So would a non zero TTL be required for the eviction to work? Thank you.
jbekas wrote: > > > yesnid wrote: >> >> I have tried this with a TTL of 0 essentially infinite for the message >> and for non-zero values, but neither yielded any results for me. Also My >> queue that I am trying to evict from has no consumer on it, does that >> make a difference, I am beginning to suspect from other posts that it is? >> >> Could you post a configuration that has worked and I could try this and >> see if I get better results with eviction? >> > > You do not need a consumer to expire the messages. However, the messages > will just sit there (expired) unless some activity occurs on your queue to > tell ActiveMQ to purge the old messages. For testing purposes, I wrote a > slow consumer (1 message per second) which allows me to rapidly send > messages with a short TTL and be able to watch them expire. > > The sample configuration you posted is probably working, but the messages > are sticking around in your queue. You can try to set an attribute on > your oldestMessageEvictionStrategy parameter, it's called > evictExpiredMessagesHighWatermark. This gives ActiveMQ the hint that it > should purge expired messages after a set threshold of expired messages > has been met, in this case, 1000. Without this set, your messages will > likely stick around until ActiveMQ decides that it needs to free up space > on your queue for incoming messages. > > <messageEvictionStrategy> > <oldestMessageEvictionStrategy > evictExpiredMessagesHighWatermark="1000" /> > </messageEvictionStrategy> > > > In my test, my slow consumer processes 1 message a second. So, if I send > 1500 messages with a TTL of 1000ms, the consumer should process 1-2 > messages, ActiveMQ will expire the rest, and somewhere between 1000-1500 > messages will be immediately removed from the queue and placed on the > ActiveMQ.DLQ (the default DeadLetterQueue). Note, some expired messages > will probably remain on the queue until the high watermark is hit again or > ActiveMQ inspects them and decides to remove them. > > I hope that helps. > > -- View this message in context: http://www.nabble.com/Queue-PolicyEntry-Eviction-Not-working-tp23026416p23031084.html Sent from the ActiveMQ - User mailing list archive at Nabble.com.
