[Alfresco 5.0] How to speed up indexing on a new file

[mtgdavidchow]

Dear All,

One of our client is using Alfresco 5.0, when we add a new file or modify a metadata in a file, it will take 1-2 hours to be able to be searched in using CMIS session query (org.apache.chemistry.opencmis). E.g. the metadata is patientNo, assuming originally it is "PN0001", after we changed to "PN0002", we can't find the record by searching "PN0002", but searching "PN001" is ok (But in alfresco UI, the metadata already updated). After 1-2 hours, can be searched by "PN0002" with correct result

When the data set is small, it just needs <5 mins. Now the data is around 10TB. and resulting the above issue.

Is that anyway to check why it is slow and any parameter or configuration I can try ?

Thank you!

[angelborroy]

Search Services is indexing the repository to update changes in the properties. The batch process performing this task is MetadataTracker. It looks like you need to provide more resources to Solr in order to get the update done faster.

[mtgdavidchow]

Is this the solr setting you mean ?  if yes, I think we have already set relatively large in the resource. but the indexing is still very slow

[mtgdavidchow]

---

@mtgdavidchow wrote:

Is this the solr setting you mean ?  if yes, I think we have already set relatively large in the resource. but the indexing is still very slow

---

Anyone can have insight on that?

[mtgdavidchow]

@angelborroy Can you advise?

[jd_Inpute]

There are lots of variables impacting indexeing performance. You say you have 10Tb of data, how is that made up? How many nodes do you have in your index? Have you got content indexing enabled? Hows the file system performing?

Have you searched for any Solr 4 performance information?  There's some good (if old) Alfresco presentations from Alfresco events on the web.

Troubleshooting something like this isn't going to be quick and should rule out all the obvious issues (I see you have lots of RAM) with infrastrucuture then look at the actual indexing performance. You may want to consider re-creating the indexes on another server, see how quickly they regenerate.

Also, really consider upgrading to Search 2.x - its much more better performing!