You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@storm.apache.org by Andrew Montalenti <an...@parsely.com> on 2014/06/03 04:07:37 UTC

Topology acked/emitted count reset

Attached you'll find two screenshots from the Storm UI, one taken this
morning, and one taken just recently. The Storm topology in question --
"cass" -- was not restarted in between.

You can see the uptime is 13h (storm_ui_healthy.png) and 26h
(storm_ui_num_reset.png), respectively. Yet, notice that in the later
screenshot, the "acked" counter for the "all-time" window has *dropped*
from 27.2 million to 3.9 million. All the other counts have also dropped.

What explains this? Shouldn't alltime emit/ack counts for a topology that's
been running 26h non-stop always be greater than the same topology 13h
earlier?

This is with Storm 0.9.1-incubating.

---
Andrew Montalenti
Co-Founder & CTO
http://parse.ly

Re: Topology acked/emitted count reset

Posted by Harsha <st...@harsha.io>.
Hi Andrew,

              From what I read in the code executor.clj (worker) is
responsible for updating the stats for bolts and spouts . If a worker
is restarted or it might be the case if a topology is rebalanced there
is a chance of loosing the stats.  Topology stats derived from spouts
and bolts there is no stats kept track for topology itself. So if a
worker / supervisor died and restarted on another node stats for that
supervisor/workers are lost.

Thanks,

Harsha





On Mon, Jun 2, 2014, at 07:07 PM, Andrew Montalenti wrote:

Attached you'll find two screenshots from the Storm UI, one taken this
morning, and one taken just recently. The Storm topology in question --
"cass" -- was not restarted in between.

You can see the uptime is 13h (storm_ui_healthy.png) and 26h
(storm_ui_num_reset.png), respectively. Yet, notice that in the later
screenshot, the "acked" counter for the "all-time" window has dropped
from 27.2 million to 3.9 million. All the other counts have also
dropped.

What explains this? Shouldn't alltime emit/ack counts for a topology
that's been running 26h non-stop always be greater than the same
topology 13h earlier?

This is with Storm 0.9.1-incubating.

---
Andrew Montalenti
Co-Founder & CTO
[1]http://parse.ly

  Email had 2 attachments:
  * storm_ui_num_reset.png
  *   566k (image/png)
  * storm_ui_healthy.png
  *   541k (image/png)

References

1. http://parse.ly/