Use a separate snapshot file per lookup tier. - #5358
Conversation
Prevents conflicts if two processes on the same machine use the same lookup snapshot directory but are in different tiers.
| public synchronized List<LookupBean> pullExistingSnapshot() | ||
| public synchronized List<LookupBean> pullExistingSnapshot(final String tier) | ||
| { | ||
| final File persistFile = getPersistFile(tier); |
There was a problem hiding this comment.
for backward compatibility...
if tier.lookupSnapshot.json is not found then we should look for lookupSnapshot.json (move that to tier.lookupSnapshot.json if exists).
let us leave some comment telling why that code exists, and, can be removed some releases after 0.12.0 .
There was a problem hiding this comment.
I was thinking backwards compatible code is not really necessary, since this feature will be able to pull lookups from the coordinator if the file is not found.
Hopefully, this is fine during a rolling upgrade, where the user is paying attention and is sure the coordinator is online and everything is healthy. Meaning snapshots aren't really needed.
There was a problem hiding this comment.
yes, if coord is up then snapshot doesn't matter. so its not very critical to be backward compatible in this case.
|
thanks, this is great. |
Prevents conflicts if two processes on the same machine use the same lookup snapshot directory but are in different tiers.
Prevents conflicts if two processes on the same machine use the same lookup snapshot directory but are in different tiers.
Prevents conflicts if two processes on the same machine use the same lookup snapshot directory but are in different tiers.
3 participants
Prevents conflicts if two processes on the same machine use the
same lookup snapshot directory but are in different tiers.
Fixes issue raised in #5344 (comment).