<html><div style='background-color:'><DIV class=RTE>We are using Spread 4 RC2 from almost 5 months in Windows XP/2003 environment. For client side API we are using C# open source library "cspread". We deploy our solution on 3 computers each running its own spread daemon. For daemon configuration we share a configuration file "spread.conf" from one of these computers so all daemon run from the same configuration file. We do this through -c switch and use network file path like this "<A href="file://\\computer1\spreadconfiguration\spread.conf">\\computer1\spreadconfiguration\spread.conf</A>". On each computer we have three applications that connect to their local daemons and so communicate with each other. Also on one </DIV>
<DIV class=RTE>computer data is continuously arrived through another server by using TCP socket and then publish on some specific group in spread message bus. Second computer applications are listening on that group and </DIV>
<DIV class=RTE>when data is arrived they again publish data on another group on which 3rd computer applications are listening. So there is no idle time for daemons on computer 1 and 2. Data is coming on rate of almost 5 </DIV>
<DIV class=RTE>messages (of 10-50 KB) per second.<BR>I explain all our scenario as it might help to understand our problem.</DIV>
<DIV class=RTE>Problem:<BR>The problem does not occur very frequently but it sometimes occur. Some times spread daemon suddenly crashed. Observed scenarios at the time of crash are:<BR>1) When we kill one spread daemon manualy from task manager the other spread daemons on other machines also crash. Note this does not happen always but sometimes.</DIV>
<DIV class=RTE>2) Extending the point 1 when we kill spread daemon manualy on one machine nothing goes bad but when we start the daemon again it starts successfully but daemon on the other computers crashed.</DIV>
<DIV class=RTE>3) We deploy and start our solution on the target computers through Remote Desktop in Windows. When the solution is started we just close the remote desktop window mean not log off or close the windows </DIV>
<DIV class=RTE>session so our solution continues to running. Now when we again logged in on the same computer through remote desktop but with different user then we see spread has crashed on that machine. Also spread </DIV>
<DIV class=RTE>daemon on second machine crashed but third is running fine. Note this also does not happen always but sometimes.</DIV>
<DIV class=RTE>4) We also deploy our solution on virtual PCs. In this environment the problem in point 1 and 2 occurs more frequently. Again not always but frequency of crash is greater than that of physical computers. </DIV>
<DIV class=RTE>Does it is correct to kill one daemon directly(trough task manager)? or it should be done through spread monitor? Also if we kill daemon manually, does it impact on other running daemons on other computers?</DIV>
<DIV class=RTE>5) When our solution is running for a long time say 2 3 days continuously spread daemon on computer 2 crashed but others are running fine. Note again this does not happen always but sometimes.<BR> Extending this point some time we take the statistic of spread daemon process through Process Explorer( from <A href="http://www.sysinternals.com">www.sysinternals.com</A> ) its memory goes to its limit (1.9GB). But again this also happen </DIV>
<DIV class=RTE>sometimes not always.</DIV>
<DIV class=RTE>Regards<BR>Danish Ahmed</DIV></div></html>