Log file Viewer
#mongodb logs for Sunday the 16th of September, 2012
[07:53:59] <abstrax> wereHamster: i was wondering whether people choose timestamps or simple date to store ?
[11:25:02] <SvenDowideit> I have added some print() statements to it, and when I call the stored js from the mongo shell, the code does what i expect, and prints the expected states into the mongodb.log file
[11:25:35] <SvenDowideit> but when its called from a query called by the perl app, nothing is printed to the log file, and the results are not as i expect
[11:26:13] <SvenDowideit> I _hope_ that there's something simple i can set to get the print() to go somewhere
[11:27:12] <SvenDowideit> i tried instead to insert the messages into a log collection, but that also only works from the shell for me
[12:34:39] <SvenDowideit> yup, Sun Sep 16 22:33:06 [conn28] query web_6e40bc35d8845d0bdbf7e2b803825b2e.current query: { orderby: { _topic: 1 }, query: { nolock: 1, _history: { $exists: 0 }, $where: "print("didja");" } } ntoreturn:0 keyUpdates:0 locks(micros) r:238 nreturned:0 reslen:20 0ms
[12:38:53] <SvenDowideit> so - if I remove nolock:1 from the query, suddenly print() _does_ print :(
[13:00:38] <SvenDowideit> I basically faked a join by having a db.getSisterDB('other').collection.find() in the $where
[13:01:16] <SvenDowideit> and i've come back to do some work on it a year or 2? later and (just blindly) installed 2.2
[14:21:12] <fotoflo> ok working!! now one more question: i have a collection with an array of tags in the schema, how do i use aggregate to get the list of unique tags?
[15:29:26] <timeturner> what is the most appropriate way to model a one to many relationship in mongodb?
[15:30:01] <timeturner> The only ways that I know if that all of the "many" documents have a field that references the "one" document via object id or something
[15:30:19] <timeturner> or the "one" document has a array of objectids that reference the "many objects"
[15:30:58] <timeturner> I don't know after how many objectIds will the max size of the document be reached
[15:38:37] <ron> also, the document size limitation isn't really that big of a problem, and if you really fear of hitting the limits, there are various ways to deal with it.
[16:08:09] <timeturner> ron: so should I put the many objects' objectids in the one document as an array and if it exceeds that then adjust the system to become more decentralized to where each of the many docs reference the one doc?
[16:20:18] <ron> timeturner: I'd love to go into it with you, but I gotta run. maybe we could talk about it in a few hours.
[16:21:46] <Vile> timeturner: if there going to be huge number of documents and you are going to use small subset of them tied to the main document => use first type of connection
[16:22:28] <Vile> if you would need to have access to all of them (and don't expect many) at any time => store them as array
[16:39:23] <timeturner> At one point I was thinking about a hybrid approach in which I store the latest 10 docs in an array in the one doc
[16:48:46] <timeturner> but the problem is that either I have to set a limit on the number of characters per document or I have to guess a hard limit so that those 10 docs don't exceed the main doc's size lmit
[16:49:37] <mrpro> when are you guys going to start adding new features instead of fixing same bugs/regressions over and over
[16:49:52] <timeturner> I was also thinking about calculating the difference between how much a particular document will take in the one doc via BSONSize and then check how much space is left in the one doc
[16:52:55] <Vile> this way you will not have to care of what to do with sub-doc that goes out of the limit
[16:55:08] <Vile> depends on your usage scenario, though. But usually you would not want to have the whole sub-doc, only the part of it that makes sense and allows to easily get it from the main storage collection
[19:58:26] <noordung> Right, I'm having some weird trouble. I open a connection, I try to save a document, and nothing happens... Never had this before...
[21:10:02] <driscoll> quick q about interpreting mongostat output. not sure what the "locked db" is measuring exactly.
[21:10:55] <driscoll> i'm currently importing some very large JSON files... and locked db is reporting "collection:900%" and higher. this seems odd but my inserts are still around 4k/s
[21:21:31] <noordung> Anyhow, my Mongoose problem seems to be in calling mongoose.model vs connection.model... :/
[21:26:29] <noordung> Right, so note this: mongoose.model(modelName, modelSchema); only stores it in the mongoose implementation, you need to call connection.model(modelName) to get the model constructor...
[22:36:38] <driscoll> for the sake of the log: here's a bug report regarding the "locked db" question i posted earlier: https://jira.mongodb.org/browse/SERVER-6507
[22:57:41] <voldial> Anyone familiar with #ask in semantic-mediawiki? Basically you can embed typed var's in mediawiki pages, and also embed queries (like return all pages that have a GPS coordinate, or sum some var across a subset of pages) that generate a view of the results (table, number, graph, map whatever). It's cool, but it's this big PHP app... and the functions are somewhat limited... I want to roll my own replacement (not in PHP, not tied to media
[22:57:42] <voldial> wiki) and expose a lower level of access to the data... my first thought was to just let users enter SQL directly, but that cant be made safe... so I am looking at mongo's query language... is there some general solution that lets users submit mongo queries in a safe (read-only) way?
[23:07:16] <voldial> as far as I can tell, mogodb queries are read-only... but I suspect a user could still dos the system
[23:09:01] <voldial> found http://docs.mongodb.org/manual/faq/developers/#how-does-mongodb-address-sql-or-query-injection