login to Infoview / cmc is slow

[Former Member]

Hi,

we upgraded a month ago from XI R2 to XI 3.1 FP 1.4 on a 2 cluster / Solaris 10 environment.

Now the login times are about 30 seconds - no matter if enterprise or ldap authentication is chosen.

We have a preprod-environment / test / or another standalone server with the same versions and the logins are fast as ever ( 1sec.).

Has anyone experienced this problem ?

[Former Member]

I saw something, that is quite strange - there is a user, identified through ldap absolutely fast login !

He was newly created -

a user also quite new, who is only member in the everyone group is connected really slow.

Probably not the LDAP makes the problem, but the cms DB.

[Former Member]

Use "top" or ps on solaris to show you the disk and CPU utilization for the BOE processes while you are logging on. This can tell you, for example, whether the application server is busy or whether the CMS is busy. If neither are busy it could be a database issue.

You should also try some easy network diagnostics between the various nodes to check throughput and latency. Try some simple file transfers between local filesystems on the machines, and ping from machine a to machine b and then login to machine b and try to ping machine a.

In the production system, is there a firewall or load balancer?

Is the hardware in the pre-production environment exactly the same as in production?

[former_member793810]

Hi Sabine,

Have you compared both server the fast one and slow one - software and hardware?

Both have same configuration and updates?

There is latest SP and FP, did you try installing them?

Have a look at the following SAP Note:

1203531 - Web Intelligence users experiencing poor response times, unable to log on or navigate between InfoView pages

Regards,

Bashir Awan

[AndrewK]

Make sure you do not have CMS tracing enabled on this environment as this will defintely slow things down, including logins

[Former Member]

no tracing

S.Soheili

[denis_konovalov]

Are you using Cluster name as system name for the login ?

try adding following entry into your hosts file :

99.88.88.88 your_cluster_name

do not use @ for the cluster name and IP field can be anything that looks like a valid IP. can be exactly what I wrote...

[Former Member]

We are having the same issue except that BOXI 3.1 R2 is installed on a Windows 2003 server. We are using the .Net Infoview because we need Windows AD authentication in NTLM mode. When the user first opens Infoview it takes forever to load the logon page and then another minute or so to authenticate and load the home page. When the user clicks on the Document List link, it takes another minute or so to load the list of folders. I've found that many reports that run efficiently on BOXI R2 server often time out when run on BOXI 3.1 R2 and the server is running at maximum CPU capacity. The server is a dual core processor with 4 Gig of RAM which is more than the server running BOXI R2. The latency we are seeing is with one user in a test environment. I hate to think what it would be like if we put this in production. I've checked the SAP Notes for information on this issue and found nothing that solves our problem. Any ideas?

[Former Member]

There were several performance issues resolved with hotfixes. Are you testing with the latest patches installed? If not, try those in your test environment.

[Former Member]

Hi,

it is the BO-Tomcat 5.5.

And yes still slow on logging from test - URL to the production cms - no change.

As I mentioned an ordinary user only in everyone / Infoview logon access right is as slow as an administrator.

There is no difference between LDAP or enterprise, slow either way.

Thanks very much

S.Soheili

[Steve_Fredell]

I would advise that you create a support ticket and have the engineer investigate if your issue is related to our internal KB 1390131 - Logon attempts to CMS become unresponsive which is associated to Bug Fix ADAPT01177087 that was resolved with XI 3.1 SP2.

From the KB

In reviewing CMS trace logs associated with these outages, messsages similar to the following appear:

LDAP: LdapQueryForEntries: QUERY result: 85 took 25000 ms

Note that 25000 ms (a made up figure for this example) is equal to 25 seconds. Queries of this type generally take much less than 1 second. When queries such this one start to take exorbitant amounts of time to process, subsequent requests queue up as the CMS worker threads stay in a waiting status. The result is a 'hanging' CMS that does not respond to new logon requests, regardless of the authentication type.

[joshua_kuhn]

What application server are you using? If you launch the TEST server's CMC url and specify the Production CMS name, do you still see the same slowdown? If not, it's probably something with your production application server. If so, the app server is probably ok and you'll need to look into the CMS and other avenues... In troubleshooting this, i would stick with enterprise logins as ldap is going to involve more components (queries back and forth to the ldap server). Here's your workflow for an enterprise login:

app server --> CMS

CMS queries the CMS db & checks rights, validates user login, etc --> sends result back to app server