Every cloud has a silver lining, and our Support team felt that more than ever this past week. After navigating their way through some technical challenges, the Support team have discovered a pretty nifty way of both fixing and improving our intranet searching.
Our software uses Elasticsearch for the indexing and searching services. This works by making a record of all the content stored in a customer’s intranet, allowing users to search for keywords to retrieve the information they need. So it’s like Google Search for your intranet.
Unfortunately there were some issues with Elasticsearch, where a particular setting (called “heap size”) was causing a few of our customer’s intranets to slow down. This heap size setting had hit its limit, and it turned out that by default, it was set to an extremely low value. And simply increasing this setting wasn’t enough to solve the problem.
Support team investigates
So the issue was more complex than first anticipated. A quick fix wasn’t possible, and instead the Support team had to turn detective.
Being advocates of sustainable solutions, the Support team worked their magic and formulated a fix that not only resolved the issue, but turned it on its head. By changing the configuration of how Elasticsearch and our software work together, including recreating data nodes with double the spec, our intranet searching is now better and faster than ever. Oh and that heap size is now 6x that of the default.
As we look to the future, precautionary measures, such as monitoring and automatic alerts, will be put into place to ensure we catch any further issues quicker.
Thank you to our customers
Of course, we’d like to thank our customers for their patience whilst we sourced a robust and long-term solution. Searching is a key component for a successful intranet, so we’re really excited that we had the chance to improve on a feature that’s so user-critical.