Not retry when receive error: Partition not available (code 11)

[bysoul]

Hi,
I'm using aerospike go client to write kv.
When the kv is being written, aerospike cluster starts to roll its nodes, the client then get err: Partition not available.
At

aerospike-client-go/command.go

Line 2781 in 7022c84

return errChain.setInDoubt(ifc.isRead(), cmd.commandWasSent)

, the client will always return this err because it is not a network err. And the client has no chance to make ptn.sequence++ to change partition id at

aerospike-client-go/command.go

Line 2604 in 7022c84

if !ifc.prepareRetry(ifc, isClientTimeout || (err != nil && err.Matches(types.SERVER_NOT_AVAILABLE))) {

Please take a look for this behavior, thanks a lot.
Best Regards

[khaf]

Hi, thanks for your report. Just to make sure: Are you encountering this issue in batch reads?

[bysoul]

Hi, This issue was encountered when I used PutBins. Batch Read was not tested. Best Regards

[mirzakaracic]

I’m having trouble reproducing the issue and could use a few more details about your environment and test scenario:

1. Rolling restarts – Where is the Aerospike cluster running, and how are you performing the rolling restarts?
2. Versions – What versions of the Aerospike server and client are you using?