You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In #7238 it looked like index corruption (checksum errors) but in fact it was simply that the user selected bloom_pulsing postings format, which we don't support yet still allow.
We recently removed documentation showing these postings format as a choice, but it's still really dangerous we allow this option at all since it creates unusable indices in ES when we migrate shards and try to check integrity. Before 1.3, ES didn't check Lucene checksums, so these postings formats worked fine, but with 1.3 any index using pulsing will fail.
The pulsing optimization has already been folded into the default postings format for quite a while now.
Should we go further and disable (for now) any custom formats that don't have backwards compatibility support from lucene? These can change across releases in such a way that looks like corruption.
We are currently trying to figure out a way in Lucene to safely provide options to the user AND backwards compatibility, but this is not going to happen overnight.
Should we go further and disable (for now) any custom formats that don't have backwards compatibility support from lucene? These can change across releases in such a way that looks like corruption.
Lucene's experimental codecs (from the codecs module) do not provide
backwards compatibility and are free to change from release to
release. When they do change, they typically cannot in general read
older indices and the resulting exceptions look like index corruption.
So, we are removing built-in support for them to prevent applications
from choosing one and then seeing strange exceptions on upgrade.
Closes#7566Closes#7604
clintongormley
changed the title
Remove pulsing/bloom_pulsing postings format
Mapping: Remove pulsing/bloom_pulsing postings format
Sep 8, 2014
In #7238 it looked like index corruption (checksum errors) but in fact it was simply that the user selected bloom_pulsing postings format, which we don't support yet still allow.
We recently removed documentation showing these postings format as a choice, but it's still really dangerous we allow this option at all since it creates unusable indices in ES when we migrate shards and try to check integrity. Before 1.3, ES didn't check Lucene checksums, so these postings formats worked fine, but with 1.3 any index using pulsing will fail.
The pulsing optimization has already been folded into the default postings format for quite a while now.
I think we should remove them; we are already removing pulsing from Lucene (https://issues.apache.org/jira/browse/LUCENE-5915)
The text was updated successfully, but these errors were encountered: