[ZOOKEEPER-1855] calls to zoo_set_server() fail to flush outstanding request queue. - ASF JIRA

Add vote

Voters

Watch issue

Watchers

Create sub-task

Link

Clone

Update Comment Author

Replace String in Comment

Update Comment Visibility

Delete Comments

XML

Word

Printable

JSON

Details

Type: Bug
Status: Open
Priority: Minor
Resolution: Unresolved
Affects Version/s: None
Fix Version/s: None
Component/s: c client
Labels:
None

Description

If one calls zoo_set_servers to update with a new server list that does not contain the currently connected server, the client will disconnect. Fair enough, but any outstanding requests on the set_requests queue aren't completed, so the next completed request from the new server can fail with an out-of-order XID error.

The disconnect occurs in update_addrs(), when a reconfig is necessary, though it's not quite as easy as just calling cleanup_bufs there, because you could then race the call to dequeue_completion in zookeeper_process and pull NULL entries for a recently completed request

I don't have a patch for this right now, but I do have a simple repro I can post when time permits.