Where communities thrive

  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
  • Jan 06 2016 17:57
  • Jan 06 2016 17:56
Thank you @jrlawhorne .....schedules is an array, can we create the index like above? I tried but it did not work
JR Lawhorne
Hi , I am facing Couchbase Panic Error in a N1QL query ... and some weird behavior that one user getting results of the query and other user randomly getting no results or Panic error ... Any thoughts ?
Hello SDK team, does Spark connector support StructuredStreaming based on DCP?

@daschl about the question you asked in reactor channel - although i am not an expert, but i believe when an outer publisher is unsubscribed, the inner will be unsubscribed too. in the past i had seen such similar behavior with cancellation. Basically reactor would cancel the inner publishers and then outer (just the sequence in which they were created/declared).

Also, if you don't want outer subscriber fail on error, then you should handle the errors inside flatMap (try..catch) or attach one of these operators onErrorResume/onErrorContinue. Won't it help?

But may be the reactor team can give a better answer. Direct IM to Simon Basle?? :-)

@ingenthr @daschl can you share some info on the question i asked about spark connector, structured-streaming, dcp?
Matt Ingenthron
not sure, @daschl or @programmatix may know
Matt Ingenthron
got an answer from @daschl - there is support but it’s marked as experimental (as it was when written) and may need to be updated
@nnanda2016 assuming you have an Enterprise subscription, you may want to open a case to ask for an escalated update in the roadmap
Thanks @ingenthr . I will check with our Cloud DBA to send an email or open a ticket to gather more information
is there any GitHub code repo for us to look into the experimental code?
Mike Damm
Good morning! Is there a more in-depth guide on performance tuning for Couchbase? I googled a bit but could only find the same recommendations for THP and swapiness
I'm running on 4 boxes each with quad core CPU, 48 GB RAM, and two enterprise drives in RAID 0 ... but seeing ~120 ops/sec (doing exclusively upserts)
Frank Lemanschik
There is a lot
You should maybe consider some one with linux knowleg ask a admin to assist you
did you set the ulimit correct already?
ulimit -n 40960        # nofile: max number of open files
ulimit -c unlimited    # core: max core file size
ulimit -l unlimited    # memlock: maximum locked-in-memory address space
what file systems are u using for what partitions
all that matters
Overall there can spend some hours with tuning
network package size also matters
it is endless
Mike Damm
If the filesystem matters to Couchbase, that should be documented somewhere. I'd also be deeply concerned if the max core file size impacts performance. But lets just assume the machines are already tuned as database class machines... is there any Couchbase specific tuning documentation?
Frank Lemanschik
i think this is covered with a subscribtion
and any database and any other software cares for disks
it needs to write and do a lot of other stuff
normal you use independent disks for different parts to maximize the IO
every engineer is aware of disk and ram io and io overall
no need to document that
IO is one of the most big bottlenecks followed by network bandwidth
Tommy Jakobsen
Where is the appropriate place to report issues with the new .NET SDK?
Brant Burnett
@tommyja if you have an account, you can make an issue directly at You could also do a forum post at
Tommy Jakobsen
@brantburnett Thanks!
Matt Ingenthron
note that accounts are Free! at :)
Jugal Shah

Hello guys! I have a newbie question, I wanted to use CouchbaseMock for unit testing instead of connecting to an actual Couchbase cluster.

I tried with Java SDK 2.7.4 and CouchbaseMock 1.5.54 and facing this issue: couchbase/CouchbaseMock#11.

Just wondering if this in fact the recommended usage?

Michael Nitschinger
@jugalps have you tried testcontainers-java instead? that might be a better fit
Jugal Shah
Ah yes I did look at it but unfortunately it requires docker as the prerequisite to be installed on all machines. Unfortunately we are not there yet, but certainly looks like a good option. Thanks @daschl.
Tommy Jakobsen
Is it possible to get "KV Error: {Name=\"KEY_ENOENT\", Description=\"Not Found\", Attributes=\"item-only\"}" when reading a document that is beeing updated by another client? We have seen this several times in a situation where one client is polling a document every 3 sec while waiting for another to update it. It has happened on different clusters (3 nodes, 1 replica)
Jeffry Morris
@tommyja I wouldn't think so under normal circumstances.
Rohat Şahin
We are using eventing functions and a couple time after (for example one day) function is not firing, when we see the statistics looks like this eventing dcp remaining queue not running . why this occur
function executing time max 10 ms
Senthil Sivanath
We are getting BootStrapNodeHasMovedError in XDCR replication.
Getting this only one node of cluster in one DC, any pointers on this could be very helpful

hi, guys if I update my couch base SDK version from v2.1.0

to v2.1.2



I get this error

panic: unambiguous timeout | {"InnerError":{"InnerError":{"InnerError":{},"Message":"unambiguous timeout"}},"OperationID":"WaitUntilReady","Opaque":"","TimeObserved":1000163045,"RetryReasons":null,"RetryAttempts":0,"LastDispatchedTo":"","LastDispatchedFrom":"","LastConnectionID":""}


Charles Dixon
Hi @RezaOptic I guess that you’re using cluster.WaitUntilReady? Unfortunately I’ve just spotted a bug that we introduced in the 2.1.2 release which prevents this from working. Annoyingly and ironically this was introduced when fixing the behaviour for making cluster ops when not using WaitUntilReady which meant that testing focus and integration tests were refocused on that. The fix at least is possibly simple, unless you are using WaitUntilReady for a reason (e.g. other than working around the cluster behaviour in 2.1.1) then you can just stop using it as the ops will now wait until the cluster object is ready anyway. I’m really sorry about that.
Antoine Huret
Hi all! I did a contribution for prometheus couchbase-exporter I'll be very happy to be review soon, do you have an idea about delays ? thanks !
Hi @daschl a question on Java SDK-3. Do you have timeout enforced anywhere in the reactive chain? I mean for SDK 2.x, there was this applyTimeout method; do you have something similar in SDK 3.x?
or it is completely to the SDK users to enforce?