Look for a useful TOE/TSO/GSO hardware engine as well as hardware TCP/UDP checksumming on both RX and TX. Hardware checksumming will reduce the cpu power requirements dramatically, a segment engine will reduce the number of packets and give you the equivelant of jumbo frames (from the driver's point of view) without actually using jumbo frames on the network . Which brings me to the next point. A bad driver will make a huge difference and may be a limiting factor especially at 500mbps. A good driver will make use of NAPI for overload conditions, and interrupt coalescing for moderate load conditions. A segment engine isn't likely on a PCI33/32 MAC, but some do have checksumming. One other neat feature I've seen on nvidia integrated MACs is it makes use of MSI-X to offer 3 different interrupts for each MAC. One for TX service, one for RX service, and the 3rd for phy events. This allows the driver to have completely separate RX/TX/misc code paths. Very cool. Anyway, I'd recommend Intel (e1000) or Broadcom (tg3, bnx2) MACs. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: dual-core vs. HyperThreading
Thomas Charron writes: On 4/25/07, Michael ODonnell [EMAIL PROTECTED] wrote: I can definitely confirm that my core 2 quad system has no HT available to turn on in the bios (but does show ht in the cpuflags, which aligns with other comments in the thread). I've heard it suggested several times now that features like HyperThreading can be enabled or disabled in the BIOS and I don't understand how that can be possible. AFAIK a CPU either does or doesn't have a given feature and there's nothing (short of modifying the CPU's microcode) that any BIOS can do to disable it, at least not in any way that would prevent a capable OS from doing as it pleased once it was booted. So, what's the story? I don't know the specifics, but the BIOS has to 'turn the bits on'. There are thousands of Toshiba laptop owners who wished Toshiba would enable the VT bit in their damned BIOS screens for their Intel Core 2 Duo chips. Unfortunately this is a feature. CPUs, Chipsets, etc.. usually have things like this in Write-Once or Write-Lock registers. After reset, the register can we written exactly once or can be set to lock in the values until the next reset. BIOS code makes sure to write all of these registers once to enable/disable features as it sees fit. This is usually good as typically they are uses for board-specific enable/disables. It makes no sense to enable a PCI device or bridge if there is nothing connected to it on the board as an example. Another use is for features that can only be enabled or modified early in the boot process before memory is configured or IO devices setup, making changes afterwards would give unpredictable results. Locking these down makes sense as any changes could lock up the system. Turning HT on/off on the fly seems unlikely to work at the OS level as I'm sure there is some initializations needed that aren't trivial or likely known to the OS. Same goes for VT. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: slow last 128MB of RAM in a 2GB system?
Ben Scott writes: On 4/20/07, Bill McGonigle [EMAIL PROTECTED] wrote: Ouch - is that simply a matter of cache impact on performance? I wouldn't have guessed it would be so high. It could be (although it could be something else I'm just not aware of). 50x isn't too bad. I'd expect you would end up with different results depending on which memory happends to end up in that last 32MB. The highmem area can be disk cache, userspace code, and userspace data, but not kernel code or kernel lowmem allocations. If you were to run a bunch of programs first then the test program you may end up with the test program in cached memory, but if some critical disk blocks end up uncached (say the test program's code segment, or even worse an important part of libc) then it'll crawl. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Dell 690 only seeing 3 GB RAM (was: slow last 128MB...)
Ben Scott writes: I'm not sure if disabling MTRR is a good idea or not. It may be important for other reasons, and sometimes disabling something that is causing a problem just leads to bigger problems down the road. Perhaps Dave or someone else who understands it better can comment on that aspect? If by disabling you mean CONFIG_MTRR, all that will do is remove /proc/mtrr and prevent video drivers from modifying MTRR. MTRR is still present in the CPU, linux just won't touch it. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: slow last 128MB of RAM in a 2GB system?
Bill McGonigle writes: Into -h units: 639K - BIOS-e820: - 0009fc00 (usable) 1K- BIOS-e820: 0009fc00 - 000a (reserved) 128K - BIOS-e820: 000e - 0010 (reserved) 2014M - BIOS-e820: 0010 - 7dfd (usable) 60K - BIOS-e820: 7dfd - 7dfdf000 (ACPI data) 132K - BIOS-e820: 7dfdf000 - 7e00 (ACPI NVS) 4K- BIOS-e820: fee0 - fee01000 (reserved) 8M- BIOS-e820: ff78 - 0001 (reserved) 1919M - limit_regions endfor: 0010 - 7800 (usable) 0MB to 2016MB is basically RAM. The remaining 32MB isn't listed in the e820 map. It's likely the 32MB is being reserved for the integrated graphics. Now this will cause a challenge to the BIOS's MTRR logic. It will want to cover that entire 2016MB exactly with the MTRR registers. The MTRR registers allow you to specify a base address and size, but the size must be a power of two. There are 2 basic ways BIOSs can do this: 1) It can list the cacheable memory using power of 2's. That would be: 1024+512+256+128+64+32. reg00: base=0x ( 0MB), size=1024MB: write-back, count=1 reg01: base=0x4000 (1024MB), size= 512MB: write-back, count=1 reg02: base=0x6000 (1536MB), size= 256MB: write-back, count=1 reg03: base=0x7000 (1792MB), size= 128MB: write-back, count=1 reg04: base=0x7800 (1920MB), size= 64MB: write-back, count=1 reg05: base=0x7C00 (1984MB), size= 32MB: write-back, count=1 2) It can list all 2048MB as cachable, then work backwards for the uncachable part by doing: reg00: base=0x ( 0MB), size=2048MB: write-back, count=1 reg01: base=0x7e00 (2016MB), size= 32MB: uncachable, count=1 Also, note it is customary to leave 1 or 2 MTRR registers available for the OS to use. On intel cpus, there are 6 MTRR registers so the BIOS will tend to only use 4. Your BIOS is using #1 above, prefering to list out each cachable part. When you had 1GB of RAM, you needed 4 MTRR registers to fully express 992MB, but with 2GB you need 5 MTRR registers to fully express 2016MB. Using #2 method is much more preferable for the case were some amount of memory at the end of RAM is reserved. With this you only need 2 MTRR registers no matter what the size of RAM. You can easially fix this using /proc/mtrr. Since the BIOS filled out the first 4 entries in #1 above, you can simply add the 5th one yourself. You could also convert the table into #2 above, but unless you plan to run Xorg you dont need any more MTRR registers. Simply add this to a startup script and you should be all set: echo base=0x7c00 size=0x200 type=write-back /proc/mtrr After that the output should look exactly like #1 above. While I feel it should be the manufacturer's responsibility to get this right, I think linux would be stronger for knowing how to deal with as much buggy BIOS as is out there. I'll run bug reports up the flagpole as necessary if you can set me on the right track. Good luck. aside: Is it just more or is this on the uptick? - I'm seeing 'buggy BIOS' as the cause of so many problems recently. I've been through nasty problems with nVidia, ATI, and SiS BIOS bugs, all in the past year, where I haven't worried much about it previously. Intel chipsets are, so far, batting 1000 recently. Yes, my ASUS M2N-SLI shipped with a buggy BIOS where it wouldn't place any memory over 4GB, limiting usable memory to 3.75GB. Upgrading the bios to the latest version fixed that, but broke ACPI. -- Dave ___ gnhlug-discuss mailing list [EMAIL PROTECTED] http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: slow last 128MB of RAM in a 2GB system?
Dave Johnson writes: Using #2 method is much more preferable for the case were some amount of memory at the end of RAM is reserved. With this you only need 2 MTRR registers no matter what the size of RAM. Make that: provided size of RAM is a power of 2 (512MB, 1GB, 2GB etc..). 1.5GB would require 1 more MTRR. Yes, my ASUS M2N-SLI shipped with a buggy BIOS where it wouldn't place any memory over 4GB, limiting usable memory to 3.75GB. Upgrading the bios to the latest version fixed that, but broke ACPI. Woops, I mean it broke the IOAPIC, not ACPI. I always get those confused. -- Dave ___ gnhlug-discuss mailing list [EMAIL PROTECTED] http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: slow last 128MB of RAM in a 2GB system?
Bill McGonigle writes: Long story short: I put 2GB in my soho server and if I use the last 128MB or RAM or so the machine is murderously slow. If I boot the machine mem=1920M it's just fine. Some googling around has lead me to look at linux himem handling and mtrr cache stuff, but I'm not really sure what's going on. 1920MB is practically just fine, but I can't leave well enough alone... $cat /proc/mtrr reg00: base=0x ( 0MB), size=1024MB: write-back, count=1 reg01: base=0x4000 (1024MB), size= 512MB: write-back, count=1 reg02: base=0x6000 (1536MB), size= 256MB: write-back, count=1 reg03: base=0x7000 (1792MB), size= 128MB: write-back, count=1 reg04: base=0x7800 (1920MB), size= 64MB: write-back, count=1 I believe this is telling me that only the first 1920MB of RAM is being cached. But I'm not sure why or what's happening to the last 128MB. 32MB of that last 128MB is allocated to the integrated graphics controller. Maybe the last 128MB is mapped to the graphics controller anyway? But then why doesn't linux detect this? Also, this wasn't a problem with 1GB in the system. Documentation/mtrr.txt hasn't been particularly helpful for me, and Googling has gotten me a fix, but not much in the way of theory (other than MTRR is an old nasty x86 hack and linux ought to dump MTRR on modern hardware and use PAT like Windows does - whatever that is). Has anybody figured this out before? I unfortunately know way too much about these types of problems having debugged and fixed a commercial BIOS W.R.T. to MTRR and E820 bugs. The MTRR map is (by default) filled out by the BIOS. It also fills out the E820 memory map. This first question is what did the bios fill out for the E820 map? This is printed at the begging of the kernel startup with the text 'BIOS-provided physical RAM map'. Also, what is the output of /proc/iomem and 'lspci -vv' (run as root please). The BIOS may be providing a memory break at 1984MB with the remaining 64MB at 4GB and forgetting to add the MTRR entry for that, or it may be reserving that 64MB for use by ACPI/SMM/etc... and forgetting to mark those regions reserved in the E820 map. Either way it's usually a BIOS bug, but linux should be able to work around either case (though that might mean some kernel changes). -- Dave ___ gnhlug-discuss mailing list [EMAIL PROTECTED] http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: SSH to one address, different ports, different hosts
Ben Scott writes: Anyone else have thoughts or ideas to offer? ssh client doesn't mind if there is more than one entry for a given host in the known_hosts file. Because of this you can simply manually edit the known_hosts file to have multiple entries (one for each actual host) all with the same hostname. ssh client won't do this automatically, but once you know the public keys for each host you can then edit the file and add all of them. Once you add them all it will accept any of them for that hostname. Example: some.host.com ssh-rsa KEY-TEXT-FOR-HOST-1. some.host.com ssh-rsa KEY-TEXT-FOR-HOST-2. You can then ssh to some.host.com on some port and ssh client will accept EITHER key listed in the file. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: DHCP question: dhclient won't request same IP.
Scott Garman writes: To test things further, I placed both machines behind a Linux DHCP server so I could look for any unusual error messages. Sure enough, I found the following in my logs after the Debian system exhibited the IP-changing behavior. The Debian system started out with IP 192.168.1.215, and ended up with 192.168.1.210: Jan 31 12:32:03 dhcpd: DHCPDISCOVER from 00:d0:c9:9e:ae:59 via eth1 Jan 31 12:32:03 dhcpd: ICMP Echo reply while lease 192.168.1.215 valid. Jan 31 12:32:03 dhcpd: Abandoning IP address 192.168.1.215: pinged before offer Jan 31 12:32:03 dhcpd: Wrote 0 deleted host decls to leases file. Jan 31 12:32:03 dhcpd: Wrote 0 new dynamic host decls to leases file. Jan 31 12:32:03 dhcpd: Wrote 9 leases to leases file. Jan 31 12:32:05 dhcpd: DHCPDISCOVER from 00:d0:c9:9e:ae:59 via eth1 Jan 31 12:32:06 dhcpd: DHCPOFFER on 192.168.1.210 to 00:d0:c9:9e:ae:59 via eth1 Jan 31 12:32:08 dhcpd: DHCPREQUEST for 192.168.1.210 (192.168.1.1) from 00:d0:c9:9e:ae:59 via eth1 Jan 31 12:32:08 dhcpd: DHCPACK on 192.168.1.210 to 00:d0:c9:9e:ae:59 via eth1 It appears the client is sending a DHCPDISCOVER while still using the old address (the server can ping it). For a renewal of an existing lease the client should first send a DHCPREQUEST instead of DHCPDISCOVER. Only if the DHCPREQUEST is ignored/lost should it send a DHCPDISCOVER in an attempt to find a new lease. The server is doing the right thing in this case. If the server recieved a DHCPREQUEST first it would extend the existing lease. The logs from the client computer may reveal the requests and if it is trying a REQUEST first with no response. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: DHCP question: dhclient won't request same IP.
Dave Johnson writes: It appears the client is sending a DHCPDISCOVER while still using the old address (the server can ping it). For a renewal of an existing lease the client should first send a DHCPREQUEST instead of DHCPDISCOVER. Only if the DHCPREQUEST is ignored/lost should it send a DHCPDISCOVER in an attempt to find a new lease. The server is doing the right thing in this case. If the server recieved a DHCPREQUEST first it would extend the existing lease. The logs from the client computer may reveal the requests and if it is trying a REQUEST first with no response. Also, one thing to note. I've found good luck with 'dhcp3-client' instead of the older v2 versions or other clients such as pump. Under debian, you can: apt-get install dhcp3-client -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: PVR-150 (1042, 1045, 1062)?
FYI: Rumor has it Hauppauge has been putting HVR-1600 into PVR-150 boxes even though the box still says PVR-150. See the most recent comments here: http://www.newegg.com/Product/Product.asp?Item=N82E16815116625 -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Linux, gobs of RAM, RAID and performance suckage...
Bill McGonigle writes: I've forgotten some 2.4 stuff but there was a big-mem version of the 2.4 kernel at one point to work around problems with too much RAM. Ah, that's right. If you're going to run 2.4 with 1GB RAM you need to apply the rmap patches or performance will get worse the more RAM you have. Severely so with 2GB. Google says rmap patches are here, but the server is giving me a 403 at the moment. http://www.surriel.com/patches/ -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Linux, gobs of RAM, RAID and performance suckage...
Paul Lussier writes: This is bizarre. We've got an NFS server with Dual 3Ghz Xeon CPUs as our NFS server connected to a Winchester OpenSAN FC-based RAID array. The array is a single 1TB partition (unfortunately). Before yesterday we were noticing lots of NFS drop-outs on the clients (300+ of them) and we correllated this pretty much to the backups (amanda). The theory was that local disk I/O was beating out nfs-client requests. We also noticed that our memory utilization was through the roof. We had 2GB of PC2300, ECC, DDR, Registered RAM. That RAM was averaging the following usage patterns: active - 532M inactive - 1.2G unused - 39M cache - 1.3G slab cache - 255M swap cache - 6M apps - 78M buffers- 350M swap - 11M We were topping out our memory usage and occasionally dipping into swap space on disk. Yesterday we added 2GB of RAM and our memory utilization now looks like this: active - 793M inactive - 2.3G unused - 213M cache - 2.9G slab cache - 194M swap cache - 2M apps - 71M buffers- 313M swap - 4.5M So, it appears we really only succeeded in doubling the cache available to the system, and a little more than halving the amount of swap that was getting touched. However, now when backups are run, the system becomes completely unresponsive from an NFS client perspective, and the load average skyrockets (e.g. into the 40s!). Does anyone have any ideas ? I'm at a complete loss on this one. General nfs server comments: 1) Make sure you are exporting the nfs shares async otherwise most operations will seem slow from the clients point of view. Check /proc/fs/nfs/exports for 'async' If not there, set it in your /etc/exports 2) Make sure you server has enough nfsd threads to handle the client load. With 300 clients, you should have at least 100 nfsd threads in my opinion. Check your init.d scripts for how to set this. Other stuff: 11MB of swap doesn't mean anything is wrong. It's actually a good thing meaning your kernel found 11MB of stuff that wasn't needed and booted it out to swap space in order to make more room available for cache. You should check the disk rates and cpu usage with 'vmstat 5' for a few minutes. This will also show how much of the cpus are spending in wait. The full outout /proc/meminfo may also be useful. I assume you are using a x86_64 kernel? Using a 32bit kernel is ok as long as you dont run out of low memory. The kernel's heap/slab as well as filesystem metadata (buffers) must come from low memory while filesystem data (cache) and userspace processes can be in high memory. If your SAN is on a dedicated lan to the server only you should investigate converting that subnet to support jumbo frames. Since your using Xeons not Opterons you should make sure irqbalance is installed and running to spread irq load across all cpu cores (this may not be a good idea on a multi-node Opteron system though) You can run top in multi-cpu mode (press '1') to see if any cpus are overloaded with irq or wait load while others are idle. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Linux, gobs of RAM, RAID and performance suckage...
Paul Lussier writes: Yesterday we added 2GB of RAM and our memory utilization now looks like this: active - 793M inactive - 2.3G unused - 213M cache - 2.9G slab cache - 194M swap cache - 2M apps - 71M buffers- 313M swap - 4.5M When you are in this condition, how large is Dirty and Writeback in /proc/meminfo? If you have a large amount of Dirty (read: 300MB) and the underlying filesystem is using ordered journaling (the default for ext3) you can cause very long delays if a process requests a fsync on a file (vi does this alot). Instead of writing a single file to disk right away the filesystem must write all other outstanding data to disk first! If your clients are write heavy and you are using ext3 you should consider data=writeback when mounting the filesystem. I've run into this issue on a server with 64GB of RAM with dirty commonly exceeding 4GB under load. Poor users of vi had to keep waiting for the server to write out 4GB of data every time they saved a file. data=writeback helped tremendously as only the previous metadata is forced to disk on a fsync() instead of everything. Correct. The nfsd's on the server get pushed to the bottom of the queue (in top) and what I see is [kupdated] and tar (from the backups) rotating between the top position. There was one or two other processes up there too, one of which I think was a kernel thread ([bdflush] ?). As far a I know, the tar's were not making processes, but I can't be totally sure of that. Nothing *else* was making progress, that's for sure. kupdated?!?! A 2.4 kernel??!?!? -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Kill-a-watt devices
Tom Buskey writes: $33 http://www.invertersrus.com/killawatt.html $21 http://www.sustainabilitysystems.com/products.php?cat=8 Measures: # Cumulative kWh over time (using built in timer) # Watts (active power) # VA (apparent power) # Volts # Current (amps) # Frequency (hertz) # Power Factor 1875VA limit 15 A limit. I googled a review that said it showed 0.06 kWh after 2 hours for a TV. I'd imagine the utility co's unit has a higher amp rating. I have a kill-a-watt too. The only thing (besides a computer interface) I don't like about it is the lack of battery backed or otherwise persistant cumulative usage. Every time you unplug it or power goes out the totals are lost, but that's what you get for the price. There are other alterntives but a bit more expensive. The 'Watts Up' and 'Watts Up Pro' go for about $100-$200 but have more features including persistant logging. See: https://www.doubleed.com/products.html -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: MythTV hw question
Derek Atkins writes: Some VGA outs can only get up to 1080i and can't get the throughput of 1080p, especially since I'm asking for nearly 995328000 B/s (unless I've miscalulated 1920 x 1080 x 4 x 120... That's the first issue, since these boards use main memory for the video you'll need high bandwidth RAM to feed the chipset. 1080p is only about 8MB so size isn't an issue bandwidth may be. The second issue is the max dot clock the chipset can drive out the VGA port. My CN400 shows: (II) VIA(0): Clock range: 20.00 to 230.00 MHz That clock must include not just the visible picture but the sync and blanking after each line needed by the actual display device on the other end. Some searching found these lines for 60hz: Modeline 1920x1080i60 74.25 1920 1976 2008 2200 1080 1083 1085 1125 +Hsync +Vsync Interlace Modeline 1920x1080p60 148.50 1920 1976 2008 2200 1080 1083 1085 1125 +Hsync +Vsync 148Mhz clock is needed for [EMAIL PROTECTED], [EMAIL PROTECTED] would require a 297Mhz clock which is beyond the VIA chipset's limit of 230Mhz. There is also good info here: http://www.linuxis.us/linux/media/howto/linux-htpc/video_card_configuration.html I plugged in the [EMAIL PROTECTED] line and got this: (II) VIA(0): ViaValidMode: Validating 1920x1080 (148500) (II) VIA(0): ViaModePrimaryVGAValid (II) VIA(0): Required bandwidth is not available. (497664000 46100) The clock was ok, but memory bandwidth was insufficent. This system has a PC2700 DIMM, a PC3200 DIMM would probably be enough and DDR2 would be fine too. The X driver actually measures the memory bandwidth on startup. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: MythTV hw question
Not sure about the CN700, but the CN400 on my SP13000 can do 720p however it isn't one of the standard resolutions the X driver will offer. You need to add a modeline to the XF86 config and it'll then output it just fine. Last I checked Unichrome Pro was still not supported in the standard XF86 or X.org distributions. Compiling the external unichrome pro X driver from CVS is quite a pain and was still a bit buggy last time I upgraded. This recent discussion prompted me to finish up this page today, something I've been meaning to finish for a few months. It has just about everything for my MythTV frontend computer. http://centerclick.org/yukon/ -- Dave Derek Atkins writes: They don't specify the VGA output resolution, and I can't seem to find that specification anywhere. I even downloaded their PDF Manual and that info isn't in it. -derek Thomas Charron [EMAIL PROTECTED] writes: http://www.via.com.tw/en/products/mainboards/mini_itx/epia_en/ They've specifically built the board above for HD signal processing, as well as 5.1 sound. Thomas ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: MythTV hw question
Paul Lussier writes: So, before I start spending a ton of money on new hardware, I'd like to play around with MythTV a little. I have an old PIII 300Mhz system collecting dust somewhere. Would that be sufficient to play with if I threw a little memory and a new PVR-350 at it? I don't expect any great performance from it, and would expect to migrate the video card over to a new system once I figure out what I was doing. That's probably fine for a backend, but may be a bit slow for decoding for the frontend. I've noticed cpu power requirements for playing are perportional to the bitrate of the MPEG2 so you can always crank down the bitrate setting when recording to find what it can handle. At least on my PVR-250 you can set any width/height/bitrate you want and it'll do MPEG2 hardware encoding on that. You should have no problems cranking them way down just for the experimentation phase. At least that'll give you time to play around with the functionality even if the video quality isn't great. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Tivo vs MythTV (was: *pout* HDTV No Recordee....)
Ben Scott writes: Those of you here who are already using MythTV, how do you find it works in day-to-day usage? I've been using MythTV for about 2 years or so. I like it alot better than the tivo I had before that. It has it's bugs, but they are usually minor and can be worked around without too much trouble. I really like the split model [backend + frontend(s)] which allows me to have a fanless, diskless, no-moving-parts frontend computer for the TV and a big server in the garage^H^H^H^H^H^H^H server-room. Note you can get unencrypted HD off of Cable in addition to OTA. My MythTV backend has a Hauppauge WinTV-PVR 250 connected to the comcrap set-top box and a pcHDTV HD-3000 connected direct to the comcrap coax. The latter can do direct digital captures of all unencrypted channels (SD and HD) with no conversion or recompression (direct MPEG2 to disk). -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Tivo vs MythTV (was: *pout* HDTV No Recordee....)
Ben Scott writes: Note you can get unencrypted HD off of Cable in addition to OTA. Any idea how many unencrypted HD channels there are? Every broadcast channel. In my case that's the HD versions of: 2, 4, 5, 7, 9, 25, 38 last I checked In addition, comcast sends 2 version of the analog channels over the coax. An analog version for for people without set-top boxes and a digital SD version for people with digital set-top boxes. The set-top box actually uses the digital version. When you tell it to tune to say channel 5, it picks up the digital version of channel 5 instead of the analog version. This eliminates the 'fuzzy' problems you get with low signal strength or interference. The pcHDTV card can get the unencrypted SD digital version too. This allows me to avoid the digital [coax] - set-top box - analog [composite] - capture card - digital conversion of the Hauppauge Card. MythTV only uses the Hauppauge card if there is a conflict or for channels that are encrypted. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Tivo vs MythTV
Paul Lussier writes: Dave Johnson [EMAIL PROTECTED] writes: You can also just plug in the coax to a HDTV with a digital tuner and get the same result. Hmm, I didn't know that. How do I find the HD channels? When I tune my TV to the channel number advertised by Comcast, I get a black screen. Presumably that number is only meaningful to their box. Sorry, should have said 'have the same issue' instead of 'get the same result'. Without a cable card (that's a differnt issue) you have the same channel numbering problem. Remember that ATSC and QAM both allow multiple subchannels per frequency band. You can get about 2 HD and about 12 SD channels per frequency band. At least on my HDTV (Samsung), channels are represented by channel-subchannel so PBS is on 87-1 or something like that even though comcast maps it as 802. Something else is on 87-2 and so forth. Just doing a full scan in the TV's tuner setup will pick up everything (including all the encrypted channels, but those are blank). -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Getting started w/ MythTV [was Re: Tivo vs MythTV]
Paul Lussier writes: So, for these 2 systems, can someone give me a rough list of the essential-to-have hardware and the cost analysis? My main goal here is ease of installation/configuration and minimizing as much as possible hardware incompatibility frustration. I.e., if there's a certain MoBo I ought to use or specific hardware combination that just works, tell me. Cost savings *is* an issue, but time is more of an issue. I don't want to fight with hardware, I want to build an appliance that will do it's job with as little interaction from me as possible :) For analog cards, getting one with MPEG2 hardware compression is a very good idea. Besides sufficient bandwidth on your PCI bus and disk controller/disks recording uses very little cpu (read: 2% on a any modern cpu). The bandwidth requirements aren't that big. SD should run you about 2-5Mbps, HD about 10-20Mbps. (that's bits not bytes) As long as you can sustain the average and not have a congested system that could cause DMA underruns any motherboard and cpu should be fine. Note that mythtv, at least IMHO, is quite bloated and the code way too complicated and confusing to look at. If you have alot of channels and capture cards the backend database can get quite large. Keeping all that in RAM will improve the performance of the frontend and mythweb especially when getting channel listings or doing searches. My /var/lib/mysql/mythconverg/ is about 400MB. The more memory for the backend the better especially because it will be doing disk caching. My frontend is a VIA EPIA SP13000 with 512MB RAM, PXE boots with nfsroot. It's plenty fast for SD, but barely fast enough for HD. I think this is actually a video dirver problem though. I'd stick with a CLE266 based VIA system though as the CN400 graphics is a pain to get working in X. CPU requirements aren't that bad for the frontend provided you have an X driver that supports all the appropriate acceleration extensions, that's the most important thing to consider. mythfrontend is using about 120MB of RAM on my frontend computer right now. Add X and other apps and 256MB or 512MB RAM should be fine. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: good distro/instructions for soekris box?
Kevin D. Clark writes: Does anybody here have a favorite distro or a pointer to a favorite set of directions for installing Linux/BSD onto a Soekris box? I plan on using a 1MB CF card for my storage (no moving parts). I'll probably end up running HTTP/SSH/DHCP/firewall and possibly some other stuff on the box. There's lots of information out there. I can find all of that. I'm just curious if somebody here has already found a particularly helpful set of instructions. I've got a soekris net4801, works great. I started with a debian install on it then modified things until writes to rootfs were completely eliminted under normal operation to save the CF from failure. Some kernel and boot patches are here, but I haven't documented the userspace changes: http://centerclick.org/net4801/ -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: CPUs with variable speed clocks ?
Thomas Charron writes: Aye, the Intel Core 2 Duo's have 'Advanced Intel Speedstep' capabilities. The clock can be dynamically modified by multipliers, I believe up to 8 different speeds. I'll give you more info as I investigate it, as the kernel I built last night I enabled for it. Thomas speedstep and powernow work fine under linux, but usually require a kernel that knows about your sepecific cpu. Auto-detection isn't fully baked, and I have to load the appropriate modules, but it works fine after that. in my case: * cpufreq_userspace * powernow-k8 On my X2 I get 4 speed choices and both cores need to run the same speed: pmg-alliance:10002:/sys/devices/system/cpu/cpu0/cpufreq/ cat scaling_available_frequencies 220 200 180 100 pmg-alliance:10003:/sys/devices/system/cpu/cpu0/cpufreq/ cat scaling_cur_freq 100 pmg-alliance:10004:/sys/devices/system/cpu/cpu0/cpufreq/ cat affected_cpus 0 1 I'd recomend powernowd (v0.97) using the userspace governor, but there are other choices out there. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
Re: Long-term (hours+) solution for power outage: PBX, lights, and computers
Ted Roche writes: A client suffered a five-hour downtime recently, and has been looking for a reasonably priced solution. He's working out of leased office space, and a generator or major structural changes are unreasonable. He happened upon: http://www.sentrypowertechnology.com/ And they offer 6 KW and 12 KW units. The client would like to keep a half-dozen computers, LCD screens and a laser printer or two running for a 4- to 6-hour period. Anyone have experience with a solution that would work? Things to look out for? For example, each computer has its own small APC UPS, so there would be two sources of battery power in series. It's possible the generated AC would not be stable enough in voltage or frequency to keep the UPSes from tripping on and off. I know this is a problem with the smaller, less-regulated generator units. Is there a good solution for a small office? Of should he be looking for a 10K VA UPS? Most UPS manufacturers have extended-run models that you can attach multiple external battery packs for hours and hours of runtime. Are we talking about equipment that is distributed throughout the office or a bunch of network equipment and servers? If they are centralized then one big extended run UPS is in order. It's generally not a good idea to chain UPSs. For really long run (4+ hours) a generator to power the UPS is usually a better solution. 3KVA-6KVA generators aren't that big, but you do need to wheel them outside before turning them on. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: kernel headers question.
Steven W. Orr writes: So having said all that, is it safe to say that for all versions of 2.4 and 2.6 series kernels, it's ok to have one old version of glibc-kernheaders? Or is there some reason that I am constrained to having seperate kernel headers for each project thereby necessitating that each version of gcc, binutils, and UClibc is dependant on the version of linux that it will run against? When building gcc/binutils you generally do not need the specific version of the kernel headers (or more specifically --with-headers=something from gcc's configure line). The only really linux specific part in gcc/binutils is the crt*.o . Even libgcc.a, libiberty.a, libstdc++.a aren't very kernel specific, but do require something in the lines of header files to build. I generally use something along the lines of: construct sys include from a 'close-enough kernel' build binutils cross build gcc cross stage 1 (only c support) build glibc cross build gcc cross stage 2 (c c++) blow away this glibc results install stage 2 gcc, binutils, and sys includes system wide. (if you want c++ support you unfortunately need to build a cross libc inbetween stage 1 and 2.) The specific kernel headers are more important when building your *libc as any change in the syscall area has a big effect on libc. Note that this part has settled down a long time ago and you can usually get away with a close-enough kernel here too. If you're not changing syscalls or other kernel headers then you can build your *libc once and forget it. workareas consist of: build kernel build *libc (rebuilding isn't needed that often) build everything else If you're not changing your libc or major changes in your kernel (that would effect *libc) you can build the libc and install it system wide along with gcc/bintuils. Also note, you need to make sure binutils and gcc dont look in /usr/include or /usr/lib by default, but instead uses your sys includes you build with and the libgcc.a, libiberty.a and libstdc++.a that you built. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: kernel headers question.
Steven W. Orr writes: =The only really linux specific part in gcc/binutils is the crt*.o =. Even libgcc.a, libiberty.a, libstdc++.a aren't very kernel =specific, but do require something in the lines of header files to =build. = =I generally use something along the lines of: = =construct sys include from a 'close-enough kernel' This is probably really the heart of the question. Is any early 2.4 set of headers going to be good enough for any *libc to be run on any 2.4 series that is equal to or later than the set of headers used? Do I need to have a different set for 2.6 kernels or can I continue use of the headers that came from a 2.4 kernel? x86 should be rather good between 2.4 and 2.6, but other architectures may be more different and have conflicts however I don't recall any issues. You may run into problems building the latest and greatest glibc against 2.4 headers as it may be looking for new syscalls such as sched_getaffinity() which were added in 2.6. =build binutils cross =build gcc cross stage 1 (only c support) =build glibc cross =build gcc cross stage 2 (c c++) =blow away this glibc results =install stage 2 gcc, binutils, and sys includes system wide. = =(if you want c++ support you unfortunately need to build a cross =libc inbetween stage 1 and 2.) I don't understand why? In order to build shared versions of libraries ld needs crt1.o as well as -lc (both built as part of *libc) to link a .so. So more correctly I should have said --enable-shared (highly desirable in stage 2) requires glibc although I recall c++ may also need it. I just ran a build skipping glibc and the first failure in stage 2 was building libgcc_s.so.1 because it couldnt' find crt1 and libc. =The specific kernel headers are more important when building your =*libc as any change in the syscall area has a big effect on libc. Agreed, but I'm not sure I understand why syscall comes into play. I get that there are constants to syscall that correspond to new system calls in newer kernels, but I'm not sure I understand why i care. see above. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: Easy to use CAD / architectural design sw suggestions?
Bruce Labitt writes: Bruce Labitt wrote: I need to make up a set of scale drawings for a workshop I'm setting up. Does anyone have some experience with architectural drawing package that is affordable / cheap / free ? I used to have an ancient copy of Generic Cadd that ran on DOS. It was quite powerful. Anything available for Linux? I need a big step up from pencil. Anyone have experience on getting one of the ~$49 CAD software packages for sale for that other OS to run in Linux? ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss I'm attempting to compile QCAD. I get an error about iso c++ not supporting 'long long's In file included from /usr/lib/qt3/include/qobjectdefs.h:42, from /usr/lib/qt3/include/qobject.h:42, from actions/rs_actioninterface.h:31, from actions/rs_actioninterface.cpp:28: /usr/lib/qt3/include/qglobal.h:714: error: ISO C++ does not support `long long' /usr/lib/qt3/include/qglobal.h:715: error: ISO C++ does not support `long long' So I googled compile qcad and found a thread at http://groups.yahoo.com/group/qcad-user/message/1406 . It said to grep -rn -pedantic the source directory go to the file with -pedantic in it and delete -pedantic (it's a compiler switch) compile, that should do it. I'm sure it is that I am a total novice at this... I go to the directory where I have the source installed and execute grep -rn -pedantic * grep tells me: grep: invalid option -- p Usage: grep [OPTION]... PATTERN [FILE]... Try `grep --help' for more information. This wasn't too helpful ;) at least not for me... Can anyone offer a helpful hint? Be gentle, I only hack about at this after hours. I did try enclosing in double quotes, same answer... Thanks! Yep, just downloaded qcad-2.0.5.0-1-community.src.tar.gz and I get: $ cd scripts $ QMAKESPEC=/usr/share/qt3/mkspecs/linux-g++ QTDIR=/usr/share/qt3 ./build_qcad.sh [...] make[2]: Entering directory `/home/pmg/tmp/qcad-2.0.5.0-1-community.src/qcadlib/src' g++ -c -pipe -pedantic -Wall -W -O2 -DRS_NO_COMPLEX_ENTITIES -DQT_NO_DEBUG -DQT_SHARED -DQT_THREAD_SUPPORT -I/usr/share/qt3/mkspecs/linux-g++ -I. -I../include -I../../dxflib/include -I../../fparser/include -I../../qcadcmd/include -I/usr/include/qt3 -Imoc/ -o obj/rs_actioninterface.o actions/rs_actioninterface.cpp In file included from /usr/include/qt3/qobjectdefs.h:42, from /usr/include/qt3/qobject.h:42, from actions/rs_actioninterface.h:31, from actions/rs_actioninterface.cpp:28: /usr/include/qt3/qglobal.h:712: error: ISO C++ does not support `long long' /usr/include/qt3/qglobal.h:713: error: ISO C++ does not support `long long' make[2]: *** [obj/rs_actioninterface.o] Error 1 make[2]: Leaving directory `/home/pmg/tmp/qcad-2.0.5.0-1-community.src/qcadlib/src' make[1]: *** [lib/libqcad.a] Error 2 make[1]: Leaving directory `/home/pmg/tmp/qcad-2.0.5.0-1-community.src/qcadlib' make: *** [all] Error 2 Building qcadlib failed $ The change needed is in: qcad-2.0.5.0-1-community.src/mkspecs/defs.pro: QMAKE_CXXFLAGS_DEBUG += -pedantic QMAKE_CXXFLAGS += -pedantic just comment out those 2 lines and it should build fine. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: Easy to use CAD / architectural design sw suggestions?
Bruce Labitt writes: I got this after 20 minutes or so... Building Translations sh: ./release_translations.sh: No such file or directory Building qcad binary failed I think the community.src doesn't contain translations? Did you get that far? You need to run './build_qcad.sh notrans' By the way, qcad is most likely included in part of your distribution, and you shouldn't need to compile it unless you want to. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: UPSes - MinuteMan, others?
Ben Scott writes: I suspect this type of device is perfect for situations where line voltage is consistently out of tolerance by a certain amount. However, they do not provide isolation from line transients or noise, nor can they cope with severe undervoltages ( 95 VAC or so). We have both here. Sometimes, when everything else is running and then the big mill kicks in, the lights go out for a fraction of a second. If the motor is powerful enough to brown out everything else in your building when it starts up chances are the transformer feeding the building can't handle the startup load and is under powered for your needs. An electrician can hookup monitoring equipment to watch the current/voltage of your feed or just the circuit to the motor and see if this is the case. Your utility can replace the transformer with a bigger one, but that may cost money and downtime. This won't solve everthing, but it's just another area to explore. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: Which CPU are we waiting for to get VM hypervisors in hw?
Ted Roche writes: It's time to start shopping for a new development laptop. I'm a big fan of big iron: the laptop is my primary workstation and I'd prefer to lug around a 4 kg machine than lack for power. I'm tempted both by the ThinkPad T60p and the MacBookPro. Both run Core Duo. While I know these are capable of running VMs, I'm hearing that a next generation chip is going to have more capability in the chips to provide more powerful VMs. The technical details are beyond me, but my questions are simple enough: does anyone know what's the timeline on these new chips, and will their delivery in laptops make all that much perceivable difference or are the features more aimed at big iron (8-way and up) machines. Merom will be out in 1-2 months. Probably worth the wait. http://www.dailytech.com/article.aspx?newsid=3543 http://www.pcper.com/article.php?type=expertaid=276 -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: LACIE USB disk and SuSE 10.1
Jon maddog Hall writes: Hi folks, It is not often I have a problem, but here one is. I recently bought a LACIE 600 GByte USB 2.0 disk, unwrapped it and plugged it into my SuSE 10.1 distribution. Normally I plug in a USB storage device and it mounts, with a nice little icon that comes up. No such luck. I hear the drive spin up and the heads load, but that is about it. So I do a dmesg and I get this out of it: usb 4-1: new high speed USB device using ehci_hcd and address 10 usb 4-1: new high speed USB device using ehci_hcd and address 11 usb 4-1: new device found, idVendor=0451, idProduct=6250 usb 4-1: new device strings: Mfr=2, Product=3, SerialNumber=1 usb 4-1: Product: TUSB6250 Boot Device usb 4-1: Manufacturer: Texas Instruments usb 4-1: SerialNumber: AAB74D460127 usb 4-1: configuration #1 chosen from 1 choice I expected to see something like /dev/sdX but that did not come up. I poked around the net, found some information about ehci_hcd drivers, etc. and it told me to do a lsusb: [EMAIL PROTECTED]:~ lsusb Bus 003 Device 001: ID : Bus 004 Device 011: ID 0451:6250 Texas Instruments, Inc. Bus 004 Device 001: ID : Bus 002 Device 001: ID : Bus 001 Device 001: ID : 'lsusb -v' will show the class of the device: Interface Descriptor: bLength 9 bDescriptorType 4 bInterfaceNumber0 bAlternateSetting 0 bNumEndpoints 3 bInterfaceClass 8 Mass Storage bInterfaceSubClass 6 SCSI bInterfaceProtocol 80 Bulk (Zip) iInterface 0 If it isn't a Mass Storage class usb device you need to create a manual mapping of vendor/device (0451/6250 in your case) to have hotplug map it to the usb-storage driver. This will tell hotplug to load the usb-storage module when that device is discovered. You can manually modprobe usb-storage to see if that helps. I'm not to familiar with suse, but you need to find one of hte mapping files usually under /etc/hotplug or /etc/hotplug/usb and add an entry mapping the vendor/device to the usb-storage driver. I can never remember what all the fields are but something like: usb-storage 0x000f 0x0451 0x6250 0x0 0x0 0x00 0x00 0x00 0x00 0x00 0x00 0x -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: Recycled computers
Ted Roche writes: Branching the One Laptop thread... On May 27, 2006, at 10:27 PM, David Ecklein wrote: I don't understand this fixation on laptops. These are commodities for the I would rather see an effort mounted to refurbish the many usable desktops that are going to the dump every day. LUG members have done this. A group of MonadLUG members lead by Tim Lind worked on this for a while last year. Tim's apparently succeeded in getting a couple refurbished machines into local libraries and elder care facilities. However, there's a lot of resistance to these contributions: concern over support, toxic waste disposal, security, etc. For recycling, there are some good options: http://www.des.state.nh.us/swtas/recycle_electronics.htm http://www.des.state.nh.us/swtas/comp_recyclers.htm Most electronics recycling companies already have re-use deals in place for working components and will then go after the materials on whatever is left over (broken or obsolete). I brought a car-load of old stuff (some working some broken) to this place about a year ago because they would take everything I had for _FREE_ as long as I transported it to them (they charge for pickup). http://www.recyclingelectronics.com/ They have a reasonable amount of info on their website about what they will accept. It was quite a drive (they are in Brockton, MA) but well worth it to avoid the hassles and fees for CRTs and other electronics from my local town or just about any other company I found. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: Liquid Cooling
Ben Scott writes: On 5/30/06, Sean [EMAIL PROTECTED] wrote: I actually figured there would have been a few people here who would be using liquid cooling, but no responses. For whatever reason, the members of this group appear to tend towards a pragmatic bent, and/or are IT professionals. Liquid cooling is generally not cost-effective (it gives you a comparatively small performance gain, for a high cost and risk). It's major application seems to be so people can indulge in games of one-up-manship. Nothing wrong with that, but you don't see it much in a business environment. -- Ben If you're after a quieter computer (as to over-clocking) there are some easier steps to take before liquid cooling: * Invest in a quieter cpu HS/Fan with adjustable speed (manual or temperature based). I've had great success with Zalman parts. * Get a case with BIG intake/exhaust fans, 120mm is very good, 60mm should be avoided at all costs. * If the fan covers are punched out of the case material cut them out especially if they are made up of tiny holes. Replace with an attachable wire grill. I've seen some cases with more than 66% of the fan covered with only very small holes cut out. You can double or triple the airflow by opening up the fan. * If your case rattles or vibrates get some sound dampening material. I've had very good success with Akasa PaxMate brand material. * If you're motherboard has a north-bridge fan, consider replacing it with a large passive heat sink. Some companies sell these but this of course isn't for the novice. Same goes for a GPU fan. * Get an adjustable fan speed controller for your case fans. If you've done the above steps you can under-volt your case fans significantly and still get more than enough airflow. A good 80mm case fan can be very quiet at 1300rpm and still push a lot of airflow if it is unobstructed. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: Using the serial port for GPIO.
Scott Garman writes: Tom Buskey wrote: People usually use the parallel port for this kind of stuff. 8 outputs and at least 5 inputs. More inputs are possible w/ the newer bidirectional ports. parpin on sourceforge lets you work on individual pins. There's tons of info on using the parallel port for digital I/O out there. Thanks Tom, this is extremely useful and exactly what I'm looking for for some of the I/O I'll be doing! /dev/parport0 is your friend. No kernel work needed. I use that for controlling my LCD screen at work. See http://centerclick.org/lcd/ Note that I've found that most modern PCs provide very poor voltages on the parallel port (but still within the TTL ranges). I've even seen different voltages on the Data and Status Pins on the same port!. Don't expect to draw very much current from it either. After much frustration I ended up adding some octal line drivers with a very wide input range between the pport and device to boost everything to acceptable levels. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: Hard drive troubleshooting
Ted Roche writes: In /var/log/messages, smartd is reporting: Device: /de/hdb, 3 Currently unreadable (pending) sectors I'm trying to figure out where they are and how to fix them or mark them as bad. Took the machine down last night and ran SpinRite 6 on it. It reported no problems on the drive. Ran the smartctl tests, both short and long, and they finished without errors. My laptop HD developed the same problem a few years ago. smartd keeps reporting pending sectors. After a few years nothing seems to be bad and the HD keeps on working, however backups are always a good idea. intrepid:~# grep unreadable /var/log/syslog May 11 08:34:43 intrepid smartd[1557]: Device: /dev/hda, 7 Currently unreadable (pending) sectors intrepid:~# smartctl -iA /dev/hda smartctl version 5.32 Copyright (C) 2002-4 Bruce Allen Home page is http://smartmontools.sourceforge.net/ === START OF INFORMATION SECTION === Device Model: IBM-DJSA-220 Serial Number:44W4ZNF4226 Firmware Version: JS4OAC0A Device is:In smartctl database [for details use: -P show] ATA Version is: 5 ATA Standard is: ATA/ATAPI-5 T13 1321D revision 1 Local Time is:Thu May 11 08:34:53 2006 EDT SMART support is: Available - device has SMART capability. SMART support is: Enabled === START OF READ SMART DATA SECTION === SMART Attributes Data Structure revision number: 16 Vendor Specific SMART Attributes with Thresholds: ID# ATTRIBUTE_NAME FLAG VALUE WORST THRESH TYPE UPDATED WHEN_FAILED RAW_VALUE 1 Raw_Read_Error_Rate 0x000b 100 100 062Pre-fail Always - 0 2 Throughput_Performance 0x0005 105 105 040Pre-fail Offline - 6041 3 Spin_Up_Time0x0007 128 128 033Pre-fail Always - 0 4 Start_Stop_Count0x0012 091 091 000Old_age Always - 14316 5 Reallocated_Sector_Ct 0x0033 100 100 005Pre-fail Always - 0 7 Seek_Error_Rate 0x000b 100 100 067Pre-fail Always - 0 8 Seek_Time_Performance 0x0005 095 095 040Pre-fail Offline - 142 9 Power_On_Hours 0x0012 093 093 000Old_age Always - 3124 10 Spin_Retry_Count0x0013 100 100 060Pre-fail Always - 0 12 Power_Cycle_Count 0x0032 100 100 000Old_age Always - 1372 191 G-Sense_Error_Rate 0x000a 100 100 000Old_age Always - 0 192 Power-Off_Retract_Count 0x0032 100 100 000Old_age Always - 2 193 Load_Cycle_Count0x0012 092 092 050Old_age Always - 86909 196 Reallocated_Event_Count 0x0032 100 100 000Old_age Always - 27 197 Current_Pending_Sector 0x0022 100 100 000Old_age Always - 7 ---HERE 198 Offline_Uncorrectable 0x0008 100 100 000Old_age Offline - 0 199 UDMA_CRC_Error_Count0x000a 200 200 000Old_age Always - 0 intrepid:~# -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: Ethernet device sit0
Jim Kuzdrall writes: On Saturday 29 April 2006 10:14 am, Michael ODonnell wrote: That laptop reportedly has a 3c905 Enet so theoretically most Linux distributions should Yes, I had YaST do its hardware search then save the results in a file. The file had 3Com.3C920 buried in a long string of other device names, but it never showed up in the GUI listing of the results. output of dmesg? dmesg | grep -i showed nothing for warn or error. But, boot.msg did say eth-id-00:06:07:00:1b:67 No interface found. I checked the installed software. All the network stuff appears to be installed. I will check some more. I have to shut down to see if the networking has to be enabled in BIOS. I don't remember that being so, but... Jim Kuzdrall There are various NIC options in the C400 bios: http://support.dell.com/support/topics/global.aspx/support/kb/en/document?dn=1062407l=enlangid=1c=uscs=s=gen If none of those helps, make sure the 3c59x kernel module is loaded (or CONFIG_VORTEX is enabled statically). 'lspci' of couse will also reveal what if anything is actually there. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: DNS migration and folks that don't play nice
Cole Tuininga writes: Preface - The folks on the sys-admin list are talking about the migration of services from the older server to the newer server. Of course, one of the issues that's come up is DNS. This led to the following snippet: On Sat, 2006-04-08 at 09:04 -0400, wrote: Well, there's at least one easy workaround for that, aside from the obvious (shorten TTL ahead of time, to force fast propagation). Unfortunately, shortening the TTL doesn't work for clients (like AOL) that cache/maintain their own DNS. I was curious - how do folks in general deal with this? While AOL can certainly constitute a large number of users, my inclination is to say hell with 'em. If they can't conform to proper netiquette, why should I be bending over backwards to support them? I was just curious to get other folks' take on this quasi-philosophical point. For HTTP you can create temporary A/PTRs that have never existed then use a 302 to redirect from old to new. For example: old server has www.example.com that responds with a 302 redirecting to www2.example.com new server hosts both www and www2 with the same content. That way people with and old cache will request a new lookup for www2 (which is new and never had the old address). This of course means you need to keep the www2 name around indefinately because it could end up in people's bookmarks/links. If bandwidth isn't an issue for the short term, the better solution is to NAT requests going to the old server to the new server. Use both SNAT and DNAT in iptables to redirect important UDP/TCP ports on the old server to the new server. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: this has to be a bug...
Paul Lussier writes:
From the 'More for your dollar' department:
[snip]
$ sudo /sbin/ifconfig eth0:0 del 10.107.33.189
$ ifconfig
[snip]
eth0:0:1 Link encap:Ethernet HWaddr 00:0C:F1:E2:A8:6B
inet addr:10.107.33.189 Bcast:0.0.0.0 Mask:0.0.0.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
Base address:0xac00 Memory:ff7e-ff80
[snip]
I ask to delete a non-existent interface, and instead, I get a totally
new one I didn't ask for :)
Ya, a bug in net-tools, but you are using an invalid syntax. Buggy
software and sudo definately dont mix :(
Calling 'del' with an IPv4 IP on an interface that doesn't exist
causes the call to for_all_interfaces() from set_ifstate() to not find
the interface. It doesn't handle this case and assumes you want to
add the address as a secondary address on the base interface.
Two bugs here:
1) del with IPv4 actually adds if what you specify doesn't exist.
2) add and del assume you only call it with a parent interface name
'eth0' not 'eth0:anything here' adding a second ':nnn' anyway.
The following actually work:
ifconfig eth0 add 10.107.33.189
ifconfig eth0 del 10.107.33.189
but run 'del' a second time and it adds it back, but with the correct
name (bug)
see below:
(gdb) r
Starting program: /tmp/net-tools-1.60/ifconfig eth0:0 del 10.107.33.189
Breakpoint 1, set_ifstate (parent=0xbf8a1a10 eth0:0, ip=3173083914, nm=0,
bc=0, flag=0) at ifconfig.c:1118
1118pt.base = parent;
(gdb) n
1119pt.baselen = strlen(parent);
(gdb)
1121pt.flag = flag;
(gdb)
1120pt.addr = ip;
(gdb)
1122memset(searcher, 0, sizeof(searcher));
(gdb)
1121pt.flag = flag;
(gdb)
1122memset(searcher, 0, sizeof(searcher));
(gdb)
1123i = for_all_interfaces((int (*)(struct interface *,void
*))do_ifcmd,
(gdb) p pt
$4 = {flag = 0, addr = 3173083914, base = 0xbf8a1a10 eth0:0, baselen = 6}
(gdb) n
1125if (i == -1)
(gdb)
1127if (i == 1)
(gdb) p i
$5 = 0
(gdb) n
1131for (i = 0; i 256; i++)
(gdb)
1132if (searcher[i] == 0)
(gdb)
1131for (i = 0; i 256; i++)
(gdb) p searcher
$6 = \001, '\0' repeats 254 times
(gdb) n
1132if (searcher[i] == 0)
(gdb)
1135if (i == 256)
(gdb)
1138if (snprintf(buf, IFNAMSIZ, %s:%d, parent, i) IFNAMSIZ)
(gdb) p parent
$7 = 0xbf8a1a10 eth0:0
(gdb) n
1140if (set_ip_using(buf, SIOCSIFADDR, ip) == -1)
(gdb) p buf
$8 = eth0:0:1\000\031\212¿W¹\004\b
(gdb) p ip
$9 = 3173083914
(gdb) p /x ip
$10 = 0xbd216b0a
(gdb) show endian
The target endianness is set automatically (currently little endian)
(gdb)
SIOCSIFADDR on 'eth0:0:1' will create the interface. Kernel is doing
what it's told although with a very non-standard interface name. As
long as it starts with 'eth0:' it considers it an alias.
--
Dave
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: Network testing and latency
Kevin D. Clark writes: Dave Johnson writes: Latency and jitter are side effects due to queuing prior to the bandwidth limited hop. I think that latency has more to do with your transmission medium. And I think that jitter has more to do with contention. Ya, I meant to say queuing introduces jitter and latency. The transmission delay of the medium adds latency regardless of congestion. Protocols such as TCP are designed to avoid introducing latency when a slow link is in the path. I think that the design of TCP more has to do with using the network efficiently and with operating reliably in the presence of congestion. I'm not sure how TCP is designed to avoid intoducing latency. The protocol tries to operate reliably but doesn't really make any guarantees that the bytestream will make it to the destination by a certain time. Should have said avoid introducing excessive congestion. Latency is a side-effect of that. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: Network testing and latency
Paul Lussier writes: Dave Johnson [EMAIL PROTECTED] writes: Note that as mentioned before, limiting bandwidth and introducing latency and/or jitter are different things. If you want to simulate a bandwidth limited link you need to both limit bandwidth and queue packets. If you simply drop and don't queue then there is no possibility of latency. Latency and jitter are side effects due to queuing prior to the bandwidth limited hop. Protocols such as TCP are designed to avoid introducing latency when a slow link is in the path. Anyway, onto the implementation. Below script limits bandwidth in both directions when forwarding through two interfaces. Note you'll need to setup the appropriate interfaces and routes. Each side has it's own bandwidth and queue with a max size in bytes. This is equivilant to a full-duplex T1 pipe using a linux box and 2 ethernet interfaces representing the two endpoints of the T1. Dave, does this add jitter as well? I assume that since you're queueing, it does to some extent. Would it also be wise to inject traffic on the simulated T1 connection by having other hosts communicating as well? For example I could have 1 linux system in between 2 switches, on which were several other systems all communicating with each other doing things like generating web requests, copying large files, etc. It seems this approach would add to the randomness of the connection to some extent, and increase both the latency and jitter experienced by the 2 systems we're really trying to test. Thanks again! Latency and jitter due to queuing will be accurate, however the only thing this won't do is introduce inherant latency due to link speed, distance, etc... Sending 1500 bytes over a T1 takes 7.8ms plus forwarding delays, distance, etc... If you chain multiple connections together you've got even more. The TC setup will forward each packet as fast as possible, just introduce delays between them to simulate a bandwidth limit. In any case if you want a 'real-world' experiance yuo'll need to send other stuff over the link at the same time as your test. TCP connections are easy. If you want to bring the link to a crawl you can use my UDP network test tool to send packets over the link at any rate: http://davej.org/programs/untt/ # server listening on port 1 ./untt -l -v -p 1 # client sending 400 byte packets at 4mbps (uni-directional) ./untt -vv -p 1 -s 400 -r 4000 -c 100 192.168.11.2 The more bursts the more jitter, the more input the more latency. Also note the queue size will directly relate to the max latency because bandwidth is fixed. The 256KB gives a max latency of 1.333 seconds each direction. Flooding the link with a constant 4mbps above fills one of the queues in just seconds and we reach max latency in no time: 64 bytes from 192.168.11.2: icmp_seq=161 ttl=63 time=1350 ms 64 bytes from 192.168.11.2: icmp_seq=162 ttl=63 time=1345 ms 64 bytes from 192.168.11.2: icmp_seq=163 ttl=63 time=1353 ms 64 bytes from 192.168.11.2: icmp_seq=164 ttl=63 time=1348 ms 64 bytes from 192.168.11.2: icmp_seq=165 ttl=63 time=1349 ms 64 bytes from 192.168.11.2: icmp_seq=166 ttl=63 time=1356 ms 64 bytes from 192.168.11.2: icmp_seq=167 ttl=63 time=1350 ms 64 bytes from 192.168.11.2: icmp_seq=168 ttl=63 time=1350 ms -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: Network testing and latency
Ben Scott writes:
On 3/7/06, Kevin D. Clark [EMAIL PROTECTED] wrote:
If all that you want to do is to introduce latency, I would suggest
using iptables dstlimit and fuzzy modules.
Will that really create a realistic reproduction of a higher latency
link, though? Assuming the throughput demands were minimal, latency
would remain low, wouldn't it? And higher throughput demands would
prolly result in TCP throttling or retransmissions or some such, which
would be seen more as a bandwidth limit or bad line, not long
transmission time, per se.
dstlimit and fuzzy are both packet-per-sec limiters not bandwidth
limiters. It makes a big difference if all you packets aren't the same
size.
Note that as mentioned before, limiting bandwidth and introducing
latency and/or jitter are different things.
If you want to simulate a bandwidth limited link you need to both
limit bandwidth and queue packets. If you simply drop and don't queue
then there is no possibility of latency.
Latency and jitter are side effects due to queuing prior to the
bandwidth limited hop. Protocols such as TCP are designed to avoid
introducing latency when a slow link is in the path.
Anyway, onto the implementation.
Below script limits bandwidth in both directions when forwarding
through two interfaces. Note you'll need to setup the appropriate
interfaces and routes. Each side has it's own bandwidth and queue
with a max size in bytes. This is equivilant to a full-duplex T1
pipe using a linux box and 2 ethernet interfaces representing the two
endpoints of the T1.
--
Dave
run this:
# the 2 interfaces you are forwarding through
SIDE_A=eth1
SIDE_B=eth2
# line rates (real and fake)
REALRATE=100mbit
FAKERATE=1500kbit
# max queue size on both ends (bytes)
QUEUESIZE=256kb
# SIDE B - SIDE A
tc qdisc del dev $SIDE_A root
tc qdisc add dev $SIDE_A root handle 2: cbq bandwidth $REALRATE avpkt 1000
tc class add dev $SIDE_A parent 2: classid 2:1 cbq bandwidth $REALRATE \
rate ${FAKERATE} allot 1514 maxburst 10 avpkt 1000 bounded prio 8
tc qdisc add dev $SIDE_A parent 2:1 handle 20: bfifo limit $QUEUESIZE
tc filter add dev $SIDE_A protocol ip parent 2: prio 100 u32 match ip \
protocol 0 0 flowid 2:1
# SIDE A - SIDE B
tc qdisc del dev $SIDE_B root
tc qdisc add dev $SIDE_B root handle 3: cbq bandwidth $REALRATE avpkt 1000
tc class add dev $SIDE_B parent 3: classid 3:1 cbq bandwidth $REALRATE \
rate ${FAKERATE} allot 1514 maxburst 10 avpkt 1000 bounded prio 8
tc qdisc add dev $SIDE_B parent 3:1 handle 30: bfifo limit $QUEUESIZE
tc filter add dev $SIDE_B protocol ip parent 3: prio 100 u32 match ip \
protocol 0 0 flowid 3:1
# show current usage
watch tc -s -d qdisc ls
___
gnhlug-discuss mailing list
gnhlug-discuss@mail.gnhlug.org
http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: using make to convert images to thumbs
Python writes: I have a client that laid out their images and thumbs into almost parallel directory structures. /img/thumb /x /img /y /x /*.jpg /y /*.jpg x and y are two digit directory names used to shorten the directory scans to retrieve files. I thought I could use a Makefile to automate the creation of thumbs, but this pattern has me stumped. I am going to just write a script to step through the source tree and check the corresponding time stamp unless someone offers a suggestion. (No they will not shuffle things around because they have lots of logic in place.) here you go: .DEFAULT: all .DELETE_ON_ERROR: .PHONY: all clean IMAGES=$(wildcard img/*/*/*.jpg) THUMBS=$(patsubst %,thumb/%,$(IMAGES)) all: $(THUMBS) thumb/%: % @mkdir -p $(@D) convert $^ -scale 160x120 -quality 25 $@ clean: $(RM) $(THUMBS) -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: Connection Reset By Peer on ssh sessions
Lloyd Kvam writes: On Sat, 2006-02-04 at 09:36 -0500, Fred wrote: I've got an annoying problem with the new Verizon Fios service. If I leave an ssh session open and sits idle for longer than 2-5 minutes, it is killed with a Connection Reset by Peer error message. I've seen this kind of behavior where there is a stateful, inspection firewall processing packets, though never with a timeout this small. When the firewall dropped the connection info from its state tables, any subsequent packets would be mangled and unacceptable to the remote end which would then close the connection - generating the Connection Reset by Peer message at the local end. I ran tcpdump at both endpoints to document what was happening. The firewall managers were unwilling to make any changes. I do not know if you will be able to get Verizon to do anything to fix the problem. At least ssh has a keep-alive feature that should be somewhat configurable. Hopefully you can send a keep-alive packet every 2 minutes. Ya, definately a firewall or NAT device is aging out your connection from it's connection table. It can be this short of a time if the device is overloaded with connections (such as in a DDoS) or its table size is simply too small for what traffic is flowing through it and it needs to throw out old connections to make room for new ones. Best bet is to use '-oProtocolKeepAlives=90' or '-oServerAliveInterval=60 -oServerAliveCountMax=3' depending on what version of openssh you are using. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
Re: Cohosting around Lebanon - suggestions?
Christopher Schmidt writes: On Fri, Jan 27, 2006 at 01:05:22AM -0500, Ken D'Ambrosio wrote: On Thu, January 26, 2006 11:54 pm, Bill McGonigle wrote: We'd probably want to fund a terminal server/remote power unit to share for decent non-driving management. I have a Zyplex with lots of serial ports but it only speaks telnet, so there would be need for a pokey ssh box in front of it, which might not be worth another U. Somewhere, I've got a power strip that allows remote access. Not sure what protocols it speaks. I think it's an APC, so that probably says something to someone. I'd be glad to contribute it for this project; I imagine poking around with the docs could get it up and running fairly quickly. [I, too, have an older-than-death Ethernet-to-RS-232 gizmo. Since it actually has an AUI port, in addition to the 10-Base-T port, I imagine it only supports telnet.] Presumably this is an APC Masterswitch. I actually wrote a perl script to talk to one of those things at Wedu. They're typically pretty simple: You telnet in, you can get a status of plugs, you can turn them off or on or cycle. We used it to do our heartbeat STONITH (Shoot the Other Node In the Head) step. Worked pretty well when we wanted to kill a machine and din't want to drive to the colo (even though it was only a mile away). It does only support telnet, and only 8 char passwords at that. (At least, ours does.) Note that this was determined by trial and error, and was not documented anywhere obvious. APC materswitches can also be controlled via snmp which is how I control mine. See http://centerclick.org/temp/ms-reboot Also, I have a spare 24port 10/100 managed rackmount ethernet switch that I can bring along. It supports dot1q so we could use a seperate vlan for local equipment such as materswitch/console server/etc.. -- Dave ___ gnhlug-discuss mailing list gnhlug-discuss@mail.gnhlug.org http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss
