Hi RabbitMQ Team,


If one of the RabbitMQ node under cluster environment facing out of disk
space problem, what will happen?


1) The cluster stop function properly or even worst?
2) Just impacted node stop sync data from others?


Thanks & Regards,
Wong
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20121024/d64905b6/attachment.htm>

Search Discussions

  • Simon MacMullen at Oct 24, 2012 at 9:14 am

    On 24/10/12 07:46, Wong Kam Hoong wrote:
    Hi RabbitMQ Team,

    If one of the RabbitMQ node under cluster environment facing out of disk
    space problem, what will happen?

    1) The cluster stop function properly or even worst?
    2) Just impacted node stop sync data from others?

    Disc space monitoring was added in 2.8.2. If the disc space falls below
    the monitored threshold then *the whole cluster* will stop accepting
    publishes until more space becomes available.


    This doesn't completely stop you from running out of disc space, but it
    makes it very unlikely that RabbitMQ will take the last disc space on
    the machine.


    Running out of disc space will cause RabbitMQ to crash.


    Cheers, Simon


    --
    Simon MacMullen
    RabbitMQ, VMware
  • Matthias Radestock at Oct 24, 2012 at 9:34 am

    On 24/10/12 10:14, Simon MacMullen wrote:
    If the disc space falls below the monitored threshold then *the whole
    cluster* will stop accepting publishes until more space becomes
    available.

    The same is true for memory alarms, i.e. when *any* node in a cluster
    encounters a memory alarm then *all* nodes will block publishers.


    Matthias.
  • Wong Kam Hoong at Oct 25, 2012 at 2:25 am
    Hi Matthias,


    Besides of stop all publishes, how about consumes? Does RabbitMQ stop the
    consumes as well?


    Regards,
    Wong




    On Wed, Oct 24, 2012 at 5:34 PM, Matthias Radestock
    wrote:

    On 24/10/12 10:14, Simon MacMullen wrote:

    If the disc space falls below the monitored threshold then *the whole
    cluster* will stop accepting publishes until more space becomes
    available.
    The same is true for memory alarms, i.e. when *any* node in a cluster
    encounters a memory alarm then *all* nodes will block publishers.

    Matthias.
    ______________________________**_________________
    rabbitmq-discuss mailing list
    rabbitmq-discuss at lists.**rabbitmq.com<rabbitmq-discuss@lists.rabbitmq.com>
    https://lists.rabbitmq.com/**cgi-bin/mailman/listinfo/**rabbitmq-discuss<https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss>
    -------------- next part --------------
    An HTML attachment was scrubbed...
    URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20121025/469700c5/attachment.htm>
  • Wong Kam Hoong at Oct 25, 2012 at 2:22 am
    Hi Simon,


    You meant the cluster not only stop all publish requests and will crash at
    the end? Means consume requests also can't be send to the RabbitMQ?


    Regards,
    Wong


    On Wed, Oct 24, 2012 at 5:14 PM, Simon MacMullen wrote:

    On 24/10/12 07:46, Wong Kam Hoong wrote:

    Hi RabbitMQ Team,

    If one of the RabbitMQ node under cluster environment facing out of disk
    space problem, what will happen?

    1) The cluster stop function properly or even worst?
    2) Just impacted node stop sync data from others?
    Disc space monitoring was added in 2.8.2. If the disc space falls below
    the monitored threshold then *the whole cluster* will stop accepting
    publishes until more space becomes available.

    This doesn't completely stop you from running out of disc space, but it
    makes it very unlikely that RabbitMQ will take the last disc space on the
    machine.

    Running out of disc space will cause RabbitMQ to crash.

    Cheers, Simon

    --
    Simon MacMullen
    RabbitMQ, VMware
    -------------- next part --------------
    An HTML attachment was scrubbed...
    URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20121025/af48adeb/attachment.htm>
  • Matthias Radestock at Oct 25, 2012 at 7:18 am

    On 25/10/12 03:22, Wong Kam Hoong wrote:
    You meant the cluster not only stop all publish requests and will
    crash at the end?

    Well, if you really *completely run out* of disk space or memory then
    yes, whichever node that is happening on will crash. But that's why we
    have alarms in place that block publishing across the entire cluster
    when disk space or memory are getting low - before rabbit is in danger
    of crashing.

    Besides of stop all publishes, how about consumes? Does RabbitMQ stop
    the consumes as well?

    No, assuming you follow the advise from http://www.rabbitmq.com/memory.html
    <quote>
    other design considerations permitting, it is advisable to only use
    individual AMQP connections for either producing or consuming.
    </quote>


    Matthias.

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
grouprabbitmq-discuss @
categoriesrabbitmq
postedOct 24, '12 at 6:46a
activeOct 25, '12 at 7:18a
posts6
users3
websiterabbitmq.com
irc#rabbitmq

People

Translate

site design / logo © 2017 Grokbase