First I must say I find your presentations and the other CB team members very helpful and meaningful thanks for that.
We are not in production yet but plan to be in end on this year so we have the time to test cb4.0 as N1QL looks amazing.
The issues we are facing are strange and our system works on CB 3.0.3 without any issues running on 4.0 beta 1 we experiencing different error mentioned here
2015-07-20T16:54:31.369Z+03:00 [Info] PRAM[:9999] Request "/adminport/shutdownTopicRequest" 2015-07-20T16:54:31.37Z+03:00 [Info] PROJ[:9999] ##1 doShutdownTopic() "MAINT_STREAM_TOPIC_a2:34:83:ba:c1:b9:4a:3d" 2015-07-20T16:54:31.37Z+03:00 [Error] PROJ[:9999] ##1 GetFeed(): projector.topicMissing 2015-07-20T16:54:31.37Z+03:00 [Info] PROJ[:9999] ##1 doShutdownTopic() returns ... 2015-07-20T16:54:31.39Z+03:00 [Info] PRAM[:9999] Request "/adminport/shutdownTopicRequest" 2015-07-20T16:54:31.391Z+03:00…
We are testing CB 4.0 and are getting Errors in the Logs and Indexing hangs we need to Stop / Start the CB server and Flush the database with fresh restore
The set up is 1 node for Data 20GB Mem 2 Buckets 1st bucket with 3 design docs 3 views 2nd bucket 6 design docs 34 views
1 node for Index 20 GB Mem 1 node for Query 20 GB Mem no primary index defined and no N1QL is currently used
Any ideas what we should do .
Port server query on node 'babysi…
Couchbase rocks but I need your help ? when i setup the cluster i used private ips but some how internally local host is used how can i overide it or fix it ?
2015-07-20T16:48:28.306Z+03:00 [Error] IndexerSettingsManager: metakv notifier failed (Post
http://127.0.0.1:8091/_metakv: dial tcp 127.0.0.1:8091: connection refused)..Restarting 2015-07-20T16:48:28.307Z+03:00 [Error] IndexerSettingsManager: metakv notifier failed (Post http://127.0.0.1:8091/_metakv: dial tcp 127.0.0.1…
the last post also has full logs attached for active for download for 15 days since 20-7-2015
also reproduced here
We believe all to be related some how we don’t know exactly the actions that lead to the errors but it must be somehow related to Compacting and Indexing
Trying different node set-ups as the Couchbase Connect videos suggest e.g. separate
each node HP 4 cores CPU 24 RAM 3 SDD-122 GB 1 for CenOS 1 for cb-data and 1 for cb-index
and we have 4
and try different configurations e.g.
1 data , 1 index and 1 query
1 with all three ( data , index and query)
1 with index
1 with data
and other permutations
All our tests show that after 1 night running the indexing hangs and the system is non responsive
configurations we do
curl -X POST -v -d ‘updateInterval=10000&updateMinChanges=100’
Swappiness is turned off
Disable Transparent Huge Pages (THP)
as per http://blog.couchbase.com/often-overlooked-linux-os-tweaks
auto compact is as is default
and 2 buckets we have we set both to High priority Disk I/O Optimization
and replication is disabled
Cihan any help is welcome at this point . If more logs or info is need please do not hesitate to ask
Also I have no problem to remove the other Topics if it helps clarity to have only one topic.
A work around we find is if we trigger a FailOver then cancel it then rebalance the system finds its way and starts to work again properly .