On Tue, Nov 22, 2011 at 4:43 AM, Szabo, Viktor (Enterprise Infrastructure) <[email protected]> wrote: > Hi CouchDB users, > > I don't know if this is a known issue or not, but I've been experiencing > strange behaviour in version 1.1.1 > after initiating compaction. Even though there's already a doc with the _id > value of "dummy" in my database, > if I want to have it re-inserted after a compact operation is started, 9 > times out of 10, I receive an "ok":true > notification. Occasionally I'm seeing the expected conflict error. > > $ curl -X POST -H "Content-type: application/json" > http://hostname:7002/test/_compact > {"ok":true} > $ curl -X POST -H "Content-type: application/json" -H "Referer: > http://hostname:7002/"; http://hostname:7002/test -d '{ "_id": "dummy" }' > {"ok":true,"id":"dummy","rev":"1-967a00dff5e02add41819138abb3284d"} > $ curl -X POST -H "Content-type: application/json" > http://hostname:7002/test/_compact > {"ok":true} > $ curl -X POST -H "Content-type: application/json" -H "Referer: > http://hostname:7002/"; http://hostname:7002/test -d '{ "_id": "dummy" }' > {"ok":true,"id":"dummy","rev":"1-967a00dff5e02add41819138abb3284d"} > How is this possible? > > Given that I sometimes get an "ok" confirmation, but sometimes get a > "conflict error" for the very same request sent to the > database with the very same content, I don't really feel secure about the > integrity of the service. > I'm thinking that the issue is related to the fact that the database already > had accepted a doc with the same id and content, > but still, undeterministic behaviour is not something I like. > > Do you have a clue on what might be going on? > > (I wasn't able to reproduce under 1.1.0. Used a single-node instance with no > replication set up and no clients connecting to the service) > > Thanks, > Viktor > > -------------------------------------------------------------------------- > NOTICE: Morgan Stanley is not acting as a municipal advisor and the opinions > or views contained herein are not intended to be, and do not constitute, > advice within the meaning of Section 975 of the Dodd-Frank Wall Street Reform > and Consumer Protection Act. If you have received this communication in > error, please destroy all electronic and paper copies and notify the sender > immediately. Mistransmission is not intended to waive confidentiality or > privilege. Morgan Stanley reserves the right, to the extent permitted under > applicable law, to monitor electronic communications. This message is subject > to terms available at the following link: > http://www.morganstanley.com/disclaimers. If you cannot access these links, > please notify us by reply message and we will send the contents to you. By > messaging with Morgan Stanley you consent to the foregoing. >
Your example here is actually hitting a very specific edge case as demonstrated by Marcello's test. As of many versions ago, revisions are generated using a hashing scheme of the document contents. In your particular case the requests you're issuing contain the same identical data in such a way that CouchDB will generate a revision of the doc. Given this, we then have to look at how this plays into replication. Basically, when we merge the revision trees we get to the case where it's "oh, we already have this version, cool" because we do already have this version. Whether or not that behavior is best, or if we should force a conflict if we don't add a leaf during a write is another question. In other words, the system is working fine, but this particular behavior can be a bit unexpected.
