Might be, what do we know about this guy's farm? The varying crunch times (the realistic ones I mean) would indicate different boxes. No way to say for sure, of course, unless he tells us.
@Donald: Sorry, maybe I was a bit unprecise here... I was not talking about having a single internet connection and a switch (or router) for multiple computers. That would be a perfectly normal setup (I use it, too). What I meant was that for example you would only plug two of your boxes into said switch but have three other boxes which are not attached to the internet at all (for whatever reason). Then you would download WUs on one of the boxes that are online and split them between the three offline hosts. When they are crunched, you would put them back onto the flash drive or whatever and use the box with internet access to upload them. I've never actually done it but it is widely accepted as a possible practice, as long as the host for receiving and uploading a result is identical.
Annika I've seen one or two people here post that they were doing that with work boxes that weren't directly on the net for security reasons. We have a number of machines like that where I work, but even ignoring the installation of unauthorized software the USB sneakernet would get me as thoroughly stomped on as I would for doing a direct net connection.
I wouldn't do it on my boss's computers, either. But as you said, some people do. And it might still be his private boxes and a network problem (maybe he is really using a modem and has to get a new switch before he attaches the boxes). I would even suggest doing the network stuff from a hotspot but this computer doesn't look anything like a laptop so I'd rule this out.
But even if it's the company's boxes, that's nothing the BOINC admins would have to get active against, is it? I mean, you can be against it morally and/or in terms of getting in trouble but it's not cheating...
@Donald: Sorry, maybe I was a bit unprecise here... I was not talking about having a single internet connection and a switch (or router) for multiple computers. That would be a perfectly normal setup (I use it, too). What I meant was that for example you would only plug two of your boxes into said switch but have three other boxes which are not attached to the internet at all (for whatever reason). Then you would download WUs on one of the boxes that are online and split them between the three offline hosts. When they are crunched, you would put them back onto the flash drive or whatever and use the box with internet access to upload them. I've never actually done it but it is widely accepted as a possible practice, as long as the host for receiving and uploading a result is identical.
Actually, I realized what you meant after I made the post, but didn't take the time to go back and correct it.
But, even if he were copying the project folders from other machines, I would think that the compute times would be part of that, and would copy over as well.
1010940 just reported the last two it had in progress at 16:00:27 UTC and the runtimes are nominal for what I'd epect for this PD. The Last Contact info hasn't updated yet so I haven't been able to look at that yet, and the host doesn't seem to have requested any new work yet.
I doubt that this is caused by a hard-drive transplant.
First off, he's running Windows. So, this would only work if he's adding the old drive as a secondary drive in the new computer. (Unlike Linux, putting it in as a primary boot drive won't work with Windows. The hardware won't get configured correctly.)
I've done this repeatedly with nVidia boards without issue. Unlike the competition nVidia's used a fairly unified driver model on all their nForce boards. While you might not have everything working perfectly after the swap until you install the ones for the new one, I've never had a problem getting a stable boot to do the actual install itself.
Most other brands are hit and miss with a moboswap, but since nVidia boards are consistantly among the fastest and most reliable out there I don't have any reason to buy the competition anyway.
Added to this, if you swap to a motherboard that has a different chipset, all you have to do before you shut down on the old motherboard is to set the hard disk controller drivers to the Windows standard drivers. Then move the drive to the other system, boot up and install the new drivers for the controller used on that board.
Or use an external controller card and move that one along to the new machine. No problem booting Windows in either of those ways.
I'm wondering what happens if you put your clock back a week, minutes (or seconds in this case) before sending off a result that's been crunching for all that time. Is the 'received' time set by the Einstein server, or is it sent with the result when you upload/report it?
I'm wondering what happens if you put your clock back a week, minutes (or seconds in this case) before sending off a result that's been crunching for all that time. Is the 'received' time set by the Einstein server, or is it sent with the result when you upload/report it?
Received time is set by the project server.
I was looking at the latest Scheduler log entry for the host, and it also reported an excessive CPU time warning for the two results just reported. I had saved the log from the last session for this host and when looking it over again, I noticed when it firt contacted the project that time it cycled through a couple of Host ID's which aren't listed in the account summary anymore.
This is clear evidence of an ID respawn and is almost always due to the RPC sequence numbers getting out of sync. The unusual thing here is it cycled through two of them in this session.
Might be, what do we know
)
Might be, what do we know about this guy's farm? The varying crunch times (the realistic ones I mean) would indicate different boxes. No way to say for sure, of course, unless he tells us.
@Donald: Sorry, maybe I was a bit unprecise here... I was not talking about having a single internet connection and a switch (or router) for multiple computers. That would be a perfectly normal setup (I use it, too). What I meant was that for example you would only plug two of your boxes into said switch but have three other boxes which are not attached to the internet at all (for whatever reason). Then you would download WUs on one of the boxes that are online and split them between the three offline hosts. When they are crunched, you would put them back onto the flash drive or whatever and use the box with internet access to upload them. I've never actually done it but it is widely accepted as a possible practice, as long as the host for receiving and uploading a result is identical.
Annika I've seen one or two
)
Annika I've seen one or two people here post that they were doing that with work boxes that weren't directly on the net for security reasons. We have a number of machines like that where I work, but even ignoring the installation of unauthorized software the USB sneakernet would get me as thoroughly stomped on as I would for doing a direct net connection.
I wouldn't do it on my boss's
)
I wouldn't do it on my boss's computers, either. But as you said, some people do. And it might still be his private boxes and a network problem (maybe he is really using a modem and has to get a new switch before he attaches the boxes). I would even suggest doing the network stuff from a hotspot but this computer doesn't look anything like a laptop so I'd rule this out.
But even if it's the company's boxes, that's nothing the BOINC admins would have to get active against, is it? I mean, you can be against it morally and/or in terms of getting in trouble but it's not cheating...
RE: @Donald: Sorry, maybe
)
Actually, I realized what you meant after I made the post, but didn't take the time to go back and correct it.
But, even if he were copying the project folders from other machines, I would think that the compute times would be part of that, and would copy over as well.
Yeah, but if he messed up,
)
Yeah, but if he messed up, and caused a resend, that might be overwritten, don't you think?
-
)
-
Hmmm... 1010940 just
)
Hmmm...
1010940 just reported the last two it had in progress at 16:00:27 UTC and the runtimes are nominal for what I'd epect for this PD. The Last Contact info hasn't updated yet so I haven't been able to look at that yet, and the host doesn't seem to have requested any new work yet.
Alinator
RE: RE: I doubt that
)
Added to this, if you swap to a motherboard that has a different chipset, all you have to do before you shut down on the old motherboard is to set the hard disk controller drivers to the Windows standard drivers. Then move the drive to the other system, boot up and install the new drivers for the controller used on that board.
Or use an external controller card and move that one along to the new machine. No problem booting Windows in either of those ways.
I'm wondering what happens if you put your clock back a week, minutes (or seconds in this case) before sending off a result that's been crunching for all that time. Is the 'received' time set by the Einstein server, or is it sent with the result when you upload/report it?
RE: I'm wondering what
)
Received time is set by the project server.
I was looking at the latest Scheduler log entry for the host, and it also reported an excessive CPU time warning for the two results just reported. I had saved the log from the last session for this host and when looking it over again, I noticed when it firt contacted the project that time it cycled through a couple of Host ID's which aren't listed in the account summary anymore.
2007-09-24 16:27:18.1558 [PID=3403 ] [debug ] REQUEST_METHOD=POST CONTENT_TYPE=application/x-www-form-urlencoded HTTP_ACCEPT=*/* HTTP_USER_AGENT=BOINC client (windows_intelx86 5.8.16)
2007-09-24 16:27:18.1558 [PID=3403 ] [debug ] CONTENT_LENGTH=136587
2007-09-24 16:27:19.6817 [PID=3403 ] [normal ] Handling request: host 1010926, platform windows_intelx86, version 5.8.16
2007-09-24 16:27:19.6817 [PID=3403 ] [debug ] OS version Microsoft Windows XP Professional Edition, Service Pack 1, (05.01.2600.00)
2007-09-24 16:27:19.6879 [PID=3403 ] [normal ] [HOST#1010926] forwarding to new host ID 1010929
2007-09-24 16:27:19.6884 [PID=3403 ] [normal ] [HOST#1010926] forwarding to new host ID 1010940
2007-09-24 16:27:19.6884 [PID=3403 ] [debug ] Request [HOST#1010926] Database [HOST#1010940] Request [RPC#1] Database [RPC#1]
This is clear evidence of an ID respawn and is almost always due to the RPC sequence numbers getting out of sync. The unusual thing here is it cycled through two of them in this session.
Alinator
RE: Yeah, but if he messed
)
Well, could be, I guess. I have to admit that I'm at a loss here.