Raghavendra Thallam

Results 77 comments of Raghavendra Thallam

First attempt to the fix caused https://github.com/yugabyte/yugabyte-db/issues/14213

Doesn't seem to be a problem anymore. Should have been fixed by some of the test stability fixes.

@adithya-yb , Shamanth's data indicates that the tsan failures could have increased after you commit landed. @shamanthchandra-yb , Can we schedule a loop of 100 executions of tsan for this...

There are other tests that run into the same error, See [report](https://detective.dev.yugabyte.com/search?error_details=Failed%3A+IO+error+%28yb%2Fclient%2Fbatcher.cc%3A110%29 ) `Failed: IO error (yb/client/batcher.cc:110) `

seems like a deadlock of some sort. Pasting a small snippet below. [Full log can be found here ](https://jenkins.dev.yugabyte.com/job/yugabyte-db-phabricator/102575/testReport/junit/(root)/ClientTest/TestServerTooBusyRetry/ ) ``` table_handle.cc:371] Failed: IO error (yb/client/batcher.cc:110): Errors occurred while reaching...

cc @ttyusupov , @lingamsandeep , if this is a known deadlock in GetTableLocations space. cc @amitanandaiyer , @bmatican

Using this issue to track the disabling of Packed Row + Colocation. https://github.com/yugabyte/yugabyte-db/issues/21244 will track the real fix.

@mdbridge , You are correct about the PG semantics, In yugabyte, we were support column level locking, but there is a increased push to head towards the row level locking...

@pilshchikov , Do you think this is a recent regression? Are we able to triangulate/narrow down the builds?