[PDB-3830] Command Queue Grows Steadily Over Time not related to load Created: 2018/02/05 Updated: 2018/08/24 Resolved: 2018/03/08
|Fix Version/s:||PDB 5.1.5, PDB 5.2.1|
|Reporter:||Nick Walker||Assignee:||Rob Browning|
|Remaining Estimate:||Not Specified|
|Time Spent:||Not Specified|
|Original Estimate:||Not Specified|
|Sprint:||Data Platform 2018-03-14|
|Method Found:||Customer Feedback|
|Release Notes:||Bug Fix|
|Release Notes Summary:||PuppetDB now correctly reduces the command queue size when a malformed message is submitted to the queue. Previously if you submitted a message that PuppetDB discards, the command queue would increase when the message came in and would not reduce when the message was discarded.|
|QA Risk Assessment:||Needs Assessment|
Over time we notice that the queue depth grows just a little bit and never drops back down. Over the course of a few days it'll be sitting around 300-400 despite capacity to work it down.
After graphing some other metrics we see that discard count seems suspiciously close to the command queue which grows over time.
I suspect that discards are not correctly removed from the counter for queue depth.
|Comment by Jeff Yarnell [ 2018/02/14 ]|
Nick Walker Have you tested your patch for this?
|Comment by Nick Walker [ 2018/02/15 ]|
|Comment by Jeff Yarnell [ 2018/02/15 ]|
Russell Mull Please see Nick's question above.
|Comment by Jeff Yarnell [ 2018/02/21 ]|
Nick Walker In what PDB version did you see this?
|Comment by Nick Walker [ 2018/02/21 ]|
I think it should go into the next z release of PE 2017.3 so I think that would mean PDB 5.1.5?