399 views

Description

When two threads on the same application server attempt to use CatalogItemCategory.populateCache() at the same time, two semaphores hang. As a result, multiple occurrences of this issue exhaust the available semaphores leading to resource issues that require the node to be restarted.

Steps to Reproduce

To reproduce the issue, a race condition must be met, which is extremely difficult to trigger on demand. Specifically, two user threads need to access the same Java code at nearly the same instant.

Customers are at risk for the issue if they have enabled the contextual search plugin. Customers on affected versions (see Seen In listed below) with the plugin enabled should read the workaround below.

Workaround

There is no current way to work around this issue fully without upgrading to the fixed patch (see Fixed In section below). However, you can reduce the impact by limiting how this issue can be instigated by following these steps to remove the catalog items from the search results:

  1. Navigate to Contextual Search > Search Contexts.
  2. Select Incident Deflection.
  3. Set the value in the Searcher field to Knowledge and pinned knowledge.
    Service catalog results that are throwing the error are removed, but knowledge results are still provided.

Related Problem: PRB642841

Seen In

Dublin Patch 8
Eureka Patch 10
Eureka Patch 10 Hot Fix 5
Fuji Patch 2 Hot Fix 1
Fuji Patch 3
Fuji Patch 4 Hot Fix 4
Fuji Patch 5
Fuji Patch 5 Hot Fix 2
Fuji Patch 6
Fuji Patch 7
Fuji Patch 7 Hot Fix 3
Fuji Patch 7 Hot Fix 5
Fuji Patch 7 Hot Fix 6
Fuji Patch 8

Fixed In

Fuji Patch 7 Hot Fix 11
Fuji Patch 7 Hot Fix 7
Fuji Patch 8 Hot Fix 3
Fuji Patch 9
Geneva
Helsinki
Istanbul

Associated Community Threads

There is no data to report.

Article Information

Last Updated:2016-06-03 15:46:08
Published:2015-08-19