After re-reading it through I have a thought. Create a vanilla server, add a couple of maps or two to it. Create a new folder on the webserver and put the custom files there. Run it.
My thoughts are this. The old saying goes that once you've eliminated all the possibilities whatever is left is the obvious answer. The obvious answer to me, from a 3rd person perspective, is the files on the site differ from the files on the server. If the new vanilla server doesn't show this sign then the answer becomes simple enough to understand. The clients join the server, the server tells it what map is running and gives them a list of files needed to join. If they have the files then all good come right in. If they don't then they're sent to the site to get them. Somewhere along the line the files don't jive. If they don't have the files they get them from the site. The site gives them the files. The server checks those files against the ones running on the server. They get wrecked for differing files. Who's at fault? The odd thing is that the server is changing its files after running. Somewhere along the line the connection between the server and the site is mixing up files. To eliminate a hardware issue run a server without custom files. If it borks then your servers are busted. (Unlikely) If they work then add a couple maps and put up a new place to pull them from with that being a direct copy of the files from the server. Don't get creative and bzip or anything fancy...just put em up for pull. The server will create ztmp files for anything not on the site anyway so don't worry about completeness at this point..just testing. I think that this will be the road that you find your issues. Another option is to create your typical setup of a server and pull the custom files for it from the web server. This should make all the files the same eliminating any discrepancy. If all that don't get you somewhere then I can't think of what else could. Rayne -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of [EMAIL PROTECTED] Sent: Saturday, January 14, 2006 1:25 PM To: hlds@list.valvesoftware.com Subject: Re: [hlds] Map Differs Offer This is a multi-part message in MIME format. -- [ Picked text/plain from multipart/alternative ] Sorry for the late reply. Both 1U`s are with Server Central.net located in the Equinox DC facility in Chicago. I do not think their would be a problem on that end as we have just moved our box`s to them recently and have had the same problem with other collocations at that facility but it is some thing I will look into. The 1U in question ran Memtest for 10 hours with no errors, Prime95 ran for 6 hours with no errors and no signs of over heating before it was taken back to Chicago. Unfortunately had to cut Prime95 test down due to some other things going on. As far as the zmtp files go as soon as Steam let us set a download url we have used that option for serving all custom content. Gary I made all the .bsp`s read only but that did not work, I will try the directory. We are using a 3ware Escalade 7000/8000 controller with the 1.15.0.14 driver. Dan we currently use the IntelR PRO/1000 MT Server Adapter with the IntelR PRO/100 S Server Adapter disabled. Chorizo we have noticed that MD5 signatures have changed on the maps with an error, but to be honest I have no clue on what to do about that. I will have to start researching. I appreciate all the tips/help given because as of last night (1.13.06) it is even worse now then before. When we find the problem/solution I will fulfill my offer one way or another. > From: <[EMAIL PROTECTED]> > To: <hlds@list.valvesoftware.com> > Date: Fri, 13 Jan 2006 15:09:48 -0600 > Subject: [hlds] Map Differs Offer > Reply-To: hlds@list.valvesoftware.com > > The Famous "Your Map de_dust differs from the server" ERROR. > It happens with all stock or at one time or another any custom map we run in > no recognizable pattern. > > We have had this annoying problem for a long long time on 2 1U box's we use > to run CSS, DOD and CS servers on. I do not know if its a Steam issue or a > OS issue of some sort. This has nothing to do with the client side files, > FTP programs or the Bz2 files hosted for downloading on a separate website > so please do not even bring it up. > We have a fresh install of Server 2000 as a set of new HD`s were just added. > All the server files were downloaded directly from Steam with the > hldsupdatetool and a NEW "STOCK" Source server was setup and running. There > was no FTP program installed or used for these files. > We decide to test the server with the stock install as we hoped we had > corrected the issue on this box, but it was not running more than 5 hours > when we started getting "Your Map any_stockmap differs from the server" Out > of the 17 Stock maps 8 of them were then updated by hldsupdatetool with > "update -game "Counter-Strike Source" -dir D:\CSS\AnB -verify_all -retry" > command to fix the problem as we have done this so many times before. > On Custom Map differs we delete and upload to the server again. This can > happen on a daily basis to any maps or just the same one for a while and > some times can go a week or so with out happening. > At about 6:30 pm this evening on the last differ of "de_dust" we deleted a > members client gfc content and then updated, connected to the server and > still had a differ problem. > I do not know if it`s an issue with our 1U`s as they were identical up to > the fresh OS install and adding new HD`s to 1 of them. > Both have Dual 2400`s Tyan Mobos, 2 GB ECC ram and run Server 2000 > 1 has Dual 7200RPM IDE > 1 has Dual 10000RPM SATA > > I am hoping someone you has had this same problem will be able to provide a > definitive answer for this so I am offering a finders fee of $50.00 dollars > sent via PayPal or Western Union to the first person who supply's me with a > solution on how to fix this very annoying problem. I will consider it > money well spent! > > I have gone thru a good number of Search topics to know we are not the only > server with this problem and as I am not a subscriber to the Linux list I do > not know if they also have "map differs" issue`s. If they do not suffer from > these same problems a change of OS may have to be considered. > > If you just have a Smart Ass Comment/Suggestion or just want to bash > Steam PLEASE do not reply. > --__--__-- > > From: "Kevin Ottalini" <[EMAIL PROTECTED]> > > This can happen with some routers, especially Actiontec and Linksys. Newer > firmware is not a guarantee of improved functionality. > > Bypass the router and run the server directly off the modem if you can to > test this. Disabling on-board router firewalls can improve the symptoms but > may not solve it. > > Because this is appears to be a chronic problem for you even with new OS > installs, there is also a chance that your 1U cases are having cooling > problems (not unusual) - check and see if the problem only occurs after > running for a while. > > Try removing the lid and add some additional external cooling temporarily > and see if things don't stabilize. > > I've also seen problems like this happen using a hub instead of a switch > (always use a switch), and I have also seen problems like this when the > servers were running mixed speeds (gigabit to a switch that then feeds a > 100MBit router/modem), try forcing all connections to run at 100MBit (or > make all connections match the lowest speed on the network) and see if that > helps. Bad NICs can also cause this, try a different NIC if you can. > > Be sure to delete all the .ztmp files before starting a new test run, > corrupt data in the .ztmp can cause map differs problems as well. > > Two last things, be sure to run memtest86 on the servers for at least 3 > passes to identify lurking memory problems (again, not unusual): > http://www.memtest86.com > bootable floppy and CD ISO images are available. > > Also, to identify lurking thermal problems in your systems, run the prime95 > torture test - any system must be able to run this without errors or > overheating: > http://www.mersenne.org/freesoft.htm > The latest release is Version 24: > ftp://mersenne.org/gimps/p95v2414.exe > > go to "just stress testing" and then "options" then "torture test" and > select the in-place large FFTs ... I highly recommend running your > motherboard's temp monitor software while you do this. > > Be sure to test with the 1U lid off and then again with it on and see what > the difference in temperature is. > > good luck! > > --__--__-- > > From: Gary <[EMAIL PROTECTED]> > > Make the directory read only :) > > I'd run prime95 to see if you get any errors. I don't use tyan > products anymore because of the high failure rate/issues (the i8254 > was acting strange on the last tyan server I had) > > Curious to know what type of SATA controller you have on that board, > though. I used to have a server with similar issues with files > becoming corrupted at random times due to an issue with the chip/driver. > > --__--__-- > > From: Dan Sorenson <[EMAIL PROTECTED]> > > At 02:07 PM 1/13/2006 -0800, Kevin wrote: > >>I've also seen problems like this happen using a hub instead of a switch >>(always use a switch), and I have also seen problems like this when the >>servers were running mixed speeds (gigabit to a switch that then feeds a >>100MBit router/modem), try forcing all connections to run at 100MBit (or >>make all connections match the lowest speed on the network) and see if that >>helps. Bad NICs can also cause this, try a different NIC if you can. > > Just an FYI, but a switch looks at your ethernet MAC address > and passes directed traffic to it. A hub basically passes all traffic. > That's not the bad part. The bad part is that ethernet essentially > works for voltage levels using four wires, so if two devices talk at > the same time the voltage goes up and they detect a collission. That's > how ethernet determines if what was said might have been unclear. > > In a hub, everybody gets what everybody else gets. The only > way to detect a collission is to run at half-duplex and listen for > that voltage spike. With a switch, you can run full-duplex and not > fret because the pair that look for collissions can now be used to > transmit data. Naturally, you can't use one pair and two pair in > a hub without clobbering this feature. > > As a side effect, there's no real standard for auto-negotiation > on a switch. You can take a $200K Nortel Passport 8600 full of 10/100 > ethernet ports and get duplex mismatching. You can do the same with > a generic 4-port unit. If you can manage the switch, lock the port at > 100 Full and do the same in your network card properties. This will > solve a lot of CRC errors, bad frames, etc... If you're into a hub, > lock at max port speed and half duplex for the same reasons. > > If you have gigabit, don't worry -- it uses all 4 pairs so > duplex issues aren't a problem. > > You'd be amazed at how many latency issues are solved by > simply setting port speeds the same on both switch and server. > > - Dan > > --__--__-- > > From: "Visitor Rodgers" <[EMAIL PROTECTED]> > > Is it possible you have a security breech and someone/process is > altering the files? You can verify the files are changing by taking a > snapshot of their MD5 signatures after a fresh download. As soon as you > get the error, run the snapshot again and see if the files have actually > changed. > > This wont tell you who/what changed the files, but it will confirm if > they have actually changed. > > There are other options to track down who/what is changing things... > > The other recommendations to check hardware are all well advised and you > should eliminate obvious hardware problems before looking for spooky > stuff. > > Hope this helps. > > Chorizo Omelet -- _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds
