No subject

Thu Feb 16 03:44:05 GMT 2012

explicitly specified

Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

Thanks for looking into it Matthias,<br><br>&quot;rabbitmq-dev&quot;: yes, =
it is load balancer.<br>Disconnects: I was too lazy to write code for it an=
d just restarted client manually.<br>I did make a short try to reproduce fa=
ilure without any load but it worked just fine. I&#39;ll try more today.<br=
<br>Vadim.<br><br><div class=3D"gmail_quote">On Sat, Apr 28, 2012 at 12:13 =
AM, Matthias Radestock <span dir=3D"ltr">&lt;<a href=3D"mailto:matthias at rab=" target=3D"_blank">matthias at</a>&gt;</span> wrote:<br=
><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1=
px #ccc solid;padding-left:1ex">
(putting the list back on cc)<div class=3D"im"><br>
On 27/04/12 23:28, Vadim Chekan wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
I&#39;ve spent some time today playing with different client settings. Seem=
ttl does not affect failures at all.<br>
That&#39;s good to know. The more factors we can eliminate as possible caus=
es the better.<div class=3D"im"><br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
I managed to reproduce crash many times today. Basic idea is: an<br>
application, 40 threads create a pub/sub exchange and publish a message<br>
every second. Under this load I bring the master node (all queues are<br>
usually created on the same node) and often it causes another node to<br>
fail. Here is my load simulator (in c#):<br>
<a href=3D"" target=3D"_blank">http://ww=<u></u>clip/0B5W</a><br>
That code connects to &quot;rabbitmq-dev&quot;. Is that a load balancer sit=
ting in front of the three nodes?<br>
How do you deal with the disconnects resulting from the shutting down of no=
des? There doesn&#39;t seem to be any code to handle that.<br>
Mind you, I suspect that the failure should still be reproducible without a=
ny subscriptions and publishes. Would be good to try that and just watch th=
e broker logs for errors.<div class=3D"im"><br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
As long as maillist does not allow zip attaches, I&#39;m mailing to you guy=
Thanks for posting these. There is an error in the logs that we haven&#39;t=
 seen before:<br>
 =A0 =A0 =A0 =A0 =A0 {{badmatch,[]},<br>
 =A0 =A0 =A0 =A0 =A0 =A0[{rabbit_mirror_queue_misc,&#39;-<u></u>remove_from=
 =A0 =A0 =A0 =A0 =A0 =A0 {mnesia_tm,apply_fun,3},<br>
 =A0 =A0 =A0 =A0 =A0 =A0 {mnesia_tm,execute_<u></u>transaction,5},<br>
 =A0 =A0 =A0 =A0 =A0 =A0 {rabbit_misc,&#39;-execute_mnesia_<u></u>transacti=
 =A0 =A0 =A0 =A0 =A0 =A0 {worker_pool_worker,handle_<u></u>call,3},<br>
 =A0 =A0 =A0 =A0 =A0 =A0 {gen_server2,handle_msg,2},<br>
 =A0 =A0 =A0 =A0 =A0 =A0 {proc_lib,wake_up,3}]}<br>
Looking at the code, this appears to indicate that there are no master/mirr=
or processes left for the queue. Which is...unexpected. That should give us=
 something to go on.<br>
</blockquote></div><br><br clear=3D"all"><br>-- <br>From RFC 2631: In ASN.1=
, EXPLICIT tagging is implicit unless IMPLICIT is explicitly specified<br>


More information about the rabbitmq-discuss mailing list