Documentation's navigation
On this page
Linchpin Intranet Suite Performance Impact compared to Vanilla Confluence
Atlassian obliges (certified) app manufacturers to carry out a mandatory performance comparison test of typical Confluence functions for each release. The measured values must be in an acceptable ratio in order to receive a Marketplace certification.
Using the Linchpin Suite results in a maximum performance loss of ~10% compared to a Confluence without Linchpin.
It should, of course, be noted that the system resources must be expanded appropriately for operation with Linchpin.
Adequate system resources
In the Linchpin Intranet Suite documentation, we provide a guide to appropriate system resources for cases. Please base your own requirements on the recommended values provided.
Java resources
As a general rule, Confluence together with the Linchpin Intranet Suite requires at least 6 GB heap space and 384 MB → Java code cache. Depending on the system load, this value can also be doubled or increased, whereby the rule of thumb is: RAM/2 = Java Heap.
Linchpin-specific configuration
Linchpin Theme
It can happen that the Linchpin theme configuration causes problems, which can manifest themselves in a slowdown of various kinds.
Sometimes a corrupt character set can lead to such behavior.
- The Linchpin UTF-8 health check should be carried out here first
The Linchpin Health Check only refers to the currently saved Linchpin theme configuration, whereby only the text content from the "Custom CSS" and "Custom Javascript" text fields is relevant. If characters are used that cannot be saved correctly with the current encoding, there is a risk that the Linchpin configuration will grow over time to a file size that has a negative impact on Linchpin performance.
You can check this using a Linchpin theme export, for example; the export file is a pure JSON text file that is usually only a few kilobytes in size. If there are unknown characters in your custom code, the file size can easily grow to several megabytes. The phenomenon is described in this article.
Encoding the configuration in Unicode with UTF-8
In this case, you should ensure that all contents of the Linchpin theme configuration are encoded in Unicode with UTF-8. If a conversion is not possible, please export the theme configuration and attach it to your reply. We will then be happy to carry out the conversion for you.
- Linchpin Configuration > Navigation Menus > Menu Structure > "..." Button > "Export saved menu"
Deactivate health check (not recommended)
If you do not run into this problem or have not saved any custom code, you can also ignore or deactivate the Linchpin health check by deactivating the module "linchpin.healthchecks.fileencodinghealthcheck.name" in the "Linchpin Theme" app.
Confluence and JVM should use UTF-8 encoding
The currently active settings can be viewed directly in the Confluence administration area:
- General Configuration > Formatting and International Settings
- System Information > sun.jnu.encoding and file.encoding
Please consult the following Atlassian articles (if necessary, make sure you have the appropriate documentation version for your Confluence version):
Ideally, not only Confluence or the JVM, but also any other software of the infrastructure (database, reverse proxy) and the file system of the operating system should use Unicode/UTF-8, see e.g.:
We cannot give you any advice on system configuration, but the Atlassian documentation mentioned and Atlassian Support can help you if you need assistance. Please also note that a conversion may have an impact on other system components and should be considered and tested by you accordingly, e.g. PDF export and application links.
Java CodeCache & batch.js
It could be that the Java code cache is too small and therefore the batch.js files have to be regenerated too often. To prevent this, increase the Java code cache (see Extended system resources for Linchpin above).
Known bottlenecks
Certain standard settings and macros supplied with Linchpin, in combination with other factors, prove to be particularly CPU-intensive. You can find a list here to check directly for performance-related anomalies.
Linchpin Macros: "Cover Stories" / "Corporate News Feed"
Searching, loading and preparing the news to be displayed generates load on the server, especially on the search index.
If you notice performance problems here and it is editorially possible, you can reduce the number of news items to be searched for:
- Try to get by with as few news macros as possible per page.
- Try to configure as few tiles as possible in the cover stories or as many tiles as possible with the same source configuration. A separate search must be started for each source configuration.
- If the "Slides" function is activated in the cover stories, reduce the number of slides or switch the option off completely.
Linchpin Macros: Customized User List
This macro is outdated and quite slow, especially with large result sets. If possible, please use the new "People Search & List" macro.
General Troubleshooting Methods
On our Infothek page with HowTo's for customers you will find instructions on how to best summarize performance problems in a report.
The following questions should be clarified - in particular, whether the problem occurs with Linchpin alone:
- How long has the problem been occurring in this way and what changes were made previously?
(e.g. in the plugins, the global Confluence configuration, infrastructure such as DBMS, Java version, server, reverse proxy, CDN, etc.). - Which apps were deactivated - was the system completely switched to safe mode?
Safe Mode
This allows interactions to be tested, e.g. by only activating the Linchpin Intranet Suite and deactivating all 3rd party apps:
Within the Confluence administration ("Manage apps" area), you can start Confluence Safe Mode via the "Switch to safe mode" link. In safe mode, but before testing, the plugin caches should also be emptied once:
- Administration > Cache management > "Clear all" button
- CONFLUENCE_ADDRESS/admin/cache/showStatistics.action
- Now please activate only the Linchpin apps, whereby the following sequence should be followed first:
- (1) Linchpin Suite - API → (2) Linchpin Communicator → (3) Linchpin Intranet Suite
You can then (re)activate all other Linchpin apps in no particular order.
If the problem no longer occurs, the problem is due to an installed app from another manufacturer. This can be determined specifically by reactivating the other 3rd party apps one by one and retesting the behavior with each reactivation.
Are customizations (stylesheets, scripts) used or removed for testing purposes?
Customizations
CSS stylesheets and self-written HTML and JavaScript code (custom code) can be integrated in various places in Confluence.
Check the following locations for custom code:
- CONFLUENCE_ADDRESS/admin/linchpin/theme/configure.action → via the Confluence administration > Linchpin > Theme > Configuration > Advanced you will find one input window each for CSS and JavaScript code, which applies the code to the entire Linchpin theme
- CONFLUENCE_ADDRESS/admin/viewstylesheet.action → via Confluence administration > Design > Style sheet you will find an input window for CSS, which applies the code to all areas
- CONFLUENCE_ADDRESS/admin/viewcustomhtml.action → via the Confluence administration > Design > Custom HTML you will find an input window for HTML, which applies the code to all areas. JavaScript scripts can also be stored here
- In the respective areas, you can also insert CSS code that only applies to the specific area via the Configure area > Design control panel and the Style sheet sub-tab
Do the problems occur with a constant intensity over the course of the day or when do peaks occur and how does this correlate with user behavior or recurring jobs? Is there a monitoring system in place to monitor the utilization of important resources?
What happens if the page content (e.g. macros) is gradually reduced? See also Coverstories macro
Are the server's general system resources adequately dimensioned?
CPU cores, RAM, heap, code cache, DB I/O
App-Caches
A screenshot of the current app caches could be useful for further troubleshooting. 
For this, please send us a screenshot of your current app caches.
- To do this, please navigate to Confluence Administration > Cache management or via
CONFLUENCE_ADDRESS/admin/cache/showStatistics.action - Click on "Show advanced view" / "Show advanced view"
- Now take a screenshot of all lines that contain "Seibert", "Linchpin" or "Microblog". The screenshot should cover the columns "Cache name" to "Hit/Miss/Evicted".
Advanced methods
Support-Zip with DEBUG-Log
Please open the menu item "Logging and Profiling" in the administration area.
Please enter the following key in the text field under "Add new entry" if we have not given you another one:
net.seibertmedia (for the logs of all Seibert Media apps),
set the accompanying drop-down field to "ALL" and add the entry (briefly check whether the key is actually contained in the list).
Now reproduce the problem (preferably with a timestamp (hh:mm) of when you did this), i.e.
- If it concerns a specific macro: Create a new Confluence test page and insert only the relevant macro with the corresponding configuration there
- Reload the website by bypassing the browser cache ("hard reload")
Then download a support ZIP file via the administration area > "Troubleshooting and support tools" > "Create support ZIP file" tab and attach the file to your response.
Note: You can upload files up to 100 MByte in tickets - for larger files we can provide you with a special upload link.
After the support ZIP file has been created, you should remove the logging key again to avoid an unnecessary logging load. (Note: every time Confluence is restarted, the logging levels are automatically reset to the "default setting")
Export HAR Data
Open the so-called "Developer Tools" ("DevTools" for short) of your web browser.
In most cases (Chrome, Edge, Firefox), the F12 key is sufficient to open the DevTools.
Within the DevTools window...
- open the "Network" tab (or "Network analysis")
- make sure that both options are active (see also picture below) :warning: "Preserve log" and "Disable cache"
- click on the :warning: "Clear" icon at the top right to delete previous logs
- now reproduce the problem (load page, perform action, etc.)
- export the HAR file (via the "arrow down" icon)
Note: in Firefox it is the gear icon on the right > "Save all as HAR"
In this screenshot (Google Chrome) all steps are shown at a glance. (for other web browsers you will find the procedure described here)
Save Javascript logs as a text file
After you have exported the HAR file, please switch to the "Console" tab of the DevTools.
To display only all Javascript errors, please click on the corresponding filter on the left.
Now right-click in an empty area of the window with the error messages to select "Save as" from the menu.
The procedure is also described in this screenshot:
