[XI 3.1] Has anyone installed SP4 on XI R3.1

We currently are on XI 3.1 SP2 FP2.6 clustered on 2 servers with @800 users. We have significant issues with the connection services and DeskI reports. Restarting it multiple times daily is not really a solution. SAP said all the problems are fixed in SP4. I know that SP3 introduced a lot of other problems and I’m nervous about being the first to try SP4. Has anyone else installed it? Any issues so far?


asnair :us: (BOB member since 2005-08-02)

If you are referring to the connection server, you’ll be (un)happy to know that the problem is not fixed in SP4 as originally planned. The new target is now FP4.1 according to the ticket I have open.


BoB LoblaW :us: (BOB member since 2007-10-23)

Oh great, now I don’t even have a bad idea to try. :cry: Thanks for the info.


asnair :us: (BOB member since 2005-08-02)

Indeed, restarting the services “often” is not what anyone wants to do.
BUT
could we have two, three “connection servers” running in parallel on the same system, and might that help ?

I’m breeding on the following idea:

a) “clone” the connection server/service, and have 2 or three CS’s running in parallel on exactly the same machine.
b) “schedule” a round-robin scheme of :
b1 - disable CS1 (for new conections)
b2 - wait for CS1 to have no connections running any more
b3 - restart CS1 as soon as it has no client process to serve
c) start same sequence for CS2 (CS3 …) once CS1 is accepting connections

Did anyone try this scheme?
And … does anyone have enough expertise with SDK to make a “program” that does this and could be published to the cmc to be “scheduled” from inside BOE ?

More services on same hardware has two reasons :

  • not to violate our “server/cpu licence”
  • not to need an -extra- VM with boxi installed, we have 3 running already (dev, acceptance, prod) and my demand for a fourth one (sandbox with throw-away / experimental stuff) already was rejected … almost (these VMs take up space, and resources, and it 'd be for playing ? This is business, sir, not university )
    :slight_smile:

RensH :belgium: (BOB member since 2007-06-18)

I have 2 servers and currently have 2 CS on each for a total of 4 CS. I have set up a bat file that restarts all 4 CS 3 times a day and still had to manually restart yesterday. I plan to create a third CS on each server today, but based on recommendations from SAP I was told I should not exceed 3 on each server.

I do not know how to create a program that would disable the services first, but that’s a good idea. I’ll follow up with SAP and see if the analyst can provide that info.


asnair :us: (BOB member since 2005-08-02)

I put some code up on BOB to disable / stop / start servers using the SDK and some javascript - it’s probably pretty close to what you need.

I used this to stop and start my Webi reportservers, job servers, on a staggered schedule.

B


bdouglas :switzerland: (BOB member since 2002-08-29)

I initially had a process in place similar to asnair’s. I modified it to scan the connection server log files every 5 minutes for the string “:218” and restart the corresponding server if it is found. This way, you only restart a connection server that isn’t working anyway and don’t have to mess with disabling then waiting for all connections to close.


BoB LoblaW :us: (BOB member since 2007-10-23)

I installed SP4 over the weekend to resolve, according to SAP, an issue that we were having with migrated reports not working in InfoView but working fine in Deski. Not only did this not fix the problem but now most users are getting a “SetPrompt” error when they try to run any reports.


kstadtman (BOB member since 2010-04-21)

I blow away all my webi.exe processes daily via a win scheduler calling a bat file doing an endtask command. Since they’re set to restart automagically in CMC they just come back alive quick like. This was done because we query from BW and it fills up the memory for the process until it pukes and it restarts harshly and often fails to come back up proper. I do it as a proactive measure at an off peak time. It’s not graceful by any means but has cut down on reported hiccups. Seems to restart better this way versus letting system do it’s own crash/restart. Caveat is if anything is actively processing it’ll kill it and user would have to reinitiate…hence the off peak time. We’re somewhat global now which doesn’t give much for off peak time…but I still run it daily 8)

Not sure if similar could be done for your connection servers…but figured I’d throw it out there.


ccermak :us: (BOB member since 2007-11-13)