Clarify error message for stale ring failures.#244
Clarify error message for stale ring failures.#244slfritchie merged 1 commit intobasho:masterfrom evanmcc:pevm_stale_ring
Conversation
doesn't have a entry for the current node name.
| Caps = | ||
| case orddict:find(node(), State#state.supported) of | ||
| {ok, Val} -> Val; | ||
| error -> error("Node name mismatch, move or remove stale ring file") |
There was a problem hiding this comment.
is it safe to just move the stale ring file aside? In the case where I forgot to rename a node and a cluster has been resized wont moving the file cause the node to startup possibly with the old ring size meaning it will have problems when it receives a gossiped ring from another member?
There was a problem hiding this comment.
We often direct people to use older ring files when theirs are corrupted.
If this is unsafe under a resize we may have some more thinking to do. In
any case, this is usually something that people see when they're just
starting up, rather than on mature clusters.
There was a problem hiding this comment.
ah yes, older ring files, that addresses my concern.
|
@evanmcc Based on the comments that I see here and on the referenced tickets, there's enough +1 to +1. +1 from me also. |
Clarify error message for stale ring failures.
3 participants
Add a meaningful warning to the failure when the ring file
doesn't have a entry for the current node name.
Fixes issue #243