topic Re: SOLR Backup scheduled jobs in Alfresco Forum

SOLR Backup scheduled jobs

angelborroy — Sat, 02 Dec 2017 08:57:00 GMT

SOLR 4 subsystem includes by default following scheduled jobs:search.indexBackupJobDetail > index.backup.cronExpression=0 0 3 * * ?search.alfrescoCoreBackupJobDetail > solr.backup.alfresco.cronExpression=0 0 2 * * ?search.archiveCoreBackupJobDetail > solr.backup.archive.cronExpression=0 0 4

Re: SOLR Backup scheduled jobs

krutik_jayswal — Sat, 02 Dec 2017 10:17:42 GMT

Not sure exact things about this stuff, but if you analyse below files you will definitely get an idea.

community-edition-old/solr-backup-context.xml at master · Alfresco/community-edition-old · GitHub

community-edition-old/SolrBackupJob.java at master · Alfresco/community-edition-old · GitHub

community-edition-old/SolrBackupClient.java at master · Alfresco/community-edition-old · GitHub

Re: SOLR Backup scheduled jobs

angelborroy — Sat, 02 Dec 2017 11:43:35 GMT

Thanks for your comments.

I analysed those source files and also the ones related with Lucene. It looks like search.indexBackupJobDetail is not used anymore, as it was only relevant for Lucene.

So in current versions we have a backup for both SOLR cores: alfresco and archive. The first one is performed at 2:00 AM and the second one at 4:00 AM. This last scheduling happens at the same time than contentStoreCleanerJobDetail, which is moving files from contentstore to contentstore.deleted. So we have Alfresco backing archive SOLR indexes and creating new entries in archive SOLR indexes at the same time.

It could be a designed behaviour but I found it weird.

Re: SOLR Backup scheduled jobs

afaust — Mon, 04 Dec 2017 09:53:43 GMT

You are misunderstanding what contentStoreCleanerJobDetail does. It does not cause any new entries in archive SOLR indexes to be created. When content is moved from contentstore into contentstore.deleted, it has already been removed from the archive (or it has never been there in case the content was not archived / transferred into the trash can). So there is no competition between the contentStoreCleanerJobDetail and the archive index backup, if that is what you find weird.