Hi All,

Pardon my poor English skill.


I'm new to Solr and testing DIH delta-import with solr-4.0.

When testing delta-import, I've found a scenario that some updates are
missing.

This scenario follows the steps below.

--------------------------------------------

No. db-update delta-import

---------------------------------------------

1. db-update1

2. commit1

3. db-update2

4. delta-import1

5. commit2

6. db-update3

7. commit3

8. delta-import2


Delta-import writes the index start time in dataimport.properties file. So,
if the DB is MySQL using TRANSACTION_READ_COMMITTED as a transaction
isolation level, the updates in db-update2 are missing in both
delta-imports.


As far as my understanding, there's two way to stop this. One is
serializing db-update and solr-update(delta-import), which is unwanted
dependency. The other is using TRANSACTION_READ_UNCOMMITTED as isolation
level of DIH, which may cause mismatch between db-data and solr-data if
db-update has been rolled back.


The better way to solve this problem, as far as I know, is using the the
time of the latest added/modified documents, which I cannot find in ZIRA.


What is the best solution?


Thanks in advance for your feedback.


Regards,

Jang

Reply via email to