Hello,
do you have a (rough) idea of when couchbase 7 final is going to be released? This is important for me because I have an app release pending, but don’t want to do it with beta software. Thank you!
1 Like
We are targeting mid to late summer for GA release of Couchbase Server 7.0. You can contact us at pm@couchbase.com if you need more details. We would like to learn more about how you are leveraging 7.0 features and any feedback you have on the features.
And the 6.6.2 with adaptive indexes bugfixes?
Hello,
are there any updates to the schedule? Couchbase 7.0 is already awesome, but the indexing service gives a huge headache right now.
@zoltan.zvara , it would be great if you can provide more details about the problems faced with index service. We can check if some of those are already addressed in 7.0 or on the roadmap for future releases.
cc @jeelan.poola , @binh.le
Some of the details here [In Couchbase 7.0.0 Beta, indexes may stuck in building 100% state - #19 by zoltan.zvara]
Another issue we found is that after node restart, some indexes stuck in Error state and must be deleted then recreated to be fixed.
With 100s of indexes currently our single 4-node Couchbase 7.0.0 beta CE cluster, it is very hard to move along without index issues when a new application starts up or when a node goes down and comes up.
Thanks @zoltan.zvara for the details. The official 7.0 is still a month or so away from release(I don’t know the exact date). You shouldn’t face the above issues with official release. Meanwhile, you can use the Beta Refresh build (7.0.0-4554). The first issue is definitely addressed in that one. Most likely, you’ll not hit into the second issue as well.
We are using build 4602 and we are still experiencing problems with indexes, unfortunately:
- applications may create duplicate indexes (indexes with the same name may be created concurrently),
- indexes may be stuck in a 100% building state,
- indexes may report Error upon node restart.
P.S.: In case you have a newer build, we would be happy to try it out.
Thanks @zoltan.zvara We’ll need to look at logs to see what is causing the errors mentioned in your post. It will be great if you can help us by uploading logs. Couchbase Server UI has option to collect logs for entire cluster. Please use that tool and pass resultant logs to us. More information on how to trigger log collection can be found at Manage Logging | Couchbase Docs
1 Like
@yogendra.acharya I triggered a collection just now with Customer Name zoltan.zvara. We are running on Kubernetes and I’m not sure ATM if we persist the logs with Pod restart, but let’s see if it would be useful to you. In addition, I will be triggering another log collection immediately after we see the problems.
@zoltan.zvara Thanks for helping with logs. During the analysis we did find few duplicate indexes. In our internal testing we have found some corner cases where this can happen - when one of the indexer node is down and index is being created with duplicate name. But I can not confirm if issue observed on your system is same, as we do not have sufficient logs. Due to log rollover the logs at time of index creation are no longer available. Hence I would like to request you, if you observe the issue again please initiate the log collection and give us a ping. Thank you.
I will do that immediately, thank you!
We did a 4-node restart, where 2 nodes came up very late.
Again I submitted logs to your servers. As I can see, somehow duplicate indexes were created and these duplicates now report “Error” in the UI. In addition, one stuck in “100 % building” state (which is also a duplicate).
@zoltan.zvara , Thank you for uploading the logs, I will be checking them shortly.
@zoltan.zvara, Thanks for the latest logs, I checked the logs and found the cause of duplicate indexes. This is a known issue and can happen in case of failed over nodes. Please refer to Loading... for more details. The MB-22002 fixed the similar behaviour for creation on buckets, scopes and collections but its not yet fixed for indexes. We do have internal ticket open to track the issue.
@yogendra.acharya which build would contain the fix for MB-22002? Will it be fixed until the release of 7.0.0?
@zoltan.zvara The fix will be available in the GA release of 7.0. Stay tuned for the release announcement in Summer
@zoltan.zvara , I would like to clarify that MB-22002 fixed similar issue for bucket, scope and collection creation (duplicate bucket/scope/collection creation in case of failed over node) but it is not fixed for index creation. We do have a internal ticket to track the issue indexes.
Sorry for the confusion caused by previous reply.
The issue happens only in case of failed over nodes and indexers not available is considered to be rare scenario. If you can help us understand your use case, it will help us in prioritization.
@yogendra.acharya I hope there is nothing special in our use case. We simply & sometimes have nodes that restart and when a node comes up again, some indexes seem to be duplicated. The application attempts to auto-create indexes periodically and during its initialization whenever it detects that any index is missing.