robertsami

Results 12 issues of robertsami

Jira Link: [DB-3162](https://yugabyte.atlassian.net/browse/DB-3162) We can detect some deadlocks without relying on distributed deadlock detection by just looking at local wait-for relationships. We can do this for a new waiting transaction...

kind/bug
area/docdb
priority/medium

Jira Link: [DB-3161](https://yugabyte.atlassian.net/browse/DB-3161)

kind/bug
area/docdb
priority/medium

Jira Link: [DB-3160](https://yugabyte.atlassian.net/browse/DB-3160) Some examples: 1. latency of deadlock detection (min, max, avg) 2. avg cycle length in deadlock detection 3. deadlocks detected per second 4. wait time in wait...

kind/bug
area/docdb
priority/medium

Jira Link: [DB-3159](https://yugabyte.atlassian.net/browse/DB-3159) Interactions of tablet splitting and pessimistic locking /should/ work, but confirm this with comprehensive tests

kind/bug
area/docdb
priority/medium

Jira Link: [DB-3157](https://yugabyte.atlassian.net/browse/DB-3157) Any waiting transaction blocked in the WaitQueue will be unblocked as soon as the WaitQueue detects that it's blockers have been resolved. We currently rely on a...

kind/bug
area/docdb
priority/medium

Jira Link: [DB-3158](https://yugabyte.atlassian.net/browse/DB-3158) Whenever a tracked blocker is resolved in the WaitQueue, our process for determining which waiters can be resumed and subsequently resuming them could be improved in a...

kind/bug
area/docdb
priority/medium

Jira Link: [DB-3156](https://yugabyte.atlassian.net/browse/DB-3156) See: https://github.com/yugabyte/yugabyte-db/blob/master/src/yb/docdb/wait_queue.cc#L352 https://github.com/yugabyte/yugabyte-db/blob/master/src/yb/docdb/conflict_resolution.cc#L656 Instead of re-running conflict resolution in these cases, just rollback the transaction and fail the operation

kind/bug
area/docdb
priority/medium

Jira Link: [DB-3155](https://yugabyte.atlassian.net/browse/DB-3155) We should delay deadlock detection for some configurable amount of time, say 1s, similar to Postgres. This will allow us to avoid the overhead of deadlock detection...

kind/bug
area/docdb
priority/medium

Jira Link: [DB-3646](https://yugabyte.atlassian.net/browse/DB-3646) We currently make a best effort to avoid rolling back transactions unnecessarily in the face of deadlock but we do not guarantee that we won't rollback multiple...

kind/bug
area/docdb
priority/medium

Jira Link: [DB-9952](https://yugabyte.atlassian.net/browse/DB-9952) ### Description With the introduction of #14165, we will start resolving deadlocks based on txn start time. We currently report the txn start time based on client/transaction.cc...

kind/bug
area/docdb
priority/medium