[ZOOKEEPER-1561] Zookeeper client may hang on a server restart

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 3.5.0
• Fix Version/s: 3.5.0
• Component/s: java client
• Release Note: It is fixed in ZOOKEEPER-1560.
• Assignee: Unassigned
• Reporter: Jacky007
• Created: 11/Oct/12 07:43
• Updated: 24/Dec/12 04:18
• Resolved: 24/Dec/12 03:54

Description

In the doIO method of ClientCnxnSocketNIO

 if (p != null) {
                    outgoingQueue.removeFirstOccurrence(p);
                    updateLastSend();
                    if ((p.requestHeader != null) &&
                            (p.requestHeader.getType() != OpCode.ping) &&
                            (p.requestHeader.getType() != OpCode.auth)) {
                        p.requestHeader.setXid(cnxn.getXid());
                    }
                    p.createBB();
                    ByteBuffer pbb = p.bb;
                    sock.write(pbb);
                    if (!pbb.hasRemaining()) {
                        sentCount++;
                        if (p.requestHeader != null
                                && p.requestHeader.getType() != OpCode.ping
                                && p.requestHeader.getType() != OpCode.auth) {
                            pending.add(p);
                        }
                    }

When the sock.write(pbb) method throws an exception, the packet will not be cleanup(not in outgoingQueue nor in pendingQueue). If the client wait for it, it will wait forever...

Issue Links

• is related to: ZOOKEEPER-1560 Zookeeper client hangs on creation of large nodes
• relates to: ZOOKEEPER-1437 Client uses session before SASL authentication complete
• relates to: ZOOKEEPER-107 Allow dynamic changes to server cluster membership