You are so right.
Sitting here, surrounded by the humming noise of rackmounted Linux servers(RH 6.2), running; qmail, Apache, Squid, ipchains+NAT/Masq., BRU, Netsaint, Phorum, Samba, ssh, etc, this statement kind of hurt;
"...the bug infested hail-the-script-kiddies Red Hat?"
Especially since our site has survived a DDoS attack, a huge multiple gigabyte mailloop (well, technically it was a mail circle-jerk, caused by an ms-exchange autoresponder), without causing our RH based servers any problems (our bandwith was sucked dry of course).
I do like RH; both the company, and their distro.
I like rpm's, they make it easy to manage our 4 servers. Not that I have any problems with tarballs, but rpm's just makes things easier, while still having strict control over the systems.
I like the way RH is involved in the OSS community.
I like my bugzilla mails, and the fact that RH on the whole, is very responsive with bug-fixes.
I have no problem whatsoever, with other peoples choice of Linux distro. If someone wants to build their very own Linux distro from scratch, then I think that that is a cool thing to do; but it is not a viable option for _me_, though.
I say, that it definitly wasn't a IRQ problem: But I did try:
1. Manual assignment of irq's (bios)
2. Automatic assignment of irq's (bios
3. Switching the NIC between different PCI slots
(yes, the order of pci-cards, do matter, e.g. when having both a normal scsi, and a hardware raid-scsi card).
4. forcing ESCD updates.
5. Edge or level pci triggering
There where only one other card in the pc (the graphic adapter).
In short, I did all the usual stuff, when something gets tricky.
At no point did Win9x complain about irq, i/o or memory allocation errors. (errors that are usually associated with lock-ups)
It also claimed, that the drivers were perfectly installed. (to different versions)
I also tried some of the more arcane stuff, like checking the "System" panel for redundant stuff while in "safe mode", or installing the driver in different ways (mostly a solution from early win95 releases). Of course, fdisk, format and a fresh install of 3 different flavours of win9x made no difference.
Forcing the NIC to use irq 9 (I have yet to see a pci graphic adapter using this irq, and both the bios boot-up message, and win9x claimed that the GA used irq 10, and the NIC irq 9). didn't help.
Put a Linux harddisk in the machine, and it would work.
Put the NIC in a another win98 box, and it would work.
Hm. I had a somewhat similar experience yesterday:
A friend had formatted his harddrive, since his win95 installation had turned ugly.
He then tried to get the pci NIC working again, but had no luck. Then I came, but even though I tried every trick I know of: (3 different ways to install the different incarnations of the driver, switching between pci-slots, tweaking the bios, reinstalling win95, then win98, then win98se), but the pc wouldn't see the network.
Tested the NIC on my own machine; It worked like a charm. Grabbed a spare hd with Dead Rat Linux 6.1, and slipped it into the machine; Linux detected the new hardware, and would ping and ftp the network, seconds later.
There we were; Windows claimed, that everything was OK, though it was not.
The NIC and pc worked with Linux, but not with 3 different versions of Windows (though it had worked before his reinstall.)
Now, this was probably a subtle (Punt & Pray) conflict between the bios, NIC and Windows, that was particular for his machine. But how I really, really hate those kind of debugging sessions, since they turn out to be more like exorcism, than rational problem debugging.
Had I been at a paying costumer, I would have declared the NIC dead (even though it was working) within the first 15 min (=3 reboots), and slipped another brand of NIC in. (and prayed that would work;-)
BTW, the NIC was a well known brand, with a decent chipset, neither the shipped driver, or the newest downloaded drivers made a difference.
I strongly suspect that this is not a question of just slabbing a few ALT tags on a few pages.
Just look at the site; a few random clicks reveals something like 3-4 pictures per page. And the size of the site is going to explode during the olympics. (Would a guestimate of >5000 pages be too far out?)
Regarding the 2 million price tag;
It seems that the IBM offer, is a full accessebility implementation for the blind and visually impaired, not just a few cosmetic ALT tags on the frontpage of the website.
We are talking about almost real-time ALT tagging of live dynamic content.
Should IBM do it for free? No, I really don't think so. Someone clearly forgot to include accessebility issues in the contract, and that is not IBM's fault. IBM should not pay for their incredible rich costumers lack of planning.
Could IBM be faulted, for not insisting on accessebility issues? Perhaps. But then again, the costumer is always right, and therefore gets what he wants.
Besides, web accessebility for the blind, seems to be a non-issue, among 99% of all commercial web sites. This is so wrong, but I do understand why it is so; Blind people are a minority, and not even associated with a strong buying power.
When I tell people about the ALT tag, the usual respons is something like "What? Can blind people surf the web?" So awarness of the problem, is a major issue too.
The irony is of course, that www.ibm.com, is one of the best ALT-tagged corporate websites I know of. They even remember to tag the little "go"-buttons.
>we have come to wonder why so many admins get so >uptight about being scanned.
It is folklore among sys admins, that portscanning is the noisy preamble of a script kiddie attack.
Therefore, they usually install some kind of "port scanning" detection device, for early warning.
A massive portscanning of such a site, may trigger all kinds alarms, and notify the sys admin, who then has to check out what triggered the alarm.
I guess, that at that point, the sys admin transforms into a BOFH. When he discovers that the portscanning came from a uni, he turns into an even more angry and paranoid BOFH, since unis traditionally have been known as script kiddies CO-LOs; lots of unsecure unix boxes, and lots of bandwith.
There is nothing wrong with portscanning, but the present climate makes it a rude thing to do, especially if it is a massive portscan (walking up and down on every port on the entire IP segment).
I think it would be a fair "Acceptable Use Policy", to state, that (massive) portscans, without prior permission from the scanned site, is a no-no. And if someone needs to play around with portscan tools (developing Netcraft like mapping tools and such), they better inform their own sys admin first.
Of course, such rules should only apply to (l)users; sys admins and other divine creatures, knows the craft, and should be allowed such things as portscanning when having a good cause (since they know the implications of portscanning, and take the heat anyway).
It must be said though, that some sys admins seems to regard even the tiniest ping or trace, as a full scale attack on their network, or at least as a personal insult. That is a too stuck up attitude of course.
You are right. The real news is, that AMD might release Durons without multiplier locks. I don't know if AMD really will do this, though the Abit press release seems to imply this. I submitted the story, and looking back, I can see it was a really sloppy submission: It was really unclear whether AMD really was going to remove the multiplier lock on new Durons. Without AMD doing this, the story becomes a non-story. (Even my old bx-board support adjustable multipliers. The focus became skewed toward a particular MB manufacturer's rather tame, (and perhaps misleading?) press release. All in all: a sloppy, underresearched, skewed and speculative submission from my side. Not good.
The version I played (550 point version by Level 9 on the ZX Spectrum 48k), would be very hard to complete without XYZZY, since your batteries would run out, after so many turns.
Colossal Cave is still around. Play it online?: www.xyzzy.com
There is also company called Zzyzx. According to an ad in the "Sys Admin" journal, its the name of "A road scratched into the middle of the desert in Southern California".
Slashdot Mirror
You are so right.
Sitting here, surrounded by the humming noise of rackmounted Linux servers(RH 6.2), running; qmail, Apache, Squid, ipchains+NAT/Masq., BRU, Netsaint, Phorum, Samba, ssh, etc, this statement kind of hurt;
"...the bug infested hail-the-script-kiddies Red Hat?"
Especially since our site has survived a DDoS attack, a huge multiple gigabyte mailloop (well, technically it was a mail circle-jerk, caused by an ms-exchange autoresponder), without causing our RH based servers any problems (our bandwith was sucked dry of course).
I do like RH; both the company, and their distro.
I like rpm's, they make it easy to manage our 4 servers. Not that I have any problems with tarballs, but rpm's just makes things easier, while still having strict control over the systems.
I like the way RH is involved in the OSS community.
I like my bugzilla mails, and the fact that RH on the whole, is very responsive with bug-fixes.
I have no problem whatsoever, with other peoples choice of Linux distro. If someone wants to build their very own Linux distro from scratch, then I think that that is a cool thing to do; but it is not a viable option for _me_, though.
I say, that it definitly wasn't a IRQ problem: But I did try:
1. Manual assignment of irq's (bios)
2. Automatic assignment of irq's (bios
3. Switching the NIC between different PCI slots
(yes, the order of pci-cards, do matter, e.g. when having both a normal scsi, and a hardware raid-scsi card).
4. forcing ESCD updates.
5. Edge or level pci triggering
There where only one other card in the pc (the graphic adapter).
In short, I did all the usual stuff, when something gets tricky.
At no point did Win9x complain about irq, i/o or memory allocation errors. (errors that are usually associated with lock-ups)
It also claimed, that the drivers were perfectly installed. (to different versions)
I also tried some of the more arcane stuff, like checking the "System" panel for redundant stuff while in "safe mode", or installing the driver in different ways (mostly a solution from early win95 releases). Of course, fdisk, format and a fresh install of 3 different flavours of win9x made no difference.
Forcing the NIC to use irq 9 (I have yet to see a pci graphic adapter using this irq, and both the bios boot-up message, and win9x claimed that the GA used irq 10, and the NIC irq 9). didn't help.
Put a Linux harddisk in the machine, and it would work.
Put the NIC in a another win98 box, and it would work.
No, a simple resource conflict it wasn't.
Regards
Peter H.S.
Hm. I had a somewhat similar experience yesterday:
A friend had formatted his harddrive, since his win95 installation had turned ugly.
He then tried to get the pci NIC working again, but had no luck. Then I came, but even though I tried every trick I know of: (3 different ways to install the different incarnations of the driver, switching between pci-slots, tweaking the bios, reinstalling win95, then win98, then win98se), but the pc wouldn't see the network.
Tested the NIC on my own machine; It worked like a charm. Grabbed a spare hd with Dead Rat Linux 6.1, and slipped it into the machine; Linux detected the new hardware, and would ping and ftp the network, seconds later.
There we were; Windows claimed, that everything was OK, though it was not.
The NIC and pc worked with Linux, but not with 3 different versions of Windows (though it had worked before his reinstall.)
Now, this was probably a subtle (Punt & Pray) conflict between the bios, NIC and Windows, that was particular for his machine. But how I really, really hate those kind of debugging sessions, since they turn out to be more like exorcism, than rational problem debugging.
Had I been at a paying costumer, I would have declared the NIC dead (even though it was working) within the first 15 min (=3 reboots), and slipped another brand of NIC in. (and prayed that would work;-)
BTW, the NIC was a well known brand, with a decent chipset, neither the shipped driver, or the newest downloaded drivers made a difference.
I strongly suspect that this is not a question of just slabbing a few ALT tags on a few pages.
Just look at the site; a few random clicks reveals something like 3-4 pictures per page. And the size of the site is going to explode during the olympics. (Would a guestimate of >5000 pages be too far out?)
Regarding the 2 million price tag;
It seems that the IBM offer, is a full accessebility implementation for the blind and visually impaired, not just a few cosmetic ALT tags on the frontpage of the website.
We are talking about almost real-time ALT tagging of live dynamic content.
Should IBM do it for free? No, I really don't think so. Someone clearly forgot to include accessebility issues in the contract, and that is not IBM's fault. IBM should not pay for their incredible rich costumers lack of planning.
Could IBM be faulted, for not insisting on accessebility issues? Perhaps. But then again, the costumer is always right, and therefore gets what he wants.
Besides, web accessebility for the blind, seems to be a non-issue, among 99% of all commercial web sites. This is so wrong, but I do understand why it is so; Blind people are a minority, and not even associated with a strong buying power.
When I tell people about the ALT tag, the usual respons is something like "What? Can blind people surf the web?" So awarness of the problem, is a major issue too.
The irony is of course, that www.ibm.com, is one of the best ALT-tagged corporate websites I know of. They even remember to tag the little "go"-buttons.
>we have come to wonder why so many admins get so >uptight about being scanned.
It is folklore among sys admins, that portscanning is the noisy preamble of a script kiddie attack.
Therefore, they usually install some kind of "port scanning" detection device, for early warning.
A massive portscanning of such a site, may trigger all kinds alarms, and notify the sys admin, who then has to check out what triggered the alarm.
I guess, that at that point, the sys admin transforms into a BOFH. When he discovers that the portscanning came from a uni, he turns into an even more angry and paranoid BOFH, since unis traditionally have been known as script kiddies CO-LOs; lots of unsecure unix boxes, and lots of bandwith.
There is nothing wrong with portscanning, but the present climate makes it a rude thing to do, especially if it is a massive portscan (walking up and down on every port on the entire IP segment).
I think it would be a fair "Acceptable Use Policy", to state, that (massive) portscans, without prior permission from the scanned site, is a no-no. And if someone needs to play around with portscan tools (developing Netcraft like mapping tools and such), they better inform their own sys admin first.
Of course, such rules should only apply to (l)users; sys admins and other divine creatures, knows the craft, and should be allowed such things as portscanning when having a good cause (since they know the implications of portscanning, and take the heat anyway).
It must be said though, that some sys admins seems to regard even the tiniest ping or trace, as a full scale attack on their network, or at least as a personal insult. That is a too stuck up attitude of course.
--
Regards
Peter H.S. (sys admin in spe)
You are right. The real news is, that AMD might release Durons without multiplier locks. I don't know if AMD really will do this, though the Abit press release seems to imply this.
I submitted the story, and looking back, I can see it was a really sloppy submission:
It was really unclear whether AMD really was going to remove the multiplier lock on new Durons. Without AMD doing this, the story becomes a non-story. (Even my old bx-board support adjustable multipliers.
The focus became skewed toward a particular MB manufacturer's rather tame, (and perhaps misleading?) press release.
All in all: a sloppy, underresearched, skewed and speculative submission from my side.
Not good.
The version I played (550 point version by Level 9 on the ZX Spectrum 48k), would be very hard to complete without XYZZY, since your batteries would run out, after so many turns.
Colossal Cave is still around. Play it online?: www.xyzzy.com
There is also company called Zzyzx. According to an ad in the "Sys Admin" journal, its the name of "A road scratched into the middle of the desert in Southern California".