| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Retired Forums
Forum: Member-to-Member Support [Read Only]
Thread: the task reaches 100% and then without sending the data anywhere starts the next task. |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 51
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I have identified two problems that are currently present on my setup:
1. My PC#2 is resetting 2. The Agent on my PC#2 is sometimes resetting its task without completing the previous one within the data crunching process – i.e. prior to completion of the present task. Let me start of by giving any overview of my setup. I have a SOHO type set up with a cable modem (1 meg down, 256 up) in my office. The cable modem is connected via an Ethernet card #1 in my PC #1. I have a second Ethernet card #2 in the same PC that works as a bridge to a D-link 5 port router. My second PC#2 is connected to the D-link router via Ethernet. All PCs run Norton’s Antivirus. PC#2 has Norton’s firewall on it. PC#1 is using the XP firewall (I know, I know – wrong way round…… maybe someday I will change it) I first started running the wcg Agent on PC #1 and have never had a problem with it. I then went on to install the Agent on my second PC #2 that accesses the internet via the D-link router (connected to the bridge in PC#1). This is the computer that I am having problems with. Up until yesterday, the issue with the PC#2 resetting was critical as the wcg Agent was also resetting – i.e. after the PC reset, the agent would loose all trace of the data it had crunched and download a new data batch. This had me worried as it could mean that data is actually being lost, or in the worst case, resource was being wasted crunching data that was later discarded. I reinstalled some drivers and the wcg agent and now the PC no longer looses data if it resets. ( I wrote about this in a previous post) Problem #2 is still present. It seems to me that if my PC#1 communicates with wcg (e.g. uploads data) then the wcg Agent on PC#2 automatically resets itself. I cannot tell weather it just starts crunching the same data over again or if it downloads a new batch of data from the grid. I am sure that Admins can see the data batch IDs and identify what is happening. My feeling is that there is a problem caused by my network setup and that even though there are two different Device names (that YES – use the same device profile – I have only one profile defined) somehow or the other a conflict in ID takes place that makes the second Agent receive a new batch of data every time the agent on PC#1 receives new data. This is just a hunch, but the two incidents definitely have something in common. It could be something to do with wcg using IP addressing for data routing. In my setup I actually have only one external IP address and the packets are managed within the LAN by XP. Does this makes sense? Is it a known problem or is my imagination doing the “Terry Pratchett” on me?  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Robert --
First of all, thanks for having the stamina to re-enter everything. As an aside, many of us have found that we are better off composing our posts in a notepad off to the side and, when complete, cut/pasting into the forums. Earlier, we had been hiting a timeout and that caused us to do this. I don't know if your network is the cause of this. I too have two systems running the WCG software. My setup is a bit different. My DSL modem connects to a Linksys BEFW11S4 802.11b Router with 4 port switch. My primary system connects by ethernet to the router and my old system has an 802.11b card and uses a wireless connection. Upstream, both would be visible to the WCG servers with the same IP address, so that should not be a problem. I do not know how the bridging in your setup might come into play and am going to request that the WCG technicians take a look at this. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
RobSab,
Urkk!!! I hate network problems, since I never feel confident that I know what I am doing. Even so, because I am brash, I tend to speak up when I should stay silent and let real experts work. The first thing to do when you suspect you have a network problem is to gather information in a form that can be easily transmitted to other people. A good place to start is to look at http://www.microsoft.com/technet/prodtechnol/winxppro/maintain/netdiag.mspx This is titled 'Network Diagnostics Tools Feature Overview'. [This was published on 1 Nov 2001 and some of the feature names used seem to be slightly different in my current Windows XP.] To access the Network Diagnostics Web Page, the instructions are: Click Start, and then click Run. Enter the Help and Support command syntax: hcp://system/netdiag/dglogs.htm and click OK. This can get you started on collecting lots of information and storing it into files.  mycrofth |
|||
|
||||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks mycrofth. Will follow the instructions through.... once I have a sec.
I donot actually think that it is a network problem, as the PC and other PCs on the network (mainly people walking in with TapLops and plugging in, don't have problems. My horrible feeling is that wcg may be using something in a process to download / upload / monitor progress that sees the IP of the eathernet card connected to the cable modem and not the internal ethernet cards of the PCs on my LAN - i.e. it sees one LARGE (not to use another word) PC and not several - thus is trying to get all of the components to do the same thing when they are actually carrying out separate and non related tasks..... so much for my theory. Will check the LAN anyway. ... one never knows with computers... ;o) ... thnks for the tip in getting my spelling right Lewis Caroll ... you have made me feel so much better just knowing that all you guys out there are also making spelling mistakes - just using a spellchecker to correct them  |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I am now 100% certain that when my PC#1 sends data to the wcg, the Agent on my PC#2 gets killed.
I have just observed it - two screens next to each other...... and PC#1 starts transmitting and the Agent on PC#2 gets killed. This time even XP acknowledged the error and sent the error data off to microsoft!! Who knows - maybe there is more to this problem then meets the eye!!! I hope this feedback is usefull. RobertS |
||
|
|
Viktors
Former World Community Grid Tech Joined: Sep 20, 2004 Post Count: 653 Status: Offline Project Badges: 
|
Perhaps you have your systems configured to share an internet connection via the second computer or something. Maybe both are sharing with each other in some bizarre loop. Or maybe both machines have the same IP address? You must have some really weird network configuration problem, to have network traffic on one machine kill a program running on another, which is not even using the network at that time. Perhaps the configuration causes windows to get in some fast allocation loop which consumes all of the resources on the second machine making applications crash. It could even be a weird hardware problem where the second network card goes nuts when it sees a lot of network traffic not directed to itself and possibly incorrectly responds to other IP addresses. All pure speculation, of course.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks Viktors - not very useful speculation unfortunately. My network has been running in this configuration for over two years - we load it up to 5 machines and work on data base entry, download graphic files, use heavily loaded emails and have never experienced this problem before.
I do not want to download the agent onto any of the other PCs as something is not right here and the environment is becoming unpredictable - but only with the Agent. Any suggestions on what to do? Should I give the two machines separate device profiles and set up different access time to wcg to ensure that they do not interfere with each other? RobertS |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
RobSab,
PC #1 is killing a process running on PC #2 when PC #1 communicates with the Internet. PC #1 is on 2 different networks. The first network connects to the cable modem and then the Internet. The second network is through a router to PC #2 and transient laptops. PC #1 is the bridge between the 2 networks. 1) You should not even be communicating with PC #2 when transmitting results from PC #1 to the WCG server. 2) The WCG client UD.EXE contains some code to kill all but one instance of itself on a computer. If it finds several instances running on a computer, it shuts down all but one. I do not know if this is what is happening, but it is the only unusual thing about the client (that I can think of) that might cause it to react to a network problem that all other programs ignore. My suggestion: Remove the WCG Agent from PC #2. When you have time, run the Microsoft Network Diagnostics on your systems, both PC #1 and PC #2. I think they might show something strange about your network. This is a problem unlike any other that has been reported. If you put the diagnostic files into a word processing document, then you can have real network experts (not me) look at them and try to figure out what is wrong. If anybody has some better ideas about handling this problem, please feel free to comment. I always try to get other people to solve network problems.   network dunce (mycrofth) |
|||
|
||||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Thanks mycrofth.
Sounds logical enough to buy. At present I have defined two different device profiles that are exclusive in their access time to wcg. If this doesn't work, then I will do exactly what you have suggested and then publish the two reports for the network gurus to look at. The very thought of having to change something on the network makes me multo  RobertS |
||
|
|
Viktors
Former World Community Grid Tech Joined: Sep 20, 2004 Post Count: 653 Status: Offline Project Badges:
|
You could give them different profiles and make them communicate and run at different times. But, I am still not sure how this could be happening in your setup without being there in person to poke around for clues.
I assume you are not using a shared disk and running both machines from the same installation directory. This would definitely not work. Also, each machine should be installed separately and registered separately. If you copy the install directory from one machine to another, this will not work properly. |
||
|
|
|