Sage 100

Connection failure to host:[tcp]Servername;####;NODELAY;STREAM"" appears when starting tasks. Started getting the above error on a couple users on a customers 4.4 MAS200 install. They have been running on the same server for awhile. I know the article on Sage KB says to do the following: Perform the following: === Have all users exit Sage MAS 200. Log on to Sage MAS 200 as Administrator. From Administrative Tools, select User Maintenance. Select the first user, and click the Preferences tab. Select the Spawn Tasks from Application Server check box, and click Accept. Proceed for remaining users. Click Done. === However, no explanation on the reason for making this change to the system. Any idea what would all of a sudden cause this to begin happening? Why does spawn tasks from Application server need to be set and do I have to do it for ALL users?

Have seen this a lot when running MAS 200 as service and load of 15-20+ users (user count is approximate).

We have as well. When we changed it to Spawn tasks from server, it caused a whole host of issues. We ended up running it as an application.

Ah...thanks Wayne and Therese. I may have to convince the IT guy to do that. He wanted to run it as a service since running as an application won't auto-start without someone logging in as user on the server after a reboot. Someone on site would have to login to the server anytime it is rebooted. I will give that a go and report back...thanks.

We have also had some success with what is outlined in SS3003-ABC. http://support.sagesoftwareonline.com/mas/query/index.cfm?action=abcdetail&abc_id=210

There's a way around that @BrianBuck - I know customers have thrown it into the autostart folder then they have a way to get it to log in. Someone will jump in with details I'm sure. We've had a few hair pulling issues with customers repeatedly calling with this problem and running it as an application made the issue disappear.

Thanks @WilliamSchlageter I didn't realize that article had been updated to include Windows 2008 server. I am hesitant to muck around with the registry on their server but I am going to push their IT to run as an application. Just need to Google how to set up autologin on a Win2008 server...or if someone knows how please pipe in. Thanks.

I've seen that one too @WilliamSchlageter. Good suggestion as well.

I'm asking my customer who has a 30 user MAS200

Alnoor has done extensive work on this.

Brian, I agree with running it as a Application, I have seen this plus if you happen to have long Custom Crystal Report names, it would crash MAS200, lock it right up, SU 8 for Version 4.4 fixed that issue of long names plus we are running it as an application and have checked on spawn tasks so a combination of things.

@RonPeterson are you saying I should have Spawn tasks on application server turned on for all users (user maintenance)? @ThereseLogeais said that caused her problems when she enabled that.

I think I found my answer to the Autologon...Windows sysinternals tool...http://technet.microsoft.com/en-us/sysinternals/bb963905.aspx I am going to test on my Win2008R2 virtual machine first. Then the fun part...get the IT guy to approve it.

We turned off the Spawn tasks option and just ran it as an application. Maybe @AlnoorCassim can provide his take on this?

Brian, I am saying that, though as with Therese, we are running it as an Application, not a service. Our main issue though was it we had Spawn Tasks Off we could not run the Crystal Reports, turned some reports ran and the ones with longer names crashed MAS200. Discussing with Sage they said the issue with Crystal Names is fixed and we should check on in User Maintenance Spawn Tasks.

Interesting @RonPeterson. Are you referring to trying to run the reports from a Custom Reports menu if they have too long of a path? I know we have had issues adding them if the path is too long but nothing regarding the actual names when running them.

I found that it was the sum of the path/filename and name given to the menu choice that would cause the error. I would generally try to make both as few characters as possible.

Therese, what we had is we had our customer who created maybe two dozen or more custom reports, they were on EES 1.3 MAS90. We moved them to MAS200 4.5, and they were getting a no delay stream error. We turned on Spawn Tasks which resolved the no delay stream though we started getting Error 41. Basically the command line was too long i.e. the Custom Report Name. When we changed one and shorten it we did not get Error 41. The customer did not want to change all her custom names and at this time while talking with Sage SU 8 fixed the named issue. So we have Spawn Tasks on, eliminates the no delay stream, SU 8 on, eliminates error 41 and having to have a shorter name for the custom reports. So far all is well except we have pinpointed a few of her custom reports are built incorrectly which she is fixing. Hope this helps explain what process I took and for what issue.

+1 to the Ss-3003. I always implement it on our customers with MAS 200. We rarely have any issues running 200 as a service.

Jim, do you have a favorite value that you change the setting to or do you ""tweak"" the settings before arriving at one that works?

Just arriving here. First, let me say everyone has posted all the answers already. At the bottom I've answered what Spawn Task actually means, whether you should turn it on for ALL or some users, and whether everyone needs to be out of MAS when changing the setting. Let me clarify the ""Connection Failure .. NODELAY;STREAM"" error is in fact different than the issue of running the App Server as a service vs desktop application (I'm the primary author of SS-3003 but we can we start a different thread for that issue). It's confusing because the error you get when you run as a service is ""Connection Failure .. NODELAY;STREAM - Timeout Exceeded"". They're almost the same. @BrianBuck - Yes use AutoLogon from SysInternals !! Point out to the IT guys it's officially a Microsoft utility and written by the same guy who wrote ProcessMon, RegMon, FileMon, etc (the amazing Mark Russonivich who was not with MS at the time he wrote them) @RonPeterson - you are right on the money. Let me summarize with Spawn Tasks ON: PROs: * Resolves the NODELAY;STREAM error * Resolves random ""xx user limit exceeded issues"" * Allows MAS 200 to open tasks properly on Server 2008 when Windows Firewall with Advanced Security is running CONs: * In few instances, slower task load times (see below) * Error 41 or a certain msg appears when running certain Custom Reports from the menu. @WilliamSchlageter is correct it was happening because Sy_Task file was storing the full path and filename to the rpt (didn't need the full path) and length of the menu description was a factor (both were passed as arguments to the App Server for running the custom report task). Prior to PU8 you have to either shorten the rpt name desc or DFDM Sy_Task ---------------------------------------------- What does Spawn Task mean? ---------------------------------------------- From 4.00 to 4.05 there was no explicit setting with this name but it was effectively on. It means the Application Server does the client/server handling and communication of both the Launcher task (when you start up MAS) and each subsequent task run off the Launcher (e.g. S/O Entry). In 4.10 beta, they were working on performance issues related to task load times and found by using the 3.x Host handler program called _windx.soa instead of the App Server it would load S/O Entry much faster (Keep in mind this is before they discovered (me included) the 50+ other BF programming flaws that were causing slow performance). So now in 4.10+ Spawn Task explicitly exists and every new install defaults to it OFF. This means only the Launcher itself gets handled by the App Server, all tasks off the Launcher are handled in 3.x style by _windx.soa (IOW we have mixed technologies). With it ON both types of tasks are handled by the App Server. Either way, it doesn't affect the total number of pvxwin32.exe tasks that run on the server just what you would see in App Server Config in Sessions tab. With it OFF, I started fielding cases on issues above. Worked with the ProvideX guys in Canada but we didn't get far because we couldn't reproduce NODELAY;STREAM error. To this day all I know is an internal untrapped communication error does not occur for most clients but for those affected it happens regularly during periods of heavy MAS activity. Often happens when previewing forms and reports. Sorry I don't know any more @BrianBuck - I do know that that NODELAY;STREAM can result in loss of a license slot and is one of the main reasons for ""xx user limit exceeded"" With it ON I found only a few actual cases of slower performance (by this time, many 4.10 slow perf bugs were fixed) so ON became the regular advice. ------------------------------------------------------------------------------ Should you turn it ON for ALL users or affected users only? ------------------------------------------------------------------------------ Either set ALL users for ON or OFF but no in between! Otherwise the users that are OFF are adversely affected by users that have it ON. Also I like turning it ON via DFDM on Sy_User for the 3 users you can't see in User Maintenance (DefaultUser, _ExternalURL, and _ExternalEBM ) ------------------------------------------------------------------------------ Do you need all users out of the system to make the change? ----------------------------------------------------------------------- Yes ALL users need to be out of system.

If @AlnoorCassim is willing we should make this a whitepaper for distribution to our more technical customers -- or just to hold in our tool bag when this issue arises. I know that @BrettZimmerman and I ran into this a year or two back out of the blue during a large upgrade. My experience is that this tends to occur with larger user count sites that may be more important customers. It could potentially be ""life saving"" if we have this to provide to our customer to explain the reasoning behind it and the resolution. I don't run into this problem a lot - but when I do it's usually pretty serious and having a cure could make us a hero. I wonder if starting some type of whitepaper in Google Docs would be the easiest way? Assuming all involved in the thread agree.

I would need to remove phrases like ""Keep in mind this is before they discovered (me included) the 50 other BF programming flaws that were causing slow performance

Alnoor, thanks for all the clarification. I would love a whitepaper if you are so inclined to share.

It would be terrific to have a white paper/tech doc on this topic. I learned so much just reading this thread - it would be great if we had it somewhere easily indexed to reference.

Helpful thread. My issue is that when I am connected from my workstation on the LAN, I will get errors with Spawn Tasks ON, when I connect via Remote Desktop I get errors with Spawn Tasks OFF. So I had to setup up two users for myself. One for RDP, one for local. This does not follow the advise above. Any thought as to why this happens?