Re: Cat 3 cabling
On Sat, Oct 25, 2003 at 05:16:34PM -0400, George Georgalis wrote: > > so use cat3 for ethernet? can't advise it but if you have some on hand > it may suffice; don't stretch, fold or otherwise push it, also concider > the cost of data loss, downtime and rewiring, should you find the need > to rewire. > As I said, cat3 is OK for 10 MBits Ethernet. "OK" means "certified". Depending on a host of factors (some of them you explained), it may be able to carry 100 MBits. Some days. Others not. Believe me, it can be a nightmare. It may work for weeks, and suddenly stop working, just because it's now closer to some other interferring device. I trashed all my cat3 patch cables long ago, after spending quite some time in a mess of cables just to discover that somebody thought that a red cat3 cable would be much prettier than a grey cat5. And we had switched this machine from a 10 MBps port to a 100 MBps one. -- Nicolas Bougues Axialys Interactive
Re: Cat 3 cabling
On Sat, Oct 25, 2003 at 05:16:34PM -0400, George Georgalis wrote: > > so use cat3 for ethernet? can't advise it but if you have some on hand > it may suffice; don't stretch, fold or otherwise push it, also concider > the cost of data loss, downtime and rewiring, should you find the need > to rewire. > As I said, cat3 is OK for 10 MBits Ethernet. "OK" means "certified". Depending on a host of factors (some of them you explained), it may be able to carry 100 MBits. Some days. Others not. Believe me, it can be a nightmare. It may work for weeks, and suddenly stop working, just because it's now closer to some other interferring device. I trashed all my cat3 patch cables long ago, after spending quite some time in a mess of cables just to discover that somebody thought that a red cat3 cable would be much prettier than a grey cat5. And we had switched this machine from a 10 MBps port to a 100 MBps one. -- Nicolas Bougues Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Cat 3 cabling
On Fri, Oct 24, 2003 at 03:27:32AM +0800, Jason Lim wrote: > > So in essense, since they are both 4-pairs, just looking at it won't let > you know which it is (without actually testing it)? > Right. And furthermore, even "testing" with 100 Mbps Ethernet equipment is not the right thing to do. > Any way to turn Cat 5 into Cat 3, and vice versa? > Cat 3 is ok for 10 MHz signal, Cat 5 for 100 MHz. There are of course other paramters in the spec, like signal attenuation, isolation, etc. There are devices to certify cables, such as Fluke Networks tools. They are not cheap, though ($5000 +). Simply put, don't use Cat3 for Ethernet. Use Cat 5. Or, since Cat 5 is deprecated, Cat 5e, which is 100 MHz too, but with enhancements. Or Cat 6 (250 MHz), or Cat 7 (500 MHz, I think). But then you need to be rich :) -- Nicolas Bougues Axialys Interactive
Re: Cat 3 cabling
On Fri, Oct 24, 2003 at 03:27:32AM +0800, Jason Lim wrote: > > So in essense, since they are both 4-pairs, just looking at it won't let > you know which it is (without actually testing it)? > Right. And furthermore, even "testing" with 100 Mbps Ethernet equipment is not the right thing to do. > Any way to turn Cat 5 into Cat 3, and vice versa? > Cat 3 is ok for 10 MHz signal, Cat 5 for 100 MHz. There are of course other paramters in the spec, like signal attenuation, isolation, etc. There are devices to certify cables, such as Fluke Networks tools. They are not cheap, though ($5000 +). Simply put, don't use Cat3 for Ethernet. Use Cat 5. Or, since Cat 5 is deprecated, Cat 5e, which is 100 MHz too, but with enhancements. Or Cat 6 (250 MHz), or Cat 7 (500 MHz, I think). But then you need to be rich :) -- Nicolas Bougues Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: two ethernet ports on one PCI NIC?
On Fri, Oct 10, 2003 at 09:29:54AM +1300, Jones, Steven wrote: > Think you will have to go to a 4 port NIC, Im not aware of a 2 port one, > > I know of 2 made but I have not tried either, one is a dlink unit > (the other We have been using the quad port DLINK in various setups for several years. They are based on the Tulip chipset, and work very well. And not too expensive. Beware : there are network boards with and integrated switch, too. They will look almost the same, with 4 ports. But they are really a single port + a switch, not four ports. -- Nicolas Bougues Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: RIPE Autonomously System: Question?
On Mon, Sep 29, 2003 at 12:01:29AM +0300, kgb wrote: > On Sun, 2003-09-28 at 23:40, Jonathan McDowell wrote: > > Yes and i thing that, do you know with mine architecture how traffic can > shift my PC without problem? And with more stronger PC architecture can > shift more traffic is that right? I mean zebra don't have problem with > big traffic if pc architecture is good? At some point, you hit the PC's architecture limitations, particularly on the PCI bus. Two main problems : latency and bandwidth. If you're talking about a few hundred Mbits, you should be ok with fairly standard hardware. If you're expecting full gigabit or more, it can get tricky. You should consider specialized hardware. Note that it has nothing to do with Zebra. The problem at stake here is packet routing/forwarding inside the kernel, and the kernel/hardware architecture limits. Zebra is just there to maintain the kernel routing table by using various (in your case, probably BGP) routing protocols. -- Nicolas Bougues Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: large disk datas and backup
On Fri, Aug 08, 2003 at 10:11:54AM +0200, François Chenais wrote: > Hello, > > I have to installed 2 servers with oracle and about 100 Go of disk. > I wonder which configuration I need to make them working. > > - DISK >- RAID1 soft >- RAID1 hard : which card/disk ? >- RAID4 hard : which card/disk ? Uh, why RAID4 ? Consider RAID5. The main questions are rather : - what kind of performances and reliability do you want ? - what's your budget ? You have another choice to make ATA or SCSI ? Soft RAID is OK performance-wise (in RAID1), but you probably won't want to do hotswap on it. The choice between RAID1 and RAID5 depends on the number of drives you plan to have. But for 100Gb, you can either go for 2x160 Gb drives, RAID1, or something like 3x80 Gb in RAID5. RAID1 is usually safer, so I'd go for RAID1 in that case. If you go the ATA way, 3ware boards are the way to go. They now have SATA ones, which seem pretty nice. On the SCSI hand, look for Mylex or Adaptec. > - Processor Again, it's a matter of performance level required against budget. From a single 2 GHz AMD to dual 3 GHz Xeons, there is a wide gap, both performance and price-wise ! > - Is there any system do backup oracle surely ? > Veritax, Legato... I'm not an expert, though. Your best bet is probably to ask Oracle. -- Nicolas Bougues Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: synchronous interface card suggestions
On Mon, Jun 16, 2003 at 08:56:47AM -0400, John Ackermann N8UR wrote: > I'm working with what the radios support -- which is synchronous RS-422. > These are some surplus radios that have quite a bit more range than > 802.11b (though they're not nearly as fast). They are 5 watt output, full > duplex narrowband radios that were spec'd for paths of up to 30 miles. We > need the range and reliability more than the speed... > Err, excuse my curiosity, but what kind of radios are these ? I mean, is it some-licensed-stuff-you-have-good-reasons-to-use, or is it a hack, or is it plain illegal ? 5W looks rather powerful for unlicensed spectrum... -- Nicolas Bougues
Re: synchronous interface card suggestions
On Mon, Jun 16, 2003 at 08:56:47AM -0400, John Ackermann N8UR wrote: > I'm working with what the radios support -- which is synchronous RS-422. > These are some surplus radios that have quite a bit more range than > 802.11b (though they're not nearly as fast). They are 5 watt output, full > duplex narrowband radios that were spec'd for paths of up to 30 miles. We > need the range and reliability more than the speed... > Err, excuse my curiosity, but what kind of radios are these ? I mean, is it some-licensed-stuff-you-have-good-reasons-to-use, or is it a hack, or is it plain illegal ? 5W looks rather powerful for unlicensed spectrum... -- Nicolas Bougues -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Antivirus license
On Tue, Jun 10, 2003 at 09:43:36AM +0200, Tomàs Núñez Lirola wrote: > Hi > I want to put an antivirus on the mail server (BugBear helped me to convince > my boss). Now is time for wondering about licenses. > > Kaspersky and F-Prot (two examples) have a product for a mail server. If I > use > their product for a personal use (wich license price is a 5% of the mail > server license) with amavis, am I doing something illegal? Does the license > permit its use with amavis? > I need to know it for sure... so can anyone help me? > > However, the open alternatives (clamav, openantivirus, etc) are stable > enough? > They get updated fast enough? > We use Amavis + McAfee uvscan. I downloaded uvscan from McAfee website (the Linux / personal version). A few days later, a rep called me, to get feedback about my product "evaluation". Him: Are you satisfied with the product ? Me: Yes, it works fairly well. Him: Do you plan to buy it for your organisation ? Me: Yes, sure ! Him: On how many systems do you plan to install the software. Me: Hmm... One ! Him: Then it's just for your personal use ? Me: No, it's for the mail serer. But we have just one (!) Him: Hmmm... I think we have a problem, we sell this product with a minimum 5 license pack. Me: One would be enough, I believe. Him: Then you should continue to use the free version you downloaded. I understand you don't want to buy a five-pack for just one machine. Me: But I believe it's not possible according to the software license. Him: If you think so, then you have to buy the five-pack. Me: Let's go for it. Note: the five pack was something like 100-200 Euros, I don't remember. Not very expensive anyway. uvscan gets an update every few days. -- Nicolas Bougues Axialys Interactive
Re: Antivirus license
On Tue, Jun 10, 2003 at 09:43:36AM +0200, Tomàs Núñez Lirola wrote: > Hi > I want to put an antivirus on the mail server (BugBear helped me to convince > my boss). Now is time for wondering about licenses. > > Kaspersky and F-Prot (two examples) have a product for a mail server. If I use > their product for a personal use (wich license price is a 5% of the mail > server license) with amavis, am I doing something illegal? Does the license > permit its use with amavis? > I need to know it for sure... so can anyone help me? > > However, the open alternatives (clamav, openantivirus, etc) are stable enough? > They get updated fast enough? > We use Amavis + McAfee uvscan. I downloaded uvscan from McAfee website (the Linux / personal version). A few days later, a rep called me, to get feedback about my product "evaluation". Him: Are you satisfied with the product ? Me: Yes, it works fairly well. Him: Do you plan to buy it for your organisation ? Me: Yes, sure ! Him: On how many systems do you plan to install the software. Me: Hmm... One ! Him: Then it's just for your personal use ? Me: No, it's for the mail serer. But we have just one (!) Him: Hmmm... I think we have a problem, we sell this product with a minimum 5 license pack. Me: One would be enough, I believe. Him: Then you should continue to use the free version you downloaded. I understand you don't want to buy a five-pack for just one machine. Me: But I believe it's not possible according to the software license. Him: If you think so, then you have to buy the five-pack. Me: Let's go for it. Note: the five pack was something like 100-200 Euros, I don't remember. Not very expensive anyway. uvscan gets an update every few days. -- Nicolas Bougues Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: BGP memory/cpu req
On Tue, Mar 11, 2003 at 12:28:04PM +0200, Valkai Elod wrote: > > Anyone running BGP with a global routing table on zebra/debian/gnu/linux? > I do. The machine has 256 Mb, and is pretty happy with it. Here is an extract from ps aux : root 616 0.0 11.3 31012 29204 ? SMar02 6:35 /usr/sbin/zebra -d root 620 1.0 14.6 39924 37876 ? SMar02 132:05 /usr/sbin/bgpd -d And here is "free" : [EMAIL PROTECTED]:~$ free total used free sharedbuffers cached Mem:257980 239160 18820 15304 135116 11216 -/+ buffers/cache: 92828 165152 Swap:0 0 0 > How much memory would it require? Does the CPU matter or is it mostly a > RAM issue? > It's a 450 MHz PIII. CPU power depends on the trafic you expect. We have two peers, and get full routing table from both (+- 120 000 routes). HTH. -- Nicolas Bougues Axialys Interactive
Re: BGP memory/cpu req
On Tue, Mar 11, 2003 at 12:28:04PM +0200, Valkai Elod wrote: > > Anyone running BGP with a global routing table on zebra/debian/gnu/linux? > I do. The machine has 256 Mb, and is pretty happy with it. Here is an extract from ps aux : root 616 0.0 11.3 31012 29204 ? SMar02 6:35 /usr/sbin/zebra -d root 620 1.0 14.6 39924 37876 ? SMar02 132:05 /usr/sbin/bgpd -d And here is "free" : [EMAIL PROTECTED]:~$ free total used free sharedbuffers cached Mem:257980 239160 18820 15304 135116 11216 -/+ buffers/cache: 92828 165152 Swap:0 0 0 > How much memory would it require? Does the CPU matter or is it mostly a > RAM issue? > It's a 450 MHz PIII. CPU power depends on the trafic you expect. We have two peers, and get full routing table from both (+- 120 000 routes). HTH. -- Nicolas Bougues Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: anti virus software for mail server
On Fri, Mar 07, 2003 at 04:13:30PM +0100, Markus Welsch wrote: > >Yes, it's a great Romanian AV software. > >But why not try a GPL software first -- ClamAV? > > I didn't take a look at ClamAV yet! I need a stable, proven-to-work > solution which will still work fine under heavy load. > I use Amavis + uvscan (McAfee) + spamassassin. There is a script to do FTP updates automatically every night. It works very well, but it requires quite a lot of resources. I believe this is more because of spamassassin (Perl interpreter), though. With 1 GB RAM and a PIV, it can handle about 100 msg/minute. -- Nicolas Bougues Axialys Interactive
Re: anti virus software for mail server
On Fri, Mar 07, 2003 at 04:13:30PM +0100, Markus Welsch wrote: > >Yes, it's a great Romanian AV software. > >But why not try a GPL software first -- ClamAV? > > I didn't take a look at ClamAV yet! I need a stable, proven-to-work > solution which will still work fine under heavy load. > I use Amavis + uvscan (McAfee) + spamassassin. There is a script to do FTP updates automatically every night. It works very well, but it requires quite a lot of resources. I believe this is more because of spamassassin (Perl interpreter), though. With 1 GB RAM and a PIV, it can handle about 100 msg/minute. -- Nicolas Bougues Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Story on IDE raids on tech-report.com & slashdot
On Thu, Dec 05, 2002 at 04:54:26PM +0100, Christian Hammers wrote: > Hi > > This might be of interest for the ones that discussed IDE raid in the > past days in this list. > > IDE RAID Examined > http://slashdot.org/article.pl?sid=02/12/04/2245253 > http://tech-report.com/reviews/2002q4/ideraid/index.x?pg=1 > You'll notice there that the 3ware board really sucks when it comes to write in RAID 5 mode. It's important to notice that until recently, 3ware had two versions of their boards : the "standard" one, and the "R5 Fusion" one, with "enhanced write performance", which was sold something like $100 more. The reviewed board is probably of the old "standard" kind. Now all the boards sold are "fusion" ones. Or at least, check when you buy. -- Nicolas Bougues Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Hardware IDE RAID-1 controller recommandation
On Wed, Dec 04, 2002 at 11:25:55AM +, Michelle Konzack wrote: > > > >On Mon, Dec 02, 2002 at 02:22:43AM +, Michelle Konzack wrote: > > >RAID-1 is mirroring. You plan 4- or 8-way mirroring ?? > > what do you mean with '4- or 8-way' ? > You said in your first post that you "need Hardware RAID-1 Controllers for two, four and eight Harddisks." And I was wondering why you would need more than 2 hard disks for mirroring. It can be either : - n-way mirroring : you want to have n identical drives, thus you can live with up to (n-1) failed drives. - some kind of RAID "10", as Russel pointed out, where you would do in fact RAID 1 over multiple RAID 0 arrays. > I have always two disks parallel (one original and one mirror) > OK, it's clearer now. > Had problems with Software RAID, because the mirror is not > bootable and after a shutdown thy Server was not starting. > > I think, I can use only Hardware RAID. > Obvisously better. > >3ware. Definetly. Although I'm not sure that it supports RAID-1 arrays > > I have not found a reseller for 3ware in my region... > 3ware website lists at least 4 dealers in Germany... > >> Is there a RAID-1 Controller which support PIO Mode 4 Drives ? > >> > > > >Why on earth would you like to do PIO ? It's awfully slow and > >ineffective, compared to DMA... > > Why use 30 Gigs if a 1 Gig does it... > Because 30 Gb drives are cheaper and faster than 1 Gb ones... > It is only one of my administration Servers and the installation > is around 95 MByte... ;-)) > If what you require is only a small drive, and you don't have too much I/O on it (basically serving webpages is OK, SQL is not), did you consider using CF cards ? You can get 256 Mb or more for about 100 Euros. Once you get a cheap CF-to-IDE adapter (something like 20 Euros), you plug them as IDE drives. Very reliable (almost no need for mirroring), almost zero power consumption, no noise, no heat. We use them in several "appliance like" servers at our customers. The only drawbacks are : - capacity (but as you said, for lots of tasks, that's not a problem) - PIO only. Thus quite slow, about 1 MByte/sec. HTH, -- Nicolas Bougues Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Hardware IDE RAID-1 controller recommandation
On Mon, Dec 02, 2002 at 02:22:43AM +, Michelle Konzack wrote: > Hello, > > curently I am installing some new Servers and I need Hardware > RAID-1 Controllers for two, four and eight Harddisks. > RAID-1 is mirroring. You plan 4- or 8-way mirroring ?? Or you'd like to have multiple drive arrays ? > Can anyone recomand some ? > 3ware. Definetly. Although I'm not sure that it supports RAID-1 arrays with more than two drives (online). But if you go the RAID-5 way, it's ok. > Ic possibel with HOTSWAP. > Possible. Depends on the enclosures you use for you drives. Simple IDE racks are OK, as long as they have a switch to turn off power before hotswap. > Is there a RAID-1 Controller which support PIO Mode 4 Drives ? > Why on earth would you like to do PIO ? It's awfully slow and ineffective, compared to DMA... -- Nicolas Bougues Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: SCSI or IDE
On Fri, Nov 29, 2002 at 07:32:52AM -0800, Eric Jennings wrote: > http://www.3ware.com/products/benchmarks.asp > > > My real world tests below: > > # hdparm -T /dev/sda1 > > /dev/sda1: > Timing buffer-cache reads: 128 MB in 1.01 seconds =126.73 MB/sec > You should probably try to time the disk reads, not the buffer cache... hdparm -t BTW, on my RAID5 setup (4 drives) with a 3Ware card, hdparm gives pretty much the same result as a single drive. I believe that the performance is enhanced in random access, not linear, like hdparm does. -- Nicolas Bougues Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: semi-long distance links
On Wed, Aug 28, 2002 at 12:02:49AM -0300, Carlos Barros wrote: > Hello! > > To make a network between 2 sites in a MAN or WAN the posibilities > would be: > - cabling (leased lines, fiber optic) > - air waves (microwaves, ax25, packet...) > - any provider providing the link via cabling or airwaves or > satelite > > Wavelan/Orinoco cards can make a link about 15km. > > Which tecnologies are available to make a link of ~200 kilometers > with speed at least 64kbps? > > Where can I find more info about this class of networks? > Im looking for the tecnology not to get the service. > If some wireless solution exists, it's definetly in a regulated spectrum area and/or power level. And you probably don't want to have to handle administrative regulatory issues (if it is at all possible to get a license in your country). Furthermore, this kind of equipment is very expensive, requires a nice installation place (like a cell tower, for instance), and is pretty tricky to set up. If I were you, I'd order a DS0 from the telco. Or use another technology, depending on your requirements (permanent link ? heavy traffic ?), and technical possibilities (VPN, Frame relay, X25, ISDN...) -- Nicolas Bougues Axialys Interactive
Re: multiple webcams via one linux box
On Sat, Aug 24, 2002 at 10:51:02AM +1200, Dave Watkins wrote: > At 18:00 23/08/2002 +0200, Nicolas Bougues wrote: > >On Fri, Aug 23, 2002 at 10:06:40AM -0500, Bernie Berg wrote: > >> Hi, I have a project that could potentialy have 85 webcams. The easy > >> thing to do would be to use an Axis network camera and just link to its > >> own webserver from my linux web server (or whatever). But these run > >> about 300 bucks, that would be about 25 grand for 85 cams. X10 on the > >> other hand (I hate their website, it looks like to is from 1994), has > >> much cheaper cameras, and they are wireless. You can get a usb adabpter > >> to input them into a computer. Ummm, anyone have luck linking 85 usb > >> webcams into one linux box? Anyother sugestions? > >> > > > >USB can't have more than 63 devices per bus. > > > FYI 127 is the max for USB, 63 is for Firewire > Thanks for the correction. BTW, it's probably cheap and easy to add a few more USB buses, using PCI boards. This would dramatically lower the number of devices per bus, and thus avoid (some) problems when the number of cams increases (bandwidth, latency, etc. caused by hubs chaining). One question is how the USB cams send data to the host. If it is not compressed at all, even a "small" 320x200x24 bit snapshot will be 192 Kb, and you won't be able to receive more than 3 such pictures on a bus per second. Axis cams send JPEG data, which is far smaller, and on a much higher bandwith bus (100 Mbps Ethernet, which can be further switched/aggregated to Gigabit on the host side), and you can easily capture one picture every few seconds on tens of cameras without problem. Oh, one last thing : I don't know X10 products, but you'll probably have some spectrum management issue when you run 85 wireless cams... -- Nicolas Bougues Axialys Interactive
Re: multiple webcams via one linux box
On Fri, Aug 23, 2002 at 10:06:40AM -0500, Bernie Berg wrote: > Hi, I have a project that could potentialy have 85 webcams. The easy > thing to do would be to use an Axis network camera and just link to its > own webserver from my linux web server (or whatever). But these run > about 300 bucks, that would be about 25 grand for 85 cams. X10 on the > other hand (I hate their website, it looks like to is from 1994), has > much cheaper cameras, and they are wireless. You can get a usb adabpter > to input them into a computer. Ummm, anyone have luck linking 85 usb > webcams into one linux box? Anyother sugestions? > USB can't have more than 63 devices per bus. But the main question is, do you want to handle real time video from each of the cams, all simultaneously, in which case you'll have serious bandwidth problems, starting on USB ? If it's just a matter of taking a snapshot every few seconds, or using only one video feed at a time, the usb solution might be ok (as long as you have at least two buses to connect all the cams). There might just be a few nasty constants in the device drivers, which should be fairly easy to deal with. -- Nicolas Bougues Axialys Interactive
Re: Mysql package is it SMP ?
On Mon, Jun 17, 2002 at 09:37:22AM +0200, Francois B. wrote: > > Hi , > > I would like know if when I do apt-get install mysql-server is SMP or > not ? > > MySQL is multi-threaded, even on UP machines. It will work well on SMP. Note howevere that threading in MySQL is meant to handle one connection/query per thread. Thus, you will get better results only if you run simultaneous queries. Finally, remember that some (not-so-fine-grained) locking happens on write operations, and that your disk subsystem is hardly multithread. Database design will make the difference. -- Nicolas Bougues Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: how to design mysql clusters with 30,000 clients?
On Fri, May 24, 2002 at 01:42:12PM -0400, Jeff S Wheeler wrote: > > While he may still need a large amount of DB muscle for other things, > using PHP/MySQL sessions for a site that really expects to have 30,000 > different HTTP clients at peak instants is not very bright. We have > cookies for this. Server-side sessions are a great fallback for > paranoid end-users who disable cookies in their browser, but it is my > understanding that PHP relies on a cookie-based session ID anyway? > What's not very bright is rather using MySQL in a somewhat audacious configuration, for which support is quite recent (and thus, probably not bugfree). In a high load / high availability environnement. An Oracle would probably be better here. At least, it has proven replication mechanisms. Cookie based *whatever* is generally not a good idea. PHP sessions can be handled using cookies or URL signatures. And BTW, they do not necessarily require database backend. They can be handled on the filesystem, though I'm not sure whether it works well in a shared NFS environnement. Note: I never implemented something like this. These are juste ideas. -- Nicolas Bougues Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: how to design mysql clusters with 30,000 clients?
On Fri, May 24, 2002 at 01:42:12PM -0400, Jeff S Wheeler wrote: > > While he may still need a large amount of DB muscle for other things, > using PHP/MySQL sessions for a site that really expects to have 30,000 > different HTTP clients at peak instants is not very bright. We have > cookies for this. Server-side sessions are a great fallback for > paranoid end-users who disable cookies in their browser, but it is my > understanding that PHP relies on a cookie-based session ID anyway? > What's not very bright is rather using MySQL in a somewhat audacious configuration, for which support is quite recent (and thus, probably not bugfree). In a high load / high availability environnement. An Oracle would probably be better here. At least, it has proven replication mechanisms. Cookie based *whatever* is generally not a good idea. PHP sessions can be handled using cookies or URL signatures. And BTW, they do not necessarily require database backend. They can be handled on the filesystem, though I'm not sure whether it works well in a shared NFS environnement. Note: I never implemented something like this. These are juste ideas. -- Nicolas Bougues Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: booting from CompactFlash Cards
On Sun, Mar 17, 2002 at 01:35:44AM -0500, Jeff S Wheeler wrote: > Where do you buy your CF cards that are bootable, and the IDE adapters? > I would like to do the same thing and would really appreciate it if you > would send me over your vendor information. Part numbers would be nice > if you have them handy, too! > I bought CF to IDE adapters from ACS (http://www.acscontrol.com). $20 each, you can order online. The CF cards I use are no-name 128 MB cards bought from a local retailer. I believe there's nothing special about them, except they were the cheapest around :) As stated on ACS website, "if your application works with IDE drives, it will work with a CF card and our CF to IDE adapter". -- Nicolas BOUGUES Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: booting from CompactFlash Cards
On Thu, Mar 14, 2002 at 01:29:29PM +0100, Christian Hammers wrote: > Hello > > I plan to replace a Cisco by a Linux router and would like to use a > "compact flash" card instead of a hard drive to minimize hardware outages. > > Can anybody recommend me a CompactFlash solution that allowes booting? > > As far as I learned those cards have build-in IDE adapters and are > connected to the PC via a simple connector-adapter to a 40pin IDE cable. > Sadly at least Verbatim do not think that their cards are able to present > a correct "master boot record" to the BIOS although I can see no difference > between requesting sector 0 on track 0 (MBR) and any other position. > We use "no name" CF cards with ACS CF<->IDE adapters (about $20 each), and it works fairly well, just like a normal IDE drive. The only thing is that the CF are PIO only, and thus are quite slow (about 1 Mb/sec), and use a lot of CPU bandwidth, just like any PIO drive. But it's fun and amazing :) -- Nicolas BOUGUES Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: firewire storage solution recommended
On Mon, Feb 18, 2002 at 02:24:09AM +0800, Patrick Hsieh wrote: > Hello "Thomas R. Shemanske" <[EMAIL PROTECTED]>, > > I bought Datafab MD2-FW-USB External 2.5 IDE HDD Enclosure today and > successfully work in my debian notebook. > > My question is-- is it possible to use 1394 HUB to connect some 1394 > storage device and some Linux servers and make it a 1394-based NAS > environment? Can two Linux servers mount the same 1394 storage device > and access at the same time? > If the question is "can two hosts access the same 1394 device at the same time ?", the answaer is probably yes, as long as everything is eletrically set up properly, and that drivers/firmwares support that kind of things. But if question is "can two hosts mount the same 1394 device (in R/W mode) at the same time ?", the answer is definetly no. At least with "conventional" FS (ext2/3, fat, etc.), for various reasons, including : kernel level read/write cache, which can assume things about what's on the drive and what's not (yet), no locking, etc. -- Nicolas BOUGUES Axialys Interactive
Re: firewire storage solution recommended
On Mon, Feb 18, 2002 at 02:24:09AM +0800, Patrick Hsieh wrote: > Hello "Thomas R. Shemanske" <[EMAIL PROTECTED]>, > > I bought Datafab MD2-FW-USB External 2.5 IDE HDD Enclosure today and > successfully work in my debian notebook. > > My question is-- is it possible to use 1394 HUB to connect some 1394 > storage device and some Linux servers and make it a 1394-based NAS > environment? Can two Linux servers mount the same 1394 storage device > and access at the same time? > If the question is "can two hosts access the same 1394 device at the same time ?", the answaer is probably yes, as long as everything is eletrically set up properly, and that drivers/firmwares support that kind of things. But if question is "can two hosts mount the same 1394 device (in R/W mode) at the same time ?", the answer is definetly no. At least with "conventional" FS (ext2/3, fat, etc.), for various reasons, including : kernel level read/write cache, which can assume things about what's on the drive and what's not (yet), no locking, etc. -- Nicolas BOUGUES Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Stupid question maybe.
On Thu, Jan 31, 2002 at 02:07:23AM +0200, Eetu Rantanen wrote: > On Wed, 30 Jan 2002, Nicolas Bougues wrote: > > > rsync --rsh="ssh ssh" :/some/dir /some/local/dir > > > > I tested this second solution, and I can't see how to make ssh ask the > > second password (for machine C). It complains it has no controlling > > tty. It works well if you setup your ssh to connect without passwords > > from B to C, however (by trusting keys). > > Try it with ssh -t, > -t Tty; allocate a tty even if command is given. > > It'll then prompt for the password. > That's what I tried. It worked well for the first hop, but not the next one. But maybe I got something wrong. -- Nicolas BOUGUES Axialys Interactive
Re: Stupid question maybe.
On Thu, Jan 31, 2002 at 02:07:23AM +0200, Eetu Rantanen wrote: > On Wed, 30 Jan 2002, Nicolas Bougues wrote: > > > rsync --rsh="ssh ssh" :/some/dir /some/local/dir > > > > I tested this second solution, and I can't see how to make ssh ask the > > second password (for machine C). It complains it has no controlling > > tty. It works well if you setup your ssh to connect without passwords > > from B to C, however (by trusting keys). > > Try it with ssh -t, > -t Tty; allocate a tty even if command is given. > > It'll then prompt for the password. > That's what I tried. It worked well for the first hop, but not the next one. But maybe I got something wrong. -- Nicolas BOUGUES Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Stupid question maybe.
On Wed, Jan 30, 2002 at 05:24:48PM +0100, Nicolas Bouthors wrote: > Hi, > > Here is my trouble : I'm working on machine A and I want to 'rsync' some > files to machine C. Machine C is on another (private) net, > unreachable from machine A. > > Machine B is in the between and is only reachable by ssh. > > So if I want to rsync from A to B, rsync -e ssh /some/dir B:/some/dir is > enough, but what should I do to go straight fro A to C ? > I see two possible solutions : - launch rsync on machine B. From machine A, this would look like : ssh "rsync --rsh=ssh :/some/dir :/some/dir" - or try to forward the ssh connection : rsync --rsh="ssh ssh" :/some/dir /some/local/dir I tested this second solution, and I can't see how to make ssh ask the second password (for machine C). It complains it has no controlling tty. It works well if you setup your ssh to connect without passwords from B to C, however (by trusting keys). -- Nicolas BOUGUES Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: IMAP design implementation (Cyrus, Courier or WU) -FB wanted
On Sat, Jan 19, 2002 at 04:36:00PM -0500, Ted Knab wrote: > windows/mac clients. > > I have decided to replace our current server with an Exim/IMAP combo > > either running Courier, Cryrus or WU-IMAP. > > I am hoping you will help me decide on the IMAP a good one. > Well, according to me, the main point is about local file storage. I think UW-IMAP uses primarily "plain old" UNIX mbox format, which is : - very well known, compatible with just about anything - easily corruptible (not NFS safe) - needs some tweaking for IMAP support - one file per mailbox, which is not very efficient for deletions / updates of individual messages Cyrus uses a proprietary storage layout, which includes local databases : - it's "meant for IMAP", one file per message - it features a complete ACL system and quota support - it's corruptible, and sometimes it's hard to figure out what's wrong, and how to rebuild the right database - I don't think it's NFS safe - it's not simple : you can't just "play with the files", you have to keep the databases in sync Courier uses Maildir (the native storage format of qmail) : - it's open, simple and efficient : one file per message, nothing else - it's fully NFS safe, using a purely fs based locking mechanism - it requires some tweaking for IMAP support I've setup various mail servers, using mbox, cyrus and Maildir. And I'm now "sold" to the Maildir format. It's "unbreakable", even when different POP/IMAP servers and various MTAs play with one spool at the same time. It's now supported by most MTAs (I use Postfix), and various POP/IMAP servers (I've just tried courier, and I'm fairly happy with it). mbox will be painful (corruption will happen). cyrus looks interesting, but once you're there, you can't use anything else, because the storage format is quite proprietary (although it's not secret). So my best advice would be : go for a Maildir based setup. Use your favorite MTA, and try Courier IMAP. You'll still be able to switch MTAs, POP or IMAP daemons later. -- Nicolas BOUGUES Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Multiple DSLs, and switching incoming route upon failure?
On Mon, Jun 25, 2001 at 06:00:44PM +1000, Jeremy Lunn wrote: > On Mon, Jun 25, 2001 at 01:09:13AM -0400, Fraser Campbell wrote: > > Why not have a DNS server on each network announcing different IPs for each > > service and then multi-home each server? DNS on DSL1 would only annouunce > > IPs from DSL1, and DNS on DSL2 would only announce IPs from DSL2. Due to > > the > > way DNS servers are used in a round-robin fashion you should get crude load > > balancing ... if DSL1 goes down only the DNS server in DSL2 would be > > reachable and therefore only DSL2 IPs handed out. > > How is that going to be any better than having multiple A records? > Apart from the fact that it'd be more complex to maintain. > There should be an almost 0 TTL on each DNS server, and both of them would be primary for the zone, but with different data. But well, here in France, one leased line is more reliable than 2 DSL links... -- Nicolas BOUGUES Axialys Interactive
Re: Multiple DSLs, and switching incoming route upon failure?
On Mon, Jun 25, 2001 at 06:00:44PM +1000, Jeremy Lunn wrote: > On Mon, Jun 25, 2001 at 01:09:13AM -0400, Fraser Campbell wrote: > > Why not have a DNS server on each network announcing different IPs for each > > service and then multi-home each server? DNS on DSL1 would only annouunce > > IPs from DSL1, and DNS on DSL2 would only announce IPs from DSL2. Due to the > > way DNS servers are used in a round-robin fashion you should get crude load > > balancing ... if DSL1 goes down only the DNS server in DSL2 would be > > reachable and therefore only DSL2 IPs handed out. > > How is that going to be any better than having multiple A records? > Apart from the fact that it'd be more complex to maintain. > There should be an almost 0 TTL on each DNS server, and both of them would be primary for the zone, but with different data. But well, here in France, one leased line is more reliable than 2 DSL links... -- Nicolas BOUGUES Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: WAN Adapters...Wan in general
On Fri, Jun 08, 2001 at 06:08:17PM +, Alex wrote: > A question to you all: > > Now, as far as ive gotten by my research, one needs to buy a WAN card > that understands the HDLC protocol or the SyncPPP protocol (depending on > your provider). Ive foung at least three that run under linux. > > Now something made me nervous my provider said he can get me a V.35 > line or a g207 line (i dont know what does that mean), i cant find docs > on bridging from this kind of interface to ethernet. > I believe you're talking about a T1/E1 link. Basically, the telco brings you the T1/E1 trunk. Then, depending on the country/operator, they provide you with a CSU/DSU, or not. It they do, the CSU/DSU will provide a sync serial port, either V35 or X21. V35 should be avoided, connectors are ugly and expensive, X21 is OK. Then you'll need a sync board with a matching serial interface (see below). If they don't, they provide you a basic G703 T1 or E1 line. You have either to buy a CSU/DSU, or to use a board that doesn't require one. In this case, your board will connect directly to the 4 telco wires, using (usually) an RJ45 plug. Such board (with or without CSU/DSU) exist for Linux. Try : www.sangoma.com, www.etinc.com, etc. -- Nicolas BOUGUES Axialys Interactive
Re: WAN Adapters...Wan in general
On Fri, Jun 08, 2001 at 06:08:17PM +, Alex wrote: > A question to you all: > > Now, as far as ive gotten by my research, one needs to buy a WAN card > that understands the HDLC protocol or the SyncPPP protocol (depending on > your provider). Ive foung at least three that run under linux. > > Now something made me nervous my provider said he can get me a V.35 > line or a g207 line (i dont know what does that mean), i cant find docs > on bridging from this kind of interface to ethernet. > I believe you're talking about a T1/E1 link. Basically, the telco brings you the T1/E1 trunk. Then, depending on the country/operator, they provide you with a CSU/DSU, or not. It they do, the CSU/DSU will provide a sync serial port, either V35 or X21. V35 should be avoided, connectors are ugly and expensive, X21 is OK. Then you'll need a sync board with a matching serial interface (see below). If they don't, they provide you a basic G703 T1 or E1 line. You have either to buy a CSU/DSU, or to use a board that doesn't require one. In this case, your board will connect directly to the 4 telco wires, using (usually) an RJ45 plug. Such board (with or without CSU/DSU) exist for Linux. Try : www.sangoma.com, www.etinc.com, etc. -- Nicolas BOUGUES Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Streaming MP3
On Mon, May 21, 2001 at 06:53:49AM -0400, Peter Billson wrote: > > If I cat /dev/dsp I just get a stream of ~~ (and cat /dev/audio > gives me \ \ \ \ \ ) so it would seem that there is actually no sound > coming from the card. The card is a real SoundBlaster 16 ISA PnP card. > Can any one offer any suggestions? > Have you ever been able to record something ? Either : - you have not set up recording source properly with a mixer (aumix for instance) - you haven't plugged the source in the right hole - there is no sound in your cable ?? HTH -- Nicolas BOUGUES Axialys Interactive
Re: Streaming MP3
On Mon, May 21, 2001 at 06:53:49AM -0400, Peter Billson wrote: > > If I cat /dev/dsp I just get a stream of ~~ (and cat /dev/audio > gives me \ \ \ \ \ ) so it would seem that there is actually no sound > coming from the card. The card is a real SoundBlaster 16 ISA PnP card. > Can any one offer any suggestions? > Have you ever been able to record something ? Either : - you have not set up recording source properly with a mixer (aumix for instance) - you haven't plugged the source in the right hole - there is no sound in your cable ?? HTH -- Nicolas BOUGUES Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Webalizer and net-acct differences
> Back to questioning: > recently i did some calculation and find out that webalizer results are > about about 85% of the net-acct results. > Ist that an realistic overhead form http-headers, ICMP (on or to port 80?), > and TCP/IP frame info, etc.? Yes. But it depends upon the kind of data served. The header size is quite fixed, but the payload size may vary. A site with loits of small replys will have a percentage more like 60%. Furthermore, apache doesn't take into account *incoming* traffic, whereas your hosting provider probably does (ie counts in both directions). There can be great differences here if you do a lot of "posting" (like posting big files, for instance). > > PS: we pay for the traffic "on the cable" and webalizer only gets the > "pay-load" from http. > Then use net-acct to get the real values. Unfortunatly, there's no way to figure out the data for various virtual servers which share the same IP. -- Nicolas BOUGUES Axialys Interactive
Re: Webalizer and net-acct differences
> Back to questioning: > recently i did some calculation and find out that webalizer results are > about about 85% of the net-acct results. > Ist that an realistic overhead form http-headers, ICMP (on or to port 80?), > and TCP/IP frame info, etc.? Yes. But it depends upon the kind of data served. The header size is quite fixed, but the payload size may vary. A site with loits of small replys will have a percentage more like 60%. Furthermore, apache doesn't take into account *incoming* traffic, whereas your hosting provider probably does (ie counts in both directions). There can be great differences here if you do a lot of "posting" (like posting big files, for instance). > > PS: we pay for the traffic "on the cable" and webalizer only gets the > "pay-load" from http. > Then use net-acct to get the real values. Unfortunatly, there's no way to figure out the data for various virtual servers which share the same IP. -- Nicolas BOUGUES Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Sendmail
On Thu, Feb 15, 2001 at 02:23:07PM +0100, Roger Abrahamsson wrote: > > Anyone here know how to 'force' sendmail to bind to one specific ip on a > machine? I've just moved it, and it works fine but for one thing, it > stubbornly wants to use the primary ip/interface when sending messages > out. It's causing problems with a few customers firewalls... > I'm not sure that sendmail can do that. I ran into this problem for BIND notifications (the primary server not using the right interface, thus the slave not accepting it). This might be dealt with the routing table, though. -- Nicolas BOUGUES
Re: Sendmail
On Thu, Feb 15, 2001 at 02:23:07PM +0100, Roger Abrahamsson wrote: > > Anyone here know how to 'force' sendmail to bind to one specific ip on a > machine? I've just moved it, and it works fine but for one thing, it > stubbornly wants to use the primary ip/interface when sending messages > out. It's causing problems with a few customers firewalls... > I'm not sure that sendmail can do that. I ran into this problem for BIND notifications (the primary server not using the right interface, thus the slave not accepting it). This might be dealt with the routing table, though. -- Nicolas BOUGUES -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Managing a mail/web server without Unix accounts
Bonjour Stéphane, Hello all, On Thu, Feb 01, 2001 at 11:08:23AM +0100, Stephane Bortzmeyer wrote: > > > I am looking for a documentation, as much detailed as possible, > on the setup of an Internet server (mail, several domains, POP > and IMAP, a Web server with FTP and DAV upload by customers, > may be Zope), *without* Unix accounts. The actual database > should be a DBMS (possibly with three-tier architectures). > I can tell you a few words about our setup; the point is that it works quite well, but is quite not "packaged", although it could be somewhat more. Incoming mail : - accounts stored in a database - "stock" postfix MTA - "heavily patched" postfix local delivery agent, in order to check account against database, enforce quotas on mailboxes and message size, and to handle our "home made" spool directory : mailboxes are basically in Maildir format, but the directory structure is somewhat hashed to handle *lots* of users. For instance, a user spool dir could be [EMAIL PROTECTED]/ (then cur/new/tmp maildir stuff). Outgoing mail : either "blind" relaying of trusted IPs, or SMTP AUTH done with the same database. POP/IMAP (and IMAP based webmail) : slightly patched Qmail POP and Courier Imapd, to handle our database. Web : apache and ftpd are "stock", but we set up a few perl scripts that are called remotely to handle config changes. Config is stored in a global database, and a server's httpd.conf can be regenerated in full at anytime, if required. FTP is handled with "normal" accounts, though. Be aware that DAV is not very server-side scripting friendly yet : there is no way for Apache to send a php file to a DAV client without parsing it. The available solutions to this problem are mostly things like creating other virtual hosts or directories (pointing to the same space), and disabling scripting for these. And although we don't like it much, users like Frontpage a lot, and that's what we use instead of Dav until now, because of these problems. Best regards, -- Nicolas BOUGUES Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Running Server From Behind Firewall
Hello, On Fri, Jan 26, 2001 at 02:06:33AM -0800, Bryan Kroll wrote: > I recently got ADSL and I tried to run a server, well, > I tried many servers but no one else can access it. > Because either my ISP has a firewall or I don't have a > real IP. Its dynamic and thats another problem but if > I can get others to access my server then I will be > satisfied. As far as my changing IP adress goes I can > update my dns every time I reboot. I use zoneedit.com > Its pretty cool. But anyway, can anyone help me or > tell me how to cirvumvent or "pierce" my firewall or > run a server without a real IP address? Thanks, Bryan > You want reverse NAT / forwarding. Current kernel / ipchains isn't very good at it. 2.4 is *much* better for that kind of stuff. IMHO, a good reason to upgrade. -- Nicolas BOUGUES Axialys -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: identd + SOCKS
Hello, On Tue, Jan 09, 2001 at 07:59:10PM +0100, Marcin Owsiany wrote: > Hi! > > First, some ascii "art": > > - -- > ||| SOCKS proxy | | IRC| > | IRC client || and |--/ /---| server | > ||| identd| | | > - -- > > Does anyone know a solution that would allow the IRC server to > get an ident reply of the client's IP number instead of the > SOCKS server username, when the client connects via the SOCKS > server? > This would require some kind of cooperation between the SOCKS proxy (which knows the real identity of the IRC client) and the identd (which gets queried back on a a different connection by the IRC server). I've never seen any such thing. -- Nicolas BOUGUES -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: E1/E3/T3/STM1 cards with Linux support
> Does anybody have any experience with E1/E3/T3 or STM1 cards which are > supported by Linux? Especially cards with a BNC, UTP/RJ345 or SC/APC > connector. I know at least three vendors of such products : - Sangoma (http://www.sangoma.com) - Emerging (http://www.etinc.com) - Lanmedia (http://www.lanmedia.com) Good experiences with the two firsts, no experience with the later. -- Nicolas BOUGUES Axialys Interactive -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]