[ https://issues.apache.org/jira/browse/PHOENIX-1954?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jan Fernando updated PHOENIX-1954: ---------------------------------- Attachment: PHOENIX-1954-wip6.patch [~jamestaylor] Following from our discussions attached is a patch that addresses issues (1) and (2). 1) We now support allocating slots in bulk directly from the client cache if the cache has sufficient slots available to service the request. I wasn't able to get rid of all the special casing as we ran into issues trying to have completely share the code for determining if the cache is exhausted across both the NEXT VALUE FOR and NEXT <n> VALUES FOR. The issues were related to cycles and overflows and underflows that created subtle differences in behavior in the 2 flows that I couldn't genericize without making the code very brittle and introducing risks of hard to detect regressions. Therefore the existing check below remains so that NEXT VALUE FOR flow remains the same which will significantly reduce risk for strange regressions like the ones I found. {code} value.currentValue == value.nextValue {code} 2) Added logic to ensure that if NEXT <n> VALUES FOR performs an illegal operation like requesting slots on a sequence with a cycle or exceeding the max and min we throw a SQLException during expression evaluation consistently in all cases. This involved adding client side validation, in lieu of forcing an RPC call each time. I added a few static methods on SequenceUtil to consolidate validation logic between server and client. The attached patch isn't rebased so it's easier for you to see changes. Let me know if you'd like a rebased version with all the changes and I'll do that once you are ready. > Reserve chunks of numbers for a sequence > ---------------------------------------- > > Key: PHOENIX-1954 > URL: https://issues.apache.org/jira/browse/PHOENIX-1954 > Project: Phoenix > Issue Type: New Feature > Reporter: Lars Hofhansl > Assignee: Jan Fernando > Attachments: PHOENIX-1954-rebased.patch, PHOENIX-1954-wip.patch, > PHOENIX-1954-wip2.patch.txt, PHOENIX-1954-wip3.patch, > PHOENIX-1954-wip4.patch, PHOENIX-1954-wip5-rebased.patch, > PHOENIX-1954-wip6.patch > > > In order to be able to generate many ids in bulk (for example in map reduce > jobs) we need a way to generate or reserve large sets of ids. We also need to > mix ids reserved with incrementally generated ids from other clients. > For this we need to atomically increment the sequence and return the value it > had when the increment happened. > If we're OK to throw the current cached set of values away we can do > {{NEXT VALUE FOR <seq>(,<N>)}}, that needs to increment value and return the > value it incremented from (i.e. it has to throw the current cache away, and > return the next value it found at the server). > Or we can invent a new syntax {{RESERVE VALUES FOR <seq>, <N>}} that does the > same, but does not invalidate the cache. > Note that in either case we won't retrieve the reserved set of values via > {{NEXT VALUE FOR}} because we'd need to be idempotent in our case, all we > need to guarantee is that after a call to {{RESERVE VALUES FOR <seq>, <N>}}, > which returns a value <M> is that the range [M, M+N) won't be used by any > other user of the sequence. My might need reserve 1bn ids this way ahead of a > map reduce run. > Any better ideas? -- This message was sent by Atlassian JIRA (v6.3.4#6332)