huaban / erlzk Goto Github PK

tests on erlzk fail where it tries to use create2 - e.g.
=ERROR REPORT==== 25-Dec-2014::12:20:53 ===
Reconnect to "localhost":2181 successful
erlzk_test: erlzk_test_...failed
in function erlzk_test:delete/1 (test/erlzk_test.erl, line 140)
**error:{badmatch,{error,unimplemented}}
..where line 140 contains.
{ok, {"/a", Stat}} = erlzk:create2(Pid, "/a"),
zookeeper 3.4.6

I ran into this problem when deliberately dropping all packets between client and one connected server - I expected to see it reconnect to one of the other servers but instead it just sat there.
If I bounce the connected server instead the fail-over is fine - but if I interrupt the connection it fails.
Looking at the code it seems like the client is not sending heartbeats to the server. Comparing it to the java I can see ping defined in ZooDefs.java but the matching operation does not appear in erlzk_codec.erl.
I'll read the java some more to understand what needs to be done to make this work - can you let me know what the position is with heartbeats from your side. --- thanks k

Hello,

I am using the erlzk library in one of my app, which I am writing in Elixir. I am trying to connect to zk node using a consul address via vpn. It gives me error and I am not able to establish connection with the zknode. The error I am seeing is as follows:

I❤iex|2|▶▶▶ :erlzk.connect([{"pipe-apps-zookeeper-0.node.us-east-1-z.consul", 2181}], 30000)
{:ok, #PID<0.150.0>}
I❤iex|3|▶▶▶
02:21:22.525 [error] Resolving "pipe-apps-zookeeper-0.node.us-east-1-z.consul":2181 meet an error: :einval

02:21:22.525 [error] Connect fail: :no_available_server, will be try again after 1000s

02:21:23.519 [error] Connect fail: :no_available_server, will be try again after 1000s

02:21:24.520 [error] Connect fail: :no_available_server, will be try again after 1000s

##########################################################################
Also how do I send chroot while creating a zk connection. Because when I try to send it as below I am getting error:

I❤iex|3|▶▶▶ :erlzk.connect([{"pipe-apps-z-0.node.us-east-1-z.consul", 2181}], 30000, [{chroot, "/pipe-apps-general-kafka"}])
warning: variable "chroot" does not exist and is being expanded to "chroot()", please use parentheses to remove the ambiguity or change the variable name
iex:3

Bug Bug ..!!** (CompileError) iex:3: undefined function chroot/0
Bug Bug ..!! (stdlib) lists.erl:1354: :lists.mapfoldl/3

Thank your for your work on this project, we've enjoyed using it. We recently had this issue happen in one of our environments (thankfully not prod). The result was that our service hung and was not processing data.

Steps to reproduce

(apologies for the elixir, my erlang vocab is not so good)

1. Set host list to be something like this:

:erlzk_conn.start_link([{'notahost.url', 2181}], ....)

2. running (with iex -S mix) will produce this output:

14:18:28.577 [error] Resolving 'notahost.url':2181 meet an error: :nxdomain
14:18:28.577 [error] Connect fail: :no_available_server, will be try again after 1000s
14:18:29.569 [error] Connect fail: :no_available_server, will be try again after 1000s
14:18:30.570 [error] Connect fail: :no_available_server, will be try again after 1000s
....
# this continues until the program is killed

Cause

erlzk discards hosts if they fail to resolve

When we try to establish a connection to zk, this branch is always hit:

Followed by this error handler:

Ideas for fix

• We could add a check for list being empty after we resolve the servers in

  erlzk/src/erlzk_conn.erl

  Line 165 in a1432aa

  ResolvedServerList = resolve_servers(ServerList),

• track how many errors we are encountering in some time interval and die if that goes to high.
  ie. allow a max of 30 errors in 60s

I'm curious to hear your thoughts.

Could you please push 0.3.1 tag to repo

Would be great to have the enhancements made in OTP 18 available in this app.

Cheers

Hi,

I'm seeing this:

17:16:30.770 [error] FORMAT ERROR: "Connection to ~p:~p timeout while waiting for connect reply: ~p~n" [{17 ,17,0,4},2181]

It seems to come from here.

Hey I had a question: I wanted to known the synatx for :erlzk.multi method if used in elixir. I tried below but I am getting an error:

:erlzk.multi(pid, [{:create, chroot, :persistent}, {:create, "#{chroot}/job_done", :persistent}, {:create, "#{chroot}/job_pending", :persistent}, {:create, "#{chroot}/job_inprogress", :persistent}, {:create, "#{chroot}/job_failed", :persistent}])

I'm getting a lot of error messages in my logs for normal operations while using this library. For example, these three show up a lot:

There are other conditions that are logged as warnings as well, such as

Would you welcome a PR to tone these down a little?

Hello.
I hadn't found other way to communicate with "kevinbombadil", sorry for that way.

At one of issue for "pybind11" you sad,that you write python binding to "libdar".
Can you share that code ?
Can we discuss that in private way ?
Thank you
fred.kalsin at gmail dot com