Riak start script

[ruichuan]

This branch is to stabilize the riak start script via adding retries and reporting errors when the cluster is started unsuccessfully.

[iakkus]

Should we also add some reachability related checks?
The 'curl' statement currently fails in a silent way and I think someone had run into this issue. Maybe we can also retry on that?

[ruichuan]

Should we also add some reachability related checks?
The 'curl' statement currently fails in a silent way and I think someone had run into this issue. Maybe we can also retry on that?

The retry for reachability check has been added.

[iakkus]

Looks good, thanks!

[manuelstein]

Quick question: when the join succeeds, it would directly test if the node is listed in the cluster status as "joining". Could that go wrong? E.g. when the ring is busy with other stuff, could it take a second "cluster status" to find the node is "joining"? On one deployment I saw the third out of 3 nodes was marked "joining".

[ruichuan]

I thought the "riak-admin cluster join" succeeds only if it has successfully marked a node as "joining". But, I guess it does no harm if we add a sleep before that "joining" check. Pushed a simple change. Thanks.

…

On Wed, Jul 15, 2020 at 7:42 PM Manuel Stein ***@***.***> wrote: ***@***.**** commented on this pull request. Quick question: when the join succeeds, it would directly test if the node is listed in the cluster status as "joining". Could that go wrong? E.g. when the ring is busy with other stuff, could it take a second "cluster status" to find the node is "joining"? On one deployment I saw the third out of 3 nodes was marked "joining". — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#73 (review)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAQSEDBAVCW4F254SJE4PRDR3XTAJANCNFSM4OZQ6VHQ> .