-
Notifications
You must be signed in to change notification settings - Fork 4
FAQ Common support tasks
As all of the searches are directed to the Solr instance, this needs to be functioning correctly. Encountered problems include: * running out of memory
- fragmented index causing too many open file handles
OutOfMemory errors are very infrequent. Restarting the Solr service is the quickest way of resolving this issue.
service Solr restart
This problem has been observed more frequently when the memory available to Solr (displayed on the dashboard/homepage of each instance) approaches the size of the index on disk or the number of segments exceeds 25.
A number of File exceptions are caused by the number of files open (Too many open files). This can be avoided by optimising the index and setting the file handle limits available to root (or whatever user Solr runs as) to a large number e.g.
root - nofiles 1000000
A periodic task to generate a sitemaps for the eMonocot portal based on the sitemaps standard for submission to search engines. via. Spring Batch: * Go to the 'Sitemap Generation' job in the eMonocot Harvester
- Manually add the parameters (copy a previous execution changing the 'attempt' parameter), e.g.
Sitemaps have been registered in major search engines such as google using the google webmaster tools account. Sitemaps should be reharvested periodically, but can be manually resubmitted if neccessary.
A catch-all for data that has been imported or search results that shouldn't appear; * Running the re-index job for the appropriate data type (see repairing indexes section) has solved a number of problems caused by inconsistent/'bad' archives and other changes
- Running the image processing job (see reprocess images section) will update image files and thumbnails that haven't been downloaded or are out of date
via. Spring Batch: * Go to the 'Re-Index' job in the eMonocot Harvester
- Manually add the parameters (copy a previous execution changing the 'attempt' parameter), e.g.
The highlighted classes should be replaced appropriately * Click launch and check back periodically while the index is updated
If it's necessary to rebuild all the indices (e.g. because the Solr server was rebuilt), these are all the entity types:
Entity | Base class | Query type |
Annotation | org.emonocot.model.Annotation | org.emonocot.model.Annotation |
Comment | org.emonocot.model.Comment | org.emonocot.model.Comment |
IdentificationKey | org.emonocot.model.IdentificationKey | org.emonocot.model.IdentificationKey |
Image | org.emonocot.model.Image | org.emonocot.model.Image |
PhylogeneticTree | org.emonocot.model.PhylogeneticTree | org.emonocot.model.PhylogeneticTree |
Place | org.emonocot.model.Place | org.emonocot.model.Place |
Taxon | org.emonocot.model.Taxon | org.emonocot.model.Taxon |
TypeAndSpecimen | org.emonocot.model.TypeAndSpecimen | org.emonocot.model.TypeAndSpecimen |
Group | org.emonocot.model.auth.Group | org.emonocot.model.auth.Group |
User | org.emonocot.model.auth.User | org.emonocot.model.auth.User |
Organisation | org.emonocot.model.registry.Organisation | org.emonocot.model.registry.Organisation |
Resource | org.emonocot.model.registry.Resource | org.emonocot.model.registry.Resource |
via. Spring Batch: * Go to the 'Image Processing' job in the eMonocot Harvester
- Manually add the parameters (copy a previous execution changing the 'attempt' parameter), e.g.
A number of tasks can be achieved by any member of the portal 'Administrators' group: * Adding users to groups
- Adding/Updating/Deleting Users, Organisations (Sources) and their Resources
N.B.Full administrator documentation is available on the Families and Orders scratchpad.
If the scheduled harvesting is not running, check whether there are any Jobs that appear to be running:
mysql> SELECT * FROM resource WHERE status NOT IN('COMPLETED', 'FAILED', 'ABANDONED', 'STOPPED') AND status IS NOT NULL
If there are results # Check Spring Batch Admin (e.g. UAT) to verify that there are no running jobs
- Stop and abandon (buttons in prominent view) any jobs that shouldn't be running, e.g. They started days ago, they didn't fail cleanly
- Update the database for jobs that aren't actually running
or if you are resonably certain that no-one will be activating a job; change the WHERE clause to update all problem jobs.