GitHub user prashanth-vasudev opened a pull request:
https://github.com/apache/incubator-trafodion/pull/463
Fix for TRAFODION-1923
Basically, we now understand why SQL was not able to close the outstanding
scanner ,
1. because of cancel and the plan involved esps, it was the esps that get
killed which did not have an option to close the scanner.
2. Because of this stale scanner is found in region when drop table is
issued.
3. Chore thread in region is able to clean the stale scanner, however it
may kick in at different times.
4. A drop table issued in the meantime appears as hung since region
observer intercepts the region close ( due to drop table) and finds a stale
scanner.
5. Fix is to detect this situation in regionobserver and validate if a
transaction state object exists for this scanner. If it exists, then the
intercept is valid, otherwise let the region close complete.
I have pushed the code and did a PR request so you can review the changes.
However, committers wait till all the regressions in my dev environment
complete ( I will send out a message once its done).
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/prashanth-vasudev/incubator-trafodion
trafodion-1923
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/incubator-trafodion/pull/463.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #463
----
----
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
