Where Does Citrix Site Services Web Interface Log Requests

CTX125715

Web Interface All-time Practices to Avoid Major Production Outages

Commodity | Other | {{likeCount}} found this helpful | Created: {{articleFormattedCreatedDate}} | Modified: {{articleFormattedModifiedDate}}

Data

Some version of Web Interface has likely been deployed in 99 percent of all Citrix implementations and the bug might have originated from their daily utilise. This article explains a few methods to avoid bug that might arise from full general deployments. XenApp Services sites (formerly known as PN Agents sites) tin also benefit from this data.

Background

Organizations adopt the rapid deployment method that Citrix Web Interface provides and thus tend to deploy it to aggregate XenApp applications to their user base of operations. As Web Interface usage grows within the organization, assemblage of applications might begin to prove occasional bug and in rare occasions cause major outages.
Rare occasions where Spider web Interface in conjunction with its dependencies cause major outages must be avoided. In very few occasions Citrix has seen Web Interface alone as the root cause of the problem. In many occasions, its dependencies (IIS server, XML broker, Secure Ticket Authority (STA) Servers, Hallmark services) are likely to be the crusade of the bottleneck.
Focusing on the about used and obvious conjunction of dependencies, the XML broker and STA servers are probable to exist on top of the list. In all new and supported XenApp and XenDesktop platforms, the XML broker is as well the STA server and these, for the well-nigh part, are likely to be part of the occasions of major outages than any other dependency.

Compilation of Problem Areas

The following areas do not cause a problem immediately and thus are challenging to pin down a few weeks or months afterwards when an outage occurs.

XML and STA broker over utilize.
Lack of IIS log clean upwardly on the log drive, causing drive space problems.
Rogue, non-standard, XenApp published applications to merely a selected set of users that occasionally employ the application.
Authentication points immune by Spider web Interface that occasionally neglect for a diversity of reasons.

Avoiding Major Outages

2 methods are currently deployed when any XML or STA broker becomes unresponsive and causes an outage.
Web Interface administrators have multiple Web Interface servers which betoken to the same one or more XML brokers and utilize the built-in mechanism to fail over or load balance the XML requests to the aforementioned set of defined XML brokers. Although this proves to be a reliable method for small to medium deployments, increased usage can occasionally cripple the back end XML brokers, causing all working Spider web Interface servers to struggle for a properly working XML broker. Following are two methods that can be deployed to handle unresponsive XML brokers.

Use an industry proven Load Balancer, similar NetScaler, which is highly recommend for its built-in XML monitors and persistence mechanism to maintain user sessions for all-time operation and user feel. The Load Balancing VIP created in NetScaler can so exist used to provide Spider web Interface assurance that the XML broker receiving asking from information technology is properly operation because NetScaler removes any unresponsive XML broker service from the list.
If a Load Balancer is not used for the XML broker servers, and then each Web Interface should point to one or ii unique and distinct XML brokers in the same farm for XenApp awarding aggregation. This method ensures that if one XML broker fails, it just affects one Web Interface server. This just works if information technology is not a subcontract-wide outage where all XML brokers fail regardless of where it is. Post-obit is a more descriptive view of this idea:
Web Interface 1 -> XML broker 1 and 2
Web Interface 2 -> XML banker three and 4
Web Interface 3 -> XML broker 5 and vi
Web Interface 4 -> XML broker vii and 8
This method provides a better adventure of affecting only one Web Interface server rather than all 4 if they would share the same XML brokers.
If Citrix Access Gateway products are used in conjunction with this type of setup, the Admission Gateway needs to contain all eight XML and STA brokers, so that it can redeem the advisable STA ticket.
Every bit an additional precaution, each Spider web Interface tin can be refined to exist more reactive of an XML banker problem by tweaking the following options from the Advance Farm Settings section of the site.
- Enable socket pooling: This option is just available when the site hallmark points are at Web Interface or Access Gateway.
  Uncheck this option to allow socket connections to the XML banker to be initiated at will rather than using an already created socket from the pool. Whatsoever reduced functioning acquired past this alter is minimal, particularly for non SSL connections to the XML broker servers.
- Socket timeout: Lowering this time interval allows faster proactive determination of a problematic XML broker.
- Attempts made to contact the XML Service: Interval number of how many attempts Web Interface takes to contact an XML broker that has met the socket timeout interval.

Additional Resources

Refer to the NetScaler device and Web Interface administrator guide or eDocs for additional details on Load Balancing mechanism and Spider web Interface specific settings.
For more information nearly Advanced Farm Settings on Spider web Interface, refer to CTX122029 - Web Interface Error: The Web site is experiencing technical difficulties
For XenApp and XML broker performance settings, refer to CTX118742 - XenApp Infrastructure Monitoring and Alerting
When there are multiple Web Interface sites consider isolating each site into its ain application puddle. Each split up application pool provides a procedure boundary so that when a Web Interface site is assigned to one application pool, bug in other application pools exercise not affect the awarding. This ensures that if a Web Interface site fails, information technology does not bear upon the Web Interface sites running in other application pools. Refer to CTX126625 - How to Isolate Web Interface or Programme Neighborhood Amanuensis Sites into Their Own Awarding Pool for Best Performance for details.