You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Go play some minecraft until you almost lose your job. Just to pass time, you know 😅
Run ./console queuedtracking:monitor
Result: trackingQueueV1 has substantionally more requests, than other queues. The option "Process during tracking request" seemingly does not affect the queue number. At least I haven't noticed any correlation.
Generally, Matomo will put all requests for the same User, Visitor and IP address into the same queue.
My question is: Is there a specific reason to do it like so? I mean, putting all requests for the same User, Visitor and IP address into the same queue. Wouldn't it be better to try to evenly distribute the requests among queues? Are there specific relations, that might break if we don't put it into the same queue?
If not, I'd go with a simple rand, which statistically would ensure uniform distribution of requests:
protected function getQueueIdForVisitor($visitorId)
{
return rand(0, $this->numQueuesAvailable - 1);
}
simply keeping $visitorId for backwards compatibility for now.
The text was updated successfully, but these errors were encountered:
Hi @kevinkopf . Thank you for taking the time to create this issue. I have confirmed that is the expected behaviour. Matomo tries to split the requests as evenly as possible between queues, but it also tries to keep requests from the same visitor/user in the same queue. We do this to ensure that we can replay events in the correct order.
I'm marking this issue as closed, but please feel free to let me know if you think there's cause to have it re-opened or if you have any further questions.
Steps to reproduce:
./console queuedtracking:monitor
Result:
trackingQueueV1
has substantionally more requests, than other queues. The option "Process during tracking request" seemingly does not affect the queue number. At least I haven't noticed any correlation.Similar problem was reported here as well:
and maybe a couple of other places I cant recall now. I managed to track the uneven distribution of requests to this function.
Now I know, that @tsteur said here
My question is: Is there a specific reason to do it like so? I mean, putting all requests for the same User, Visitor and IP address into the same queue. Wouldn't it be better to try to evenly distribute the requests among queues? Are there specific relations, that might break if we don't put it into the same queue?
If not, I'd go with a simple
rand
, which statistically would ensure uniform distribution of requests:simply keeping
$visitorId
for backwards compatibility for now.The text was updated successfully, but these errors were encountered: