Fixed an issue where the client can sometimes attempt write to a read…

[whatsup2100]

…only node.

[aravindyeduvaka]

We're considering this PR but we would like a little more context as when you usually see this problem.

[whatsup2100]

Hello, Aravind. I found this to be a common issue when using Redis Labs Enterprise in a cluster environment under high load (100k+ users). It most often occurred when when a master<->slave rollover occurred (managed or due to failure) and possibly during discovery of the master by the client. I have not been able to reproduce this issue with the open source version of Redis. I have used this solution for ~6 months in a production environment with a lot of success.

…

On Fri, Jun 21, 2019 at 6:13 PM Aravind Yeduvaka ***@***.***> wrote: We're considering this PR but we would like a little more context as when you usually see this problem. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#102?email_source=notifications&email_token=AJSREX3JN6GCPP7WIKP3DNLP3VOCRA5CNFSM4FEML3OKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODYJZRVI#issuecomment-504600789>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AJSREX3GXB5FVTXHVUJLZCLP3VOCRANCNFSM4FEML3OA> .

[mapalan]

Hi @whatsup2100, are you seeing the same issues as #22 ? It might be due to timeout configurations. Sorry for the incredibly long delay.

[thomas-hayden]

@mapalan Yes, this appears this appears to be the same issue we were having. We've since switched away to no longer using a master/slave setup, so I can't say for certain if the timeout configs would make a difference. Thanks for finally getting back on this.