[zardosht]

Roger,

I work at Tokutek (and wrote the post above). I'm sorry you ran into issues trying out TokuMX. I assure you, we are "ready", as we have users running in production.

Nevertheless, you ran into problems and that is unfortunate. If you have details, can you please share them with the tokumx-user google group? We might be able to help. I suspect the transition to using a transactional system like TokuMX where entire statements are transactional is resulting in some "gotchas", but that is just an educated guess.

-Zardosht

[rogerbinns]

I mean ready in the sense that pointing code that worked flawlessly against MongoDB to TokuMX then just works flawlessly too.

I uninstalled Toku and went back to MongoDB so I can't provide any further testing. (The mongorestore takes days.)

I can tell you want code was running at the time. It reads events sorted by user id and timestamp, and then discovers session boundaries in that. A new session object (in a different collection) is written out with all the events as a subdocument list. (In rarer cases an existing session object is updated.) This was happening in 8 separate processes all in Python/pymongo. There are no statements running that affect more than one document, nor any need for transactions.

[leif]

If you were using upserts I expect you were having problems due to the optimizer retrying all possible plans (including table scan) periodically. This is reflected in https://github.com/Tokutek/mongo/issues/796 and is fixed in 1.4.0. If you'd like to try another evaluation, get in touch with us and we can help you track down whatever problems you see.

Not all mongodb code will optimally use tokumx without any changes. Concurrency is hard and mongodb encourages some patterns that are bad for any concurrent database. For example, count() for an entire collection is not, and could never be, as cheap in a concurrent database like tokumx as it is in mongodb.