This is a follow-up blog post to the Tech Talk Live 123 titled “Discovering the ‘2’ in Search Services 2.0”. If you missed the webinar, you can see the recording here. If you are interested in the slides, you can find them here.
During the webinar, many questions have been raised and as we ran out of time to answer all the questions, we've included those along with the ones that were addressed. Below you can find the list, together with the answers from the speakers.
Q: A quick question about the new Solr trackers, is there any batching for metadata retrieval or is it only one by one per thread in the ForkJoin pool?
A: Metadata Retrieval is batched using the size of “alfresco.*.tracker.maxParallelism” value in solrcore.properties file.
Q: Passing passwords via environment variables in Docker images is considered unsafe AFAIK. Are Alfresco images at the least providing an option to use secrets / reference mounted files to handle secrets?
A: Currently there is no option to handle Docker secrets in Alfresco Images, using password files could be a safer alternative for those deployments.
Q: How does disabling solr index compare to "noindex" configuration on the ACS side?
A: Disabling the trackers is a feature in Search Services 2.0, whereas the "noindex" subsystem is an option in ACS Repository. The main benefit of disabling the trackers is that it provides a way to cancel large in-flight maintenance jobs (e.g. perhaps trying to fix a set of transactions is taking longer than expected and is having an impact on general Search Services performance).
The "noindex" option provides a way to ensure queries are directed through the DB rather than Search Services (the "index"). It does not actually stop Search Services from indexing documents.
Q:Will this work with Alfresco Community 6.2 and (sidepoint) any hopes of getting an installer tool for 6.2?
A: Yes, Search Services 2.0 has been tested with Alfresco Community 6.2.
We are currently looking at Ansible rather than the old style installer Alfresco used to supply.
Q: Since I have never worked with Insight Engine / JDBC support, I wonder how well do the date functions support timezones? I.e. SOLR typically only indexes in UTC, but any BI would have to use user-/customer-specific timezones for evaluation...Strike the "typically" in the last - SOLR only indexes in UTC
A: The SQL endpoint does not currently support timezones. See SEARCH-2145 for more details.
Q: That is relevant for query-level timezones. With SQL, I wonder as well if there is some support for timezones in specific sections of the conditions / evaluation + aggregation functions like QUARTER which were mentioned in a slide as new features…
A: Currently there is no support for timezones in SQL functions.
Q: Can the community help with your velocity around supporting solr 8?
A: Supporting SOLR 8 is not an easy task, we had some previous work on that. Anyway, if some external contributions are coming, we’ll support the merging process in order to add this feature to the product.
Last but not least let me thank Francesco Corti, Tom Page, Angel Borroy, Keerat Laila and Martin Stanford for leading the conversation with the external community.