GSM sites probably have quite a good idea about the current time, if only to do proper handovers. However, the spec apparently does not include a transmission of this time on the radio interface. The mobile network here is exclusively GSM, and all phones come with a user interface to set the date and time (and a lack of support of daylight saving time). Once set, the phone runs off its own internal clock and quickly wanders off. (I think I once read somewhere that a single manufacturer had implemented an extension to the protocol in its base stations and handsets, so that these particular ones did not have to be set to current time by the user. But it is not generally available)
Anyway, GSM is useless as a reference. And so are many other systems that potentially could provide accurate time. Fortunately there are good radio sources. I have one DCF77 (longwave) and two GPS receivers connected to my system. Very enlightening when timekeepers decide to insert a leapsecond, and a shock goes through most of the NTP network:-)
Even ONE receiver (GPS) can be a problem in an office building with metalized glass windows and no access to the roof. Also, not everyone wants to setup an antenna on the roof and wire it into the computer room.
For typical computer network purposes (where relative time accuracy is more important than absolute accuracy), NTP is a very good solution. It will get all systems on your lan within milliseconds or better, and the whole network within tens of milliseconds. It will be better than a message-based (non-PPS) GPS receiver connected to all your systems!
When you require nanosecond accuracy, you probably don't need it on all systems in your network.
The GPS receiver used internal to these units often has a PPS signal. It is the USB interface that provides only a serial message interface to the system. When you find an RS232-interfaced receiver, chances are that it provides PPS on the DCD pin.
I am using OEM module GPS receivers here (bare printed circuit boards that are/were used by system integrators to build systems) and the three different types I have all provide a PPS signal. These are sometimes available as surplus components.
It is considered abusive to scan for an NTP server and then use it as a reference. A "public" NTP server is one that is explicitly listed as public and available for external users.
Listing it on pool.ntp.org of course is an explicit permission to use it.
GSM does not seem to have these time transmissions. If it does, the typical GSM handset does not take advantage of it.
On a LAN, you can expect submillisecond accuracy out of NTP. At least when your OS clock can be phaselocked to an external reference. Of course it will be difficult when you are syncing over an asymetrically loaded Internet link.
Also look at how they seem to be defining spam only as an "incoming messages" related problem. They have installed spamfilters, but only on the input. Every Nigerian can create a hotmail account and start spamming, and their filters don't bother to act. Receivers of those mails can complain at abuse@hotmail.com, but it will take two weeks for them to process the complaint and lock the account, at which time the spammer just opens a new one.
Is it really that difficult to scan outgoing mail, rate-limit the mail especially on new accounts, and implement other filters to guard the world from the spam sent by hotmail users (rather than the hotmail users from the spam sent to them)???
Flash disks solve this problem by using a translation table between the blocknumbers that you write to, and the real flashblocks on the chip. When you write to a block many times, another block is picked on the chip and the blocknumber is written to the table (also in flash). Then you start wearing out a different block on the chip. With a clever manager routine this can balance the number of writes all over the chip, and it looks like you can write billions of times to one single sector.
Maybe not all versions of this display have this characteristic?
You could try this: when using Linux, poweroff the monitor, wait maybe 30 seconds, and switch it back on. Then look at the end of/var/log/messages. Do you see a lot of messages about new USB hardware, new diskdrives, read errors on these diskdrives, etc?
This would indicate it has lost and re-discovered all this USB hardware. Maybe not all distributions have this autodetect-and-try-to-automount-everything behaviour. I am using SuSE Linux.
I always wonder when their large customers start asking why Cisco-produced hardware is so much more expensive than, say, Dell or HP-produced hardware, and they still require service contracts to keep it running.
I would say either you make and sell it at todays electronics prices and then let the customer pay for service, you you have them pay premium prices and then service it for free.
It is the same with memory modules... a memory expansion for a Cisco router is just a bogstandard PC memory module at a 20-50 times price markup.
Of course they blackmail you by withdrawing all service and support when you install thirdparty memory (or their distributors do this).
When you ask about this they will babble about extra testing, certification and whatever has been done to these modules. It will probably be the same with your ethernet card.
PC manufacturers sometimes try this as well, but at least their markup is less than Cisco's.
Today's governments see the terrorist threat as a very nice excuse to tighten their hold on the citizens. Orwell was just 20 years early in his predictions...
This is insane. All resources that better be spent on combatting ordinary crime are spent on this invisible "terrorism" thing. What has the average Dutch citizen seen of terrorism? Nothing. The killing of Theo van Gogh? Describing that as an act of terrorism is just bending the definition to force a fit. Meanwhile, people are robbed on the street, burglars break into houses, bikes and cars are stolen, cash is collected for goods sold on Internet and never delivered, and nothing is being done because "it is not a priority" or "it is too difficult to research".
You are right, when they really wanted to avoid terrorism there would be much easier and cheaper ways to do it. Like not sending troops to countries where they are unwelcome. And investigating what those terrorists have against us, gaining the insight that they too have a point.
Thanks for the tip, until now I am using the "xset" command to control things (e.g. "xset dpms force off").
Unfortunately it looks like the version of xscreensaver on this system (SuSE 9.2) does not have a pointerhysteresis option, so I'll see if a more recent version is in SuSE 10 (which I intend to install anyway).
Well, it is a Linux system but now that I have paid some attention again, I have been able to create an icon that shuts down the display when I click on it.
Unfortunately the problem remains that the tiniest motion of the mouse will turn it back on. Apparently Steve (grc.com) has recognized this, but in Linux I don't see a way to work around it (yet).
Fortunately not. That is why the use of the input select switch is a workaround for the problem. It is merely a convenience issue now. The input select needs 5 presses for a complete cycle (5 inputs), and it reacts slowly when the monitor is in standby.
This monitor does not have an external AC adapter. It uses about 59W when running normally, 3W in standby, 1W when "powered off" using the frontpanel button. Of course the powerconsumption does not drop merely by blanking the display, as with a CRT monitor. Powermanagement can set it to standby.
These numbers are consistent with observation: it gets moderately warm when normally operating, and is cold when it is in standby.
That is very true, I easily got about 20% off my Dell 2405FPW just by ordering it from work (calling our Dell contact person and asking for a better deal, he set up a new account so it was billed to me personally).
Obviously, using them as a full-time solution is not optimal.
There is nothing plugged in to these slots and connectors normally. I use them for my USB key and would use them for cards if I had those. The problem still occurs when everything is empty because there still is a hub and a couple of cardreader USB devices that are being detected, and an eager "hwscand" process that likes to find out what exciting new hardware there is to be found, automount cards, etc.
In this, Linux is becoming more like Windows. By default, it is doing things that you don't want it to, and you have to find out how to make it stop doing that.
I have powermanagement enabled, but only after 45 minutes as I find it a nuisance when the screen shuts off while I am watching something or doing other work. Also, when I leave I want the display to blank immediately. I have tried to use the "activate screensaver when moving mousepointer to screen corner" function, but it seems it can only activate a screensaver, not the powermanagement. Furthermore, leaving the system in inactivity time-out powersaving is somewhat unreliable. Sometimes it wakes up because it believes the mouse has moved.
This is a Linux system. But my experience at work with Windows systems shows the detection time and system slowdown for USB is similar, so it is probably caused by USB design defects, rather than operating systems.
Well, with my workaround the display is usable. But the 3007FPW does not allow this workaround (I believe), so I thought one better be warned.
This monitor also sports four USB ports and a media card reader.
Be careful, there is a problem with that! I have a Dell 2405FPW and it has the same ports and readers. When the monitor is switched off, the power supply to this subsystem is cut as well (and apparently it is not powered from the PC USB bus).
I leave my PC switched on all the time, and switch off the monitor when I am not using it. The PC continues to perform server functions. The result is that every switch off and on of the display it will go through the USB hardware discovery cycle, find all the cardreaders, and try to read all card types. This results in a massive amount of log messages and a very slow PC for 5-10 seconds.
The manual tells about this, but I think many users would not think about it when reading the feature list. Fortunately, the monitor has multiple inputs (VGA, DVI, S-Video, Composite, Component) and when switching to one of the TV inputs it goes to standby mode when no signal is present. So as a workaround, I switch it off by setting S-Video input and back on by selecting DVI again. Not as convenient, as it needs multiple button clicks to do so.
We are not an IT organisation, we are a small user with an IT department and the need to connect some offices. We have 3 Cisco routers. Of course we do not have a lab or a test enviroment. Why would an end-user need a testing environment to be able to install firmware upgrades? We are just a user, not the testing department. When installing new firmware breaks everything, of course we go back to a working version and wait for more development.
Our supplier does not have a test environment either.
And now the shocking news: even Cisco does not have a testing environment. When we submitted our bug they insisted on testing in our production environment because it would be too difficult for them to setup a testing environment. Yeah, right. They wanted an ISDN dialin to our router to test. When we told them our policy is to have only dialins with call-back, their reply was that their policy is not to allow dialins or dialbacks to their systems, and this whole idea of testing was called off. Talk about customer support...
Our environment is not that complicated, but it is more complicated than the examples you typically find in cisco documents.
The thing that does not work is running IPsec tunnels over ADSL interfaces that use PPPoA, and enabling IP CEF. As soon as IP CEF is enabled in this configuration, end-to-end routing (i.e. between the LANs) stops. Pinging between the routers or between one LAN and the far router still works. This clearly is a bug, and it is becoming more and more of a nuisance because cisco is tying all the advanced features to the enabling of CEF. Our routers have ADSL connections to an ISP, and the only choice for 99% of ADSL connections here is to have PPPoA. It is no problem to do this with low-end routers like Draytek or with a Linux box, but with a Cisco 1700 or 3700 with ADSL WIC it causes a lot of trouble and can only be working with CEF off, which cripples the functionality.
- someone who has bought cisco equipment cannot open a TAC case. he has to do this via his supplier. imagine a Windows user get his patches via the shop where he bought it? that is insane.
- the naming and content of the feature sets is changing. they seem to be unable to consistently refer to a feature set on their website. when a new version has a different feature set naming, it is unclear what features your existing contract exactly entitles you to.
- the fact that the kernel is monolytic (as the Linux kernel is as well) does not mean you have to distribute your entire software as one image. Cisco is distributing software as if Linux software would exclusively be distributed as Knoppix CDs. For every fix in kernel or some application you would have to download a new CD and hope that besides fixing your problem no other stuff is broken or needs attention because the details have changed.
Aside from these issues, I have serious doubts about Cisco quality control. I have a case with Cisco that has been open for nearly two years, and that they have deferred as a request for new feature while it really is a bug. When I install a newer version it introduces new problems every time. Every time. I have not had a single upgrade that went smoothly without breaking something that worked OK before. Of course this is to be expected when using this method of software upgrade and release; when I install a completely new Linux or Windows version there are problems as well. But at least with those systems you have the choice of updating only the parts you have problems with.
Even more likely, the MS patch will not fix the real problem. This has happened before, you know. Also, the problem may seem duplicate to you because it was found before, but was not yet fixed. And now it is exploited. There does not seem to be a hurry inside MS to release patches for bugs that are not yet exploited.
Cisco is much worse than that! For every update they are releasing a gazillion different versions, feature sets, platforms etc, and you have to figure out yourself what you need. Furthermore, they (can) only release full new versions, no patches. And they don't offer them for free download, no you need an account or special arrangement.
Slashdot Mirror
GSM sites probably have quite a good idea about the current time, if only to do proper handovers. However, the spec apparently does not include a transmission of this time on the radio interface. The mobile network here is exclusively GSM, and all phones come with a user interface to set the date and time (and a lack of support of daylight saving time). Once set, the phone runs off its own internal clock and quickly wanders off.
:-)
(I think I once read somewhere that a single manufacturer had implemented an extension to the protocol in its base stations and handsets, so that these particular ones did not have to be set to current time by the user. But it is not generally available)
Anyway, GSM is useless as a reference. And so are many other systems that potentially could provide accurate time.
Fortunately there are good radio sources. I have one DCF77 (longwave) and two GPS receivers connected to my system.
Very enlightening when timekeepers decide to insert a leapsecond, and a shock goes through most of the NTP network
Try ntpdate instead of netdate. It is a different protocol, not all NTP servers support that.
Even ONE receiver (GPS) can be a problem in an office building with metalized glass windows and no access to the roof.
Also, not everyone wants to setup an antenna on the roof and wire it into the computer room.
For typical computer network purposes (where relative time accuracy is more important than absolute accuracy), NTP is a very good solution. It will get all systems on your lan within milliseconds or better, and the whole network within tens of milliseconds. It will be better than a message-based (non-PPS) GPS receiver connected to all your systems!
When you require nanosecond accuracy, you probably don't need it on all systems in your network.
The GPS receiver used internal to these units often has a PPS signal. It is the USB interface that provides only a serial message interface to the system.
When you find an RS232-interfaced receiver, chances are that it provides PPS on the DCD pin.
I am using OEM module GPS receivers here (bare printed circuit boards that are/were used by system integrators to build systems) and the three different types I have all provide a PPS signal.
These are sometimes available as surplus components.
It is considered abusive to scan for an NTP server and then use it as a reference. A "public" NTP server is one that is explicitly listed as public and available for external users.
Listing it on pool.ntp.org of course is an explicit permission to use it.
GSM does not seem to have these time transmissions. If it does, the typical GSM handset does not take advantage of it.
On a LAN, you can expect submillisecond accuracy out of NTP. At least when your OS clock can be phaselocked to an external reference.
Of course it will be difficult when you are syncing over an asymetrically loaded Internet link.
Also look at how they seem to be defining spam only as an "incoming messages" related problem.
They have installed spamfilters, but only on the input. Every Nigerian can create a hotmail account and start spamming, and their filters don't bother to act.
Receivers of those mails can complain at abuse@hotmail.com, but it will take two weeks for them to process the complaint and lock the account, at which time the spammer just opens a new one.
Is it really that difficult to scan outgoing mail, rate-limit the mail especially on new accounts, and implement other filters to guard the world from the spam sent by hotmail users (rather than the hotmail users from the spam sent to them)???
Flash disks solve this problem by using a translation table between the blocknumbers that you write to, and the real flashblocks on the chip.
When you write to a block many times, another block is picked on the chip and the blocknumber is written to the table (also in flash). Then you start wearing out a different block on the chip.
With a clever manager routine this can balance the number of writes all over the chip, and it looks like you can write billions of times to one single sector.
Maybe not all versions of this display have this characteristic?
/var/log/messages. Do you see a lot of messages about new USB hardware, new diskdrives, read errors on these diskdrives, etc?
You could try this: when using Linux, poweroff the monitor, wait maybe 30 seconds, and switch it back on.
Then look at the end of
This would indicate it has lost and re-discovered all this USB hardware.
Maybe not all distributions have this autodetect-and-try-to-automount-everything behaviour. I am using SuSE Linux.
I always wonder when their large customers start asking why Cisco-produced hardware is so much more expensive than, say, Dell or HP-produced hardware, and they still require service contracts to keep it running.
I would say either you make and sell it at todays electronics prices and then let the customer pay for service, you you have them pay premium prices and then service it for free.
It is the same with memory modules... a memory expansion for a Cisco router is just a bogstandard PC memory module at a 20-50 times price markup.
Of course they blackmail you by withdrawing all service and support when you install thirdparty memory (or their distributors do this).
When you ask about this they will babble about extra testing, certification and whatever has been done to these modules.
It will probably be the same with your ethernet card.
PC manufacturers sometimes try this as well, but at least their markup is less than Cisco's.
Today's governments see the terrorist threat as a very nice excuse to tighten their hold on the citizens.
Orwell was just 20 years early in his predictions...
This is insane. All resources that better be spent on combatting ordinary crime are spent on this invisible "terrorism" thing.
What has the average Dutch citizen seen of terrorism? Nothing. The killing of Theo van Gogh? Describing that as an act of terrorism is just bending the definition to force a fit.
Meanwhile, people are robbed on the street, burglars break into houses, bikes and cars are stolen, cash is collected for goods sold on Internet and never delivered, and nothing is being done because "it is not a priority" or "it is too difficult to research".
You are right, when they really wanted to avoid terrorism there would be much easier and cheaper ways to do it. Like not sending troops to countries where they are unwelcome. And investigating what those terrorists have against us, gaining the insight that they too have a point.
Thanks for the tip, until now I am using the "xset" command to control things (e.g. "xset dpms force off").
Unfortunately it looks like the version of xscreensaver on this system (SuSE 9.2) does not have a pointerhysteresis option, so I'll see if a more recent version is in SuSE 10 (which I intend to install anyway).
Well, it is a Linux system but now that I have paid some attention again, I have been able to create an icon that shuts down the display when I click on it.
Unfortunately the problem remains that the tiniest motion of the mouse will turn it back on. Apparently Steve (grc.com) has recognized this, but in Linux I don't see a way to work around it (yet).
Fortunately not. That is why the use of the input select switch is a workaround for the problem.
It is merely a convenience issue now. The input select needs 5 presses for a complete cycle (5 inputs), and it reacts slowly when the monitor is in standby.
This monitor does not have an external AC adapter.
It uses about 59W when running normally, 3W in standby, 1W when "powered off" using the frontpanel button.
Of course the powerconsumption does not drop merely by blanking the display, as with a CRT monitor. Powermanagement can set it to standby.
These numbers are consistent with observation: it gets moderately warm when normally operating, and is cold when it is in standby.
That is very true, I easily got about 20% off my Dell 2405FPW just by ordering it from work (calling our Dell contact person and asking for a better deal, he set up a new account so it was billed to me personally).
Obviously, using them as a full-time solution is not optimal.
There is nothing plugged in to these slots and connectors normally. I use them for my USB key and would use them for cards if I had those.
The problem still occurs when everything is empty because there still is a hub and a couple of cardreader USB devices that are being detected, and an eager "hwscand" process that likes to find out what exciting new hardware there is to be found, automount cards, etc.
In this, Linux is becoming more like Windows. By default, it is doing things that you don't want it to, and you have to find out how to make it stop doing that.
I have powermanagement enabled, but only after 45 minutes as I find it a nuisance when the screen shuts off while I am watching something or doing other work.
Also, when I leave I want the display to blank immediately. I have tried to use the "activate screensaver when moving mousepointer to screen corner" function, but it seems it can only activate a screensaver, not the powermanagement.
Furthermore, leaving the system in inactivity time-out powersaving is somewhat unreliable. Sometimes it wakes up because it believes the mouse has moved.
This is a Linux system. But my experience at work with Windows systems shows the detection time and system slowdown for USB is similar, so it is probably caused by USB design defects, rather than operating systems.
Well, with my workaround the display is usable. But the 3007FPW does not allow this workaround (I believe), so I thought one better be warned.
That won't give you 2560x1600 pixels of resolution...
This monitor also sports four USB ports and a media card reader.
Be careful, there is a problem with that!
I have a Dell 2405FPW and it has the same ports and readers. When the monitor is switched off, the power supply to this subsystem is cut as well (and apparently it is not powered from the PC USB bus).
I leave my PC switched on all the time, and switch off the monitor when I am not using it. The PC continues to perform server functions.
The result is that every switch off and on of the display it will go through the USB hardware discovery cycle, find all the cardreaders, and try to read all card types. This results in a massive amount of log messages and a very slow PC for 5-10 seconds.
The manual tells about this, but I think many users would not think about it when reading the feature list.
Fortunately, the monitor has multiple inputs (VGA, DVI, S-Video, Composite, Component) and when switching to one of the TV inputs it goes to standby mode when no signal is present. So as a workaround, I switch it off by setting S-Video input and back on by selecting DVI again. Not as convenient, as it needs multiple button clicks to do so.
We are not an IT organisation, we are a small user with an IT department and the need to connect some offices. We have 3 Cisco routers. Of course we do not have a lab or a test enviroment. Why would an end-user need a testing environment to be able to install firmware upgrades? We are just a user, not the testing department.
When installing new firmware breaks everything, of course we go back to a working version and wait for more development.
Our supplier does not have a test environment either.
And now the shocking news: even Cisco does not have a testing environment. When we submitted our bug they insisted on testing in our production environment because it would be too difficult for them to setup a testing environment. Yeah, right.
They wanted an ISDN dialin to our router to test. When we told them our policy is to have only dialins with call-back, their reply was that their policy is not to allow dialins or dialbacks to their systems, and this whole idea of testing was called off. Talk about customer support...
Our environment is not that complicated, but it is more complicated than the examples you typically find in cisco documents.
The thing that does not work is running IPsec tunnels over ADSL interfaces that use PPPoA, and enabling IP CEF. As soon as IP CEF is enabled in this configuration, end-to-end routing (i.e. between the LANs) stops. Pinging between the routers or between one LAN and the far router still works.
This clearly is a bug, and it is becoming more and more of a nuisance because cisco is tying all the advanced features to the enabling of CEF.
Our routers have ADSL connections to an ISP, and the only choice for 99% of ADSL connections here is to have PPPoA. It is no problem to do this with low-end routers like Draytek or with a Linux box, but with a Cisco 1700 or 3700 with ADSL WIC it causes a lot of trouble and can only be working with CEF off, which cripples the functionality.
You seem to disregard that:
- someone who has bought cisco equipment cannot open a TAC case. he has to do this via his supplier. imagine a Windows user get his patches via the shop where he bought it? that is insane.
- the naming and content of the feature sets is changing. they seem to be unable to consistently refer to a feature set on their website. when a new version has a different feature set naming, it is unclear what features your existing contract exactly entitles you to.
- the fact that the kernel is monolytic (as the Linux kernel is as well) does not mean you have to distribute your entire software as one image. Cisco is distributing software as if Linux software would exclusively be distributed as Knoppix CDs. For every fix in kernel or some application you would have to download a new CD and hope that besides fixing your problem no other stuff is broken or needs attention because the details have changed.
Aside from these issues, I have serious doubts about Cisco quality control. I have a case with Cisco that has been open for nearly two years, and that they have deferred as a request for new feature while it really is a bug. When I install a newer version it introduces new problems every time. Every time. I have not had a single upgrade that went smoothly without breaking something that worked OK before.
Of course this is to be expected when using this method of software upgrade and release; when I install a completely new Linux or Windows version there are problems as well. But at least with those systems you have the choice of updating only the parts you have problems with.
Even more likely, the MS patch will not fix the real problem. This has happened before, you know.
Also, the problem may seem duplicate to you because it was found before, but was not yet fixed. And now it is exploited. There does not seem to be a hurry inside MS to release patches for bugs that are not yet exploited.
Cisco is much worse than that!
For every update they are releasing a gazillion different versions, feature sets, platforms etc, and you have to figure out yourself what you need.
Furthermore, they (can) only release full new versions, no patches.
And they don't offer them for free download, no you need an account or special arrangement.
I would not call their policy much better.