Webi Dying

server administration Web Intelligence 1.x/2.x Servers
1

We are about to let loose the managers on a simplified universe to be able to create their own reports. In testing I am creating simple reports that run just fine. I will go back and edit the report and when I try to apply changes or try to rerun the report it will SOMETIMES simply “hang.” While the report is hung there no one can log into webi and I can’t kill the session because the java applet just sits there and does nothing. This does not happen every time I try and apply changes or refresh the report just occasionally. Any ideas? Right now the only way to fix it is to stop webi and restart it. Also I SERIOUSLY doubt reinstalling will fix anything :slight_smile:

Webi 2.5.1
IE 5.01
NT Server SP 5

Ian

Ian Humphries
Decision Support Specialist
PowerQuest Corp.
Phone : (801) 705-8527
Fax : (801) 705-4527
Email : ian.humphries@powerquest.com

Listserv Archives (BOB member since 2002-06-25)

2

I have to stop and restart the webi service usually 3 times a week due to the fact that it “dies”. People can reach the login page and enter in their username and password but then it just sits there untill the page times out. Stopping and restarting the service brings it back into working order but its getting tiresome for not only me but my users :\ Any ideas?

Webi 2.5.2
Win NT 4.0

Ian

Listserv Archives (BOB member since 2002-06-25)

3

Ian,

What version of B.O. do you have loaded on the server?

Aris

Listserv Archives (BOB member since 2002-06-25)

4

How often do you reboot the whole mess? We go weekly, that helps

Brent

I have to stop and restart the webi service usually 3 times a week due to
the fact that it “dies”. People can reach the login page and enter in their
username and password but then it just sits there untill the page times out.
Stopping and restarting the service brings it back into working order but
its getting tiresome for not only me but my users :\ Any ideas?

Webi 2.5.2
Win NT 4.0

Ian

Listserv Archives (BOB member since 2002-06-25)

5

Ian,
Seems like we had a similar problem a while back. Have you tried checking the services user ID and passwords? Also check the DCOM Config. We had a problem with the WebIntelligence Manager and WebIntelligence Orb services. We had to have these services Log On as a valid user ID and password with the appropriate administrative rights.

Also If you might want to run the Servconf.exe utility Click on the Run DCOM Config button and check the default security. The account you are using for the services needs to be listed here.

One other thing might be getting in the way, Are these problems happening after a system backup? If so have the NT scheduler stop and restart the services after the backup runs.

Hope this helps.

James F. Lafferty
AERA Energy LLC
Applications Specialist
Office: 661.665.5126
Pager: 661.665.7566
e-mail jflafferty@aeraenergy.com

Listserv Archives (BOB member since 2002-06-25)

6

I also had problems like this.

Make sure the administrator ID is “CLEAN”, i.e., it does not run an automatic logon script that activates any services that you do not want to run. I was logging on to the WEBI server with my normal user ID which had administrator rights on the WEBI server. My normal LAN logon script attached 12 network drives with 200+ gig of storage and launched Norton Antivirus. Norton Antivirus scanned all those network drives. One virus was found, and no one could logon to WEBI until the server was rebooted.

If you can’t get a separate ID just to perform administration on the WEBI server, then after you logon to the server, go into My Computer immediately and disconnect any drives that were connected as part of your logon script.

Use an administrator ID that does not launch any extraneous services.

Don’t run any Antivirus software.

Reboot on a regular schedule and make sure your users know that schedule.

Listserv Archives (BOB member since 2002-06-25)

7

How many computers are you running??? Is it a cluster?

If you are running a cluster what I have seen is the node gets a “ghost” WIQT process that uses all the memory and cpu until the node dies. The only process is to reboot.

This problem has been logged and I am activly working with Tech support and Paris to resolve the issue.

Stay tuned…

Sincerely,

Tom Nather
Data Warehouse Group
Penske Logistics
email:tom.nather@penske.com
Phone: 216-765-5787
Fax: 216-765-5666

Listserv Archives (BOB member since 2002-06-25)

8

Hi all,

I have 2, somewhat related issues, any help in this regard would be great:

We too notice that once every week or so that the performance begins to degrade; and when that happens I see a bunch of WIQT processes just lying there. I know one solution is to bounce the server; We do that. but Do you think that just killing all the WIQT processes will achieve the same purpose ?

And as for the second issue - this concerns retreiving large reports in WebI. I notice that in good times, I am able to retreive reports, as large as 4-5MB large, quite easily. Is this size the norm ? What I notice is that when the server begins to die, I can no longer retreive these large (?) reports and was wondering if these are what cause the server to hang, albeit indirectly. Say a user is trying to retreive a report but since its taking long, just clicks stop, and reloads, thus starting a vicious circle.

Any help in this regard would be appreciated.

Savio

#-----Original Message-----

#How many computers are you running??? Is it a cluster? #
#If you are running a cluster what I have seen is the node gets #a “ghost”
#WIQT process that uses all the memory and cpu until the node #dies. The only
#process is to reboot.

#This problem has been logged and I am activly working with #Tech support and
#Paris to resolve the issue.

#Stay tuned…

#Sincerely,

#Tom Nather
#Data Warehouse Group
#Penske Logistics
#email:tom.nather@penske.com
#Phone: 216-765-5787
#Fax: 216-765-5666

#-----Original Message-----

#How often do you reboot the whole mess? We go weekly, that helps #
#Brent

I have to stop and restart the webi service usually 3

#times a week
#due to

the fact that it “dies”. People can reach the login

#page and enter
#in their

username and password but then it just sits there

#untill the page
#times out.

Stopping and restarting the service brings it back into working

#order but

its getting tiresome for not only me but my users :\

#Any ideas?

Webi 2.5.2

Win NT 4.0

Ian

Listserv Archives (BOB member since 2002-06-25)

9

We recently did some lab trials and found exactly the same symptom. Even if you kill the process, CORBA restarts it and the cycle goes on. We let one run on to see what would happen but gave up when it reached a size of 800Mb! We did discover that it could be corrected without a reboot. You have to stop Webi on the server in question and then use Task Manager to kill the rogue process. Because Webi is no longer running, the process is not restarted.

At the time, we had one of BOs UK Tech. Product Managers on site and he has also passed details to Paris (no resolution yet).

We are about to commission some new servers and I am thinking of requesting a weekly reboot for all of them. Any thoughts?

Regards

Graham Boden
EDS

Tom wrote:
If you are running a cluster what I have seen is the node gets a “ghost” WIQT process that uses all the memory and cpu until the node dies. The
only
process is to reboot.

This problem has been logged and I am activly working with Tech support
and
Paris to resolve the issue.

Tom Nather

Listserv Archives (BOB member since 2002-06-25)

10

this is all very interesting - we have experienced most of this for a year or so…

we boot our clusters every week, after a full back up… managers then the nodes a few minutes later… seems to help, or if we dont do it, sort of guarantees a bad week afterwards.

brent

From: Boden, Graham (EDS) [SMTP:Graham.Boden@IRD.EDS.COM] Sent: Tuesday, May 09, 2000 5:22 AM

We recently did some lab trials and found exactly the same symptom. Even if
you kill the process, CORBA restarts it and the cycle goes on. We let one run on to see what would happen but gave up when it reached a size of 800Mb!
We did discover that it could be corrected without a reboot. You have to stop Webi on the server in question and then use Task Manager to kill the rogue process. Because Webi is no longer running, the process is not restarted.

At the time, we had one of BOs UK Tech. Product Managers on site and he has
also passed details to Paris (no resolution yet).

We are about to commission some new servers and I am thinking of requesting
a weekly reboot for all of them. Any thoughts?

Regards

Graham Boden
EDS

Tom wrote:
If you are running a cluster what I have seen is the node gets a “ghost” WIQT process that uses all the memory and cpu until the node dies. The
only
process is to reboot.

This problem has been logged and I am activly working with Tech support
and
Paris to resolve the issue.

Tom Nather

Listserv Archives (BOB member since 2002-06-25)

11

I have to stop and restart the webi service usually 3 times a week due to the fact that it “dies”…

Ian

Listserv Archives (BOB member since 2002-06-25)

12

heh what a pain…I have asked out IT dept to schedule the box to bounce weekly…crossing the fingers.

–
Ian,

We are experiencing the same problem. First our configuration:

Webi 2.5.2
NT 4.0 svc pack 4
IIS 4.0
Oracle 7.3.4 running on a HP 9000 system

We have a Master install on a dual 400 mhz Pentuim III Compaq Proliant 5500 server and a Node install on a quad 200 mhz Pentuim Pro Compaq Proliant 5500 server. RAM is 640 MB and disk storage of 25 GB on both machines.

We were experiencing problems where the ‘system’ is down in the mornings. On a side note, the Oracle database is brought down for nightly backups and the HP9000 is rebooted to clear memory/buffers.

The problem centered around Webi allowing you to login, but you could not open a report.(Other times WEBI is down completely, meaning no logon screen is available). To fix this problem, we would shutdown both systems, Node then Master. Then we would start the Master and then bring up the Node. We were doing this manually every morning until we set up two automated processes using the ‘AT’ command.

We use ‘AT’ to shutdown and restart Webi. The WIMANAGER and the WIORB processes are stopped and started using a ‘net stop’ and ‘net start’ command.

This seemed to fix our problem, but then we starting experiencing problems with the ‘system’ being down every other day, in no particular frequency.

Using the NT Resource Kit CD, we are shutting down Webi using the before mentioned process. This is done before the Oracle database is brought down.

After the HP9000 is backuped and rebooted, we then reboot both the Master and the Node server, using the ‘Shutdown’ command located on the NT Resource Kit CD, and then start Webi using the before mentioned process. The ‘AT’ command drives all of these processes.

Currently our situation has stabilized and we are not having the same problems with Webi not working in the mornings.

We also, on occasion, see the similar issue that Webi goes down (Dies) during the day, and we have experienced the phenomenon where a WIQT process starts taking all of the RAM and not releasing, which brings ultimately brings Webi down. This situation normally takes place on the Node server. To correct this problem, we shut the Node down, and reboot the system. It seems to correct the problem, until the next time it does it. We have logged a call to Tech Support, but to no avail…

Is anyone else out there encountering these types or problems??? If so, what are you doing to correct/work around them?? Any and all information would be appreciated…

Listserv Archives (BOB member since 2002-06-25)

13

man, we had a WEIRD version of this yesterday - kind of a runaway…

by the time we killed BO, the single node in our 2 box cluster had almost 200 WIQTs on it (for 5 users)… it was weird.

the manager went down shortly afterwards (after rebooting the node) - had to reboot both boxes again to clear up the problem.

seen this before?? anyone? nt4, sp4, webi 2.51 sp 32, 33

brent

From: Ian Humphries [SMTP:Ian.Humphries@POWERQUEST.COM] Sent: Wednesday, May 10, 2000 10:19 AM

heh what a pain…I have asked out IT dept to schedule the box to bounce weekly…crossing the fingers.

–
Ian,

We are experiencing the same problem. First our configuration:

Webi 2.5.2
NT 4.0 svc pack 4
IIS 4.0
Oracle 7.3.4 running on a HP 9000 system

We have a Master install on a dual 400 mhz Pentuim III Compaq Proliant
5500 server and a Node install on a quad 200 mhz Pentuim Pro Compaq Proliant 5500 server. RAM is 640 MB and disk storage of 25 GB on both
machines.

We were experiencing problems where the ‘system’ is down in the mornings. On a side note, the Oracle database is brought down for nightly backups and the HP9000 is rebooted to clear memory/buffers.

The problem centered around Webi allowing you to login, but you could not open a report.(Other times WEBI is down completely, meaning no logon screen is available). To fix this problem, we would shutdown both systems, Node then Master. Then we would start the Master and then bring up the Node. We were doing this manually every morning until we set up two automated processes using the ‘AT’ command.

We use ‘AT’ to shutdown and restart Webi. The WIMANAGER and the WIORB
processes are stopped and started using a ‘net stop’ and ‘net start’ command.

This seemed to fix our problem, but then we starting experiencing problems with the ‘system’ being down every other day, in no particular frequency.

Using the NT Resource Kit CD, we are shutting down Webi using the before mentioned process. This is done before the Oracle database is brought down.

After the HP9000 is backuped and rebooted, we then reboot both the Master and the Node server, using the ‘Shutdown’ command located on the NT Resource Kit CD, and then start Webi using the before mentioned
process. The ‘AT’ command drives all of these processes.

Currently our situation has stabilized and we are not having the same problems with Webi not working in the mornings.

We also, on occasion, see the similar issue that Webi goes down (Dies)
during the day, and we have experienced the phenomenon where a WIQT process starts taking all of the RAM and not releasing, which brings ultimately brings Webi down. This situation normally takes place on the Node server. To correct this problem, we shut the Node down, and reboot the system. It seems to correct the problem, until the next time it does it. We have logged a call to Tech Support, but to no avail…

Is anyone else out there encountering these types or problems??? If so, what are you doing to correct/work around them?? Any and all information would be appreciated…

Listserv Archives (BOB member since 2002-06-25)

14

man, we had a WEIRD version of this yesterday - kind of a runaway…

brent

That is very similiar to one of the problems that we are experiencing. But the frequency of this is about 3 times a week. Could be a difference in our user base, server workload or ???

Listserv Archives (BOB member since 2002-06-25)