[
https://issues.apache.org/jira/browse/ZOOKEEPER-3466?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16899262#comment-16899262
]
Michael Han commented on ZOOKEEPER-3466:
----------------------------------------
I think so far we have the intuition of what went wrong, but we don't know how.
If we can have the complete log files of all servers with DEBUG enabled, then
it might us something. For now, I don't have more clue on my side. I hope if
someone else in community can jump in and take a look at this, as this seems
pretty serious that could prevent adoption of 3.5.5.
> ZK cluster converges, but does not properly handle client connections (new in
> 3.5.5)
> ------------------------------------------------------------------------------------
>
> Key: ZOOKEEPER-3466
> URL: https://issues.apache.org/jira/browse/ZOOKEEPER-3466
> Project: ZooKeeper
> Issue Type: Bug
> Affects Versions: 3.5.5
> Environment: Linux
> Reporter: Jan-Philip Gehrcke
> Priority: Major
>
> Hey, we explore switching from ZooKeeper 3.4.14 to ZooKeeper 3.5.5 in
> [https://github.com/dcos/dcos].
> DC/OS coordinates ZooKeeper via Exhibitor. We are not changing anything
> w.r.t. Exhibitor for now, and are hoping that we can use ZooKeeper 3.5.5 as a
> drop-in replacement for 3.4.14. This seems to work fine when Exhibitor uses a
> so-called static ensemble where the individual ZooKeeper instances are known
> a priori.
> When Exhibitor however discovers individual ZooKeeper instances ("dynamic"
> back-end) then I think we observe a regression where ZooKeeper 3.5.5 can get
> into the following bad state (often, but not always):
> # three ZooKeeper instances find each other, leader election takes place
> (*expected*)
> # leader election succeeds: two followers, one leader (*expected*)
> # all three ZK instances respond IAMOK to RUOK (*expected*)
> # all three ZK instances respond to SRVR (one says "Mode: leader", the other
> two say "Mode: follower") (*expected*)
> # all three ZK instances respond to MNTR and show plausible output
> (*expected*)
> # *{color:#ff0000}Unexpected:{color}* any ZooKeeper client trying to connect
> to any of the three nodes observes a "connection timeout", whereas notably
> this is *not* a TCP connect() timeout. The TCP connect() succeeds, but then
> ZK does not seem to send the expected byte sequence to the TCP connection,
> and the ZK client waits for it via recv() until it hits a timeout condition.
> Examples for two different clients:
> ## In Kazoo we specifically hit _Connection time-out: socket time-out during
> read_
> generated here:
> [https://github.com/python-zk/kazoo/blob/88b657a0977161f3815657878ba48f82a97a3846/kazoo/protocol/connection.py#L249]
> ## In zkCli we see _Client session timed out, have not heard from server in
> 15003ms for sessionid 0x0, closing socket connection and attempting reconnect
> (org.apache.zookeeper.ClientCnxn:main-SendThread(localhost:2181))_
> # This state is stable, it will last forever (well, at least for multiple
> hours and we didn't test longer than that).
> # In our system the ZooKeeper clients are crash-looping. They retry. What I
> have observed is that while they retry the ZK ensemble accumulates
> outstanding requests, here shown from MNTR output (emphasis mine):
> zk_packets_received 2008
> zk_packets_sent 127
> zk_num_alive_connections 18
> zk_outstanding_requests *1880*
> # The leader emits log lines confirming session timeout, example:
> _[myid:3] INFO [SessionTracker:ZooKeeperServer@398] - Expiring session
> 0x2000642b18f0020, timeout of 10000ms exceeded [myid:3] INFO
> [SessionTracker:QuorumZooKeeperServer@157] - Submitting global closeSession
> request for session 0x2000642b18f0020_
> # In this state, restarting any one of the two ZK followers results in the
> same state (clients don't get data from ZK upon connect).
> # In this state, restarting the ZK leader, and therefore triggering a leader
> re-election, almost immediately results in all clients being able to connect
> to all ZK instances successfully.
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
