Log file Viewer
#mongodb logs for Tuesday the 28th of October, 2014
[00:49:33] <mgeorge> is there a best practice configdb dir for rhel? the default selinux blocks mongodb from accessing /data/configdb/
[01:59:59] <mgeorge> so apparently to fix 2.6.5 on rhel7 to use /data/configdb you need to change the context of /data to "mongod_var_lib_t" using the semanage fcontext -a -t mongod_var_lib_t '/data(/.*)?'
[02:16:57] <mgeorge> or just use "semanage permissive -a mongod_t" which i guess is a little easier till dev's create a context "mongod_configdb_t"
[02:28:43] <Streemo> does anyone know why when I do db.posts.insert({title:'blah'}), the _id field is populated with something like ObjectId("ejlksfhskljr32kj4wjasd")
[02:29:14] <GothAlice> http://docs.mongodb.org/manual/reference/object-id/ — technically, it's a type.
[02:34:05] <Streemo> but i call something like db.insert(...) without _id and then my db gets populated with an object with a string id
[02:37:34] <GothAlice> AFIK inserting without specifying an _id will *always* result in a true ObjectId as the value. Only if you explicitly insert a value for _id (or accidentally set it in an upsert) will it differ from this.
[02:46:18] <Streemo> GothAlice: when i call Framework.collection.insert(...) (from the JS framework) the resulting _id is a string. When i do it straight in mongo it's an ObjectId.
[02:47:11] <GothAlice> A haaa. Your JS driver must only expose them in their encoded string form. Are results in the shell sane?
[02:49:09] <Streemo> accessing documents in mongo shell which were created in the driver results in _id being a string type
[02:49:54] <GothAlice> That's very not right. Can you pastebin one of the insert() lines from your code?
[02:50:54] <Streemo> Posts = new Meteor.Collection('posts') ===> creates a 'posts' colleciton in mongo (db.posts)
[02:51:56] <GothAlice> What version of Meteor? https://github.com/meteor/meteor/issues/61 < are you doing something like this?
[02:53:35] <GothAlice> Streemo: A ha: http://stackoverflow.com/questions/15041128/meteor-0-5-7-how-to-handle-use-meteor-collection-objectid — new Meteor.Collection('posts', {idGeneration: 'MONGO'})
[02:55:08] <GothAlice> Note that what this documentation specifies ("random Mongo.ObjectID values") is also wholly incorrect. ObjectIds are anything but random.
[03:00:06] <Streemo> how many hex chars are there, 16? (1/16) ^ what, 24 characters? thats superrrrrrrr improb. to have say 22 / 24 to be the same
[03:00:31] <GothAlice> Streemo: No legitimate source anywhere on the planet will call an ObjectId random.
[03:01:54] <Streemo> "Create a Mongo-style ObjectID. If you don't specify a hexString, the ObjectID will generated randomly (not using MongoDB's ID construction rules)."
[03:06:59] <GothAlice> Streemo: Meteor Considered Harmful™ — breaks expected usage and data validation standards conformed to by every other driver/wrapper I have ever seen for every language ever.
[03:07:50] <GothAlice> I like to reinvent the wheel; but at least my wheels still conform to the expectation of being round.
[03:11:30] <GothAlice> 0.1 + 0.1 (ten times total) = 0.9(rep) Which mathematically is exactly 1, but not to a computer…
[03:11:55] <lpghatguy> Most RDMS I've used implement a decimal type. Otherwise, using two integer fields (one for the integer part and one for the decimal part) is okay
[03:12:05] <GothAlice> nateritter: I store them as carefully formatted strings with some adapter glue to translate seamlessly to Python's Decimal type while still supporting querying.
[03:13:13] <GothAlice> https://github.com/MongoEngine/mongoengine/blob/master/mongoengine/fields.py#L279-L307
[03:14:14] <nateritter> kinda lame currency is still something we have to deal with on a regular basis
[03:16:16] <GothAlice> lpghatguy: Pro tip: Store the whole part of a BTC value as an integer, the decimal part as a string. ;)
[03:17:37] <lpghatguy> sounds like it'd be a pain to do numeric operations on it without a proper decimal implementation in the end
[03:18:01] <GothAlice> lpghatguy: There are some tricks and shortcuts you can use with a split representation like that. Does come down to regexen a bit, though.
[03:18:38] <GothAlice> (But the integer whole portion can be use for to limit the number of strings needing to be matched.)
[03:18:42] <nateritter> GothAlice: Why the decimal part as a string? Oh, the leading zeros… interesting
[03:21:11] <lpghatguy> All the recommendations I see just say to use cents as your currency unit instead of dollars
[03:22:19] <GothAlice> lpghatguy: That works if you don't need to deal with smaller fractions. Just gets confusing if you start mixing cents and other subdivisions.
[03:23:25] <nateritter> GothAlice: So, back to my original question, I have this… https://gist.github.com/nateritter/7c4d025a7f30280376a4 … and I’d like to add the condition that we only aggregate those values that are greater than or equal to zero… how would I modify this?
[03:23:53] <nateritter> GothAlice: Sorry for the newb question. Took me forever to get to this point.
[03:24:16] <GothAlice> Uhm… don't try to stuff absolutely everything into a $group… that way lies madness and unexpectedly failing queries.
[03:25:40] <GothAlice> It'll also run a lot faster as the first $match in an aggregate can make use of indexes.
[03:25:46] <Boomtime> with aggregation, unless the first operation is $match the rest of the pipeline has to trawl the entire database
[03:26:39] <GothAlice> nateritter: This query ($match on .account and .amount) would benefit most strongly on a compound [account, amount] index. :)
[03:28:08] <Streemo> GothAlice i looked at the meteor/issues/61 thing. its said that they want to focus less on mongodb and make meteor db agnostic
[03:29:25] <GothAlice> Streemo: What upsets me about that is that even if they did absolutely nothing, they'd get ObjectIds back automatically.
[03:30:32] <lpghatguy> (if what I said didn't make sense just ignore me, I'm not sure I'm quite up-to-speed in this conversation)
[03:32:43] <lpghatguy> It looked to be the official standard judging by the mongodb docs when I was getting started with node and MongoDB
[03:39:46] <GothAlice> You'll effectively be locked into only ever using the layer to interact with your DB, at least in terms of inserting records or risk mixing and matching which won't work as expected.
[03:40:12] <GothAlice> If you care about record creation times, you suddenly need to add an extra field for that.
[03:41:09] <Streemo> OK. lets make it one step better, let's say I use actualy mongodb objectId's using the {idGeneration: 'mongo'} (ofc theyll have no meaning other than being a unique id, cause they lack the generation order)
[03:43:38] <GothAlice> Streemo: Comes down to "I'd recommend a dead equine over that framework", really.
[03:46:18] <GothAlice> darkblue_b: If you're just catching up, we discovered to our joy that Meteor generates truly random ObjectIds instead of following the spec, and defaults to random strings for _ids. That's weird.
[03:46:52] <darkblue_b> I got distracted by string search indexes and havent been moving fwd on mongodb
[03:47:56] <GothAlice> http://mongoosejs.com/docs/api.html#schema-objectid-js — doesn't mention "random", which is a good sign.
[03:50:12] <GothAlice> Streemo: More to do with retrieval. Natural order in non-capped collections != insert order, AFIK.
[03:54:31] <Streemo> he said he wanted to memorize pi up to that place (maybe in the 700's if i remember right?) to recite it to people and trick them into thinking its rational
[04:11:59] <GothAlice> Law 57: If something is worth doing twice, it's worth building a tool to do it.
[04:14:18] <GothAlice> Just because it's easy for you doesn't mean it can't be hard on your clients.
[04:15:01] <Streemo> even if yo udidnt remember all 127 if you were able to conjure 127 slightly profound statements on the fly id be impressed
[04:15:10] <GothAlice> https://gist.github.com/amcgregor/9f5c0e7b30b7fc042d81 — Collected from various sources; avoided skippy's list references with extreme difficulty. :P
[04:15:29] <GothAlice> Not very well organized, but there are several consistent themes. I need to tag them individually or something. ¬_¬
[04:15:53] <GothAlice> Streemo: https://gist.github.com/amcgregor/9f5c0e7b30b7fc042d81/revisions < Revision tracking, of course.
[04:17:18] <Streemo> this is cray i used to come up with things like this, didnt know that was a common thing.
[04:18:07] <GothAlice> A few of them, not many. Zen of Python is included wholesale, plus my common addendum. Various common project management phrases, plus a bunch of "Milspec+" quotes.
[04:21:57] <GothAlice> Styled after Lefler's 102 Laws to Live By from ST:TNG episode The Game: http://en.memory-alpha.org/wiki/Robin_Lefler
[04:30:17] <darkblue_b> is there a "string distance" routine in mongodb somewhere? take two strings and return dissimlarity by some measure..
[04:31:20] <GothAlice> If you need something like that, you can pre-store soundex data (and such) numerical representations of the text for easier in-database comparison.
[04:49:11] <Streemo> thats an interesting question darkblue_b i wonder if there are ways to compute a "norm" of a string via some measure - you could probably define one. maybe even vectorize them and compute inner products for 'dissimilarity' measure.
[04:49:56] <darkblue_b> .. oh more than that.. thats how "datamining" is done.. there are many specific variations on that..
[04:50:46] <darkblue_b> I am just trying some basics on equipment I have, to get a sense of the whole
[04:50:52] <Streemo> i never knew what datamining referred to, but im sure the math behind it isnt too bad
[04:58:49] <nateritter> Where should I put a config variable that I want accessible only on the server side, and how would I access it?
[05:15:05] <Boomtime> nateritter: not sure what you mean, can you provide an intended usage of such a variable?
[05:16:01] <nateritter> Boomtime: spent too much time on PHP frameworks… this JS stuff is too easy… haha
[08:54:26] <joannac> sorry, http://docs.mongodb.org/manual/core/security-interface/#http-status-interface
[09:24:18] <Yoda-BZH> I miss-thought that a particular user with the rights roles could access the monitoring web interface
[10:01:16] <Luser> Hi guys. I am currently working on ruby script and woud like to search through the database for one phrase, but am not sure how to do that. I've seen that method: session.with(database: "db_name").command() but am not sure what to insert into command
[11:55:38] <yopp> I have ancestry based tree collection. What the best way to check if there subchildrens of current tree leaf?
[14:02:12] <GothAlice> yopp: Tree structures are one of those things that MongoDB does strangely. (Then again, nested set adjacency lists in relational databases are kinda nuts, too.)
[14:02:47] <GothAlice> yopp: https://github.com/marrow/contentment/blob/develop/web/extras/contentment/components/asset/model/__init__.py#L58-L63
[14:04:05] <GothAlice> Two ways of doing "descendants" searching, by string: https://github.com/marrow/contentment/blob/develop/web/extras/contentment/components/asset/model/__init__.py#L269 or by parents list membership. (I.e. asset X is in the parents list of all children at all levels below asset X)
[14:05:20] <GothAlice> "children" is a list of child ObjectIds (to preserve order) whereas "contents" is your "single-level of children below X" query.
[14:08:41] <yopp> GothAlice, in fact I need to make a aggregation operaion for regexp. For example, in this case for $1 in ^(/a\/.*)\/(:?[^\/]+)$
[14:09:50] <yopp> But I'm not sure how to mapreduce on index only, because I don't need a full document
[14:10:32] <GothAlice> yopp: Well, you didn't give an example document in your structure. Storage of coalesced string paths is unusual and inefficient; my Asset document class uses it primarily for direct lookup. (I.e. given a URL like /foo/bar look up /, /foo, and /foo/bar in reverse order, take the first, most specific element to use as the handler.)
[14:13:00] <GothAlice> yopp: In the case where ^/a/.* is your query, I'd recommend the parents list approach.
[14:14:35] <yopp> but still, if we have 100 root nodes, and each node has like 100 childrens, we will have 10k nodes in result.
[14:14:41] <GothAlice> yopp: So your paths aren't even normalized. (May contain an unlimited nubmer of additional empty separators.)
[14:15:15] <GothAlice> yopp: The "complex" structure I linked you currently contains 24TiB of data in my MongoDB cluster at home. ;) It's quite efficient.
[14:17:16] <GothAlice> yopp: ^/a/.*/[^/]+/ -> [start of string] "/a/" [one or more of any character including slash] "/" [one or more character that isn't slash] "/" -> /a///////////////27/ is valid.
[14:19:27] <yopp> in my case path name is normalized, and there no way that path will have more than one consequential slash
[14:20:03] <GothAlice> yopp: I'm suggesting that such a method of storage is inefficient and unnessicariy in a system that provides rich list types. What does that string represent, exactly?
[14:20:16] <GothAlice> yopp: That string represents a series of parent element references in the form of a path.
[14:23:21] <GothAlice> yopp: It makes things more complex, difficult, inefficient, and painful, actually. It is the wrong representation for the data based on how you wish to query it. (Which is what MongoDB *actually* cares about.) Unfortunately I need to transition networks in a moment. Be back in a bit.
[16:07:20] <GothAlice> ryanwmedlin: Always, always, always, always store your date/time objects in canonical UTC form.
[16:09:15] <ryanwmedlin> run the dat field through an aggregation pipeline to convert the date fielkd to that timezone then group on that?
[16:09:38] <ryanwmedlin> in sql you run it througha convertUTC function as part of the query and do a group by day on that
[16:10:14] <GothAlice> ryanwmedlin: MongoDB doesn't provide functions like that. Instead of passing in localtime references to the query and hoping MongoDB has a way to make those UTC, *make those values UTC in your code*, not MongoDB.
[16:11:19] <GothAlice> ryanwmedlin: Note from my link that I needed to work with ISO DOTW, but MongoDB doesn't use the standard ISO representation, so I converted them *outside* MongoDB.
[16:16:00] <GothAlice> ryanwmedlin: (P-code is good too; it doesn't have to work, I just need a better picture of exactly what you're trying to do. SQL would be good, too, if you only have that.)
[16:19:29] <GothAlice> http://docs.mongodb.org/manual/reference/operator/aggregation-date/ are the sum total of currently available operations on dates.
[16:21:24] <ryanwmedlin> so if your ont he west coast.. want ot group by day by using dayOfMonth.. your basically getting 7 hours of data you dont want? because mongo will group by the UTC data in the db?
[16:29:32] <niczak> This may sound crazy, but what if I make a secondary replica set the primary w/o running the touch command? Will things just load slowly at first until they find their way into memory or will there be other side effects?
[16:31:17] <GothAlice> niczak: Any time you swap primaries there is the possibility of completely rebuilding the new secondaries (i.e. having each pull a complete copy of the dataset from the new primary). This can hose your network a bit while things re-settle.
[16:31:48] <niczak> Goth: Ahh, okay, defintiely want to avoid that. So running the touch command will ensure that everything is "warmed up" and ready to go?
[16:31:51] <GothAlice> (Usually only happens if the oplog becomes so out-of-sync that there is no commonality between them.)
[16:54:38] <niczak> Dataset can definitely fit into RAM but I am thinking about letting the application warm up the database by just running it's most typical queries.
[17:12:03] <joaovagner> https://gist.github.com/JoaoVagner/41b13ba09328735c7e0b how I can perform a search using the dates between this array?
[18:08:09] <GothAlice> Where "date" is the name of the field. If this is embedded under "ticket_answers" the solution becomes: db.collection.find({"ticket_answers.date": {$gt: someDate, $lt: someOtherDate}})
[18:10:36] <joaovagner> however it still fails, I do not know how to turn that into something that your query PHP understands. :(
[18:10:42] <GothAlice> joaovagner: Oh. If you only want the nested records that match back (i.e. not the whole top-level document) you're going to have to turn that into an aggregate query with $match, $unwind, and $match stages.
[18:11:17] <GothAlice> Unfortunately I don't PHP. (And feel nobody else should either.) So I can't really help you with the specific syntax.
[18:12:05] <GothAlice> (Not being able to help with the syntax comes down to not knowing the syntax. Personal preference is just a nice bonus. ;)
[18:13:18] <GothAlice> I'm sure you've seen http://eev.ee/blog/2012/04/09/php-a-fractal-of-bad-design/ by now. :3 Back in the day I ran into some hilarious issues with the str_parse function being exactly the same as register globals. ^_^;
[18:14:11] <GothAlice> (Oh, and that was parse_str, not str_parse, illustrating one of the many points from that article. ;^)
[18:15:11] <GothAlice> I'm freely willing to admit the JVM is a work of art. The rest of the language, though… less so. ;^P
[18:15:34] <cheeser> i like the language, actually, though I flirt with other JVM languages fairly regularly.
[18:16:37] <joaovagner> GothAlice: PHP-FPM and PHP-NG has brought great advances in language that is adopted today the largest in the world :)
[18:17:26] <joaovagner> hhvm (facebook), tested in a client's ecommerce stores, and unfortunately it is far away from reality for a good developer.
[18:19:09] <mike_edmr> php was the first/best at some things and has since grown large but without fixing the issues that make it a poor basis for what it is today
[18:19:46] <GothAlice> I <3 Python so much in that it can produce better interpreters for other languages than the other languages' native runtimes.
[18:19:51] <joaovagner> mike_edmr: at all scales. PHP for command line tools composes, (dependencie manager), web, and other things. Despite being in love with Python, who pays the bills is PHP :)
[18:20:00] <GothAlice> PHP took the C global namespace and exposed it through a template language. That's it, and really, all PHP is.
[18:20:53] <GothAlice> cheeser: from concurrent.futures import ThreadPoolExecutor; with ThreadPoolExecutor(10) as e: e.submit(print, "Hello world!") — BAM, multithreading.
[18:20:57] <darkblue_b> I am a senior editor for a linux "distribution" with over 50 FOSS projects.. php is used in a small number, and that number is getting smaller
[18:21:41] <joaovagner> GothAlice: php-cli yes, including those famous crontab tasks ... I have great applications integrating with the API ingresse.com site and Amazon Web Services
[18:21:45] <GothAlice> (Replace ThreadPoolExecutor with ProcessPoolExecutor and you have split-process deferred execution.)
[18:22:34] <joaovagner> GothAlice: The consultancy work which manages the Route53 all customers via panel with AWS;
[18:23:06] <joaovagner> However we have applications in other languages like java / python doing work BigData.
[18:24:43] <joaovagner> GothAlice: The biggest problem here in Brazil is keeping these things, because the enormous difficulty of keeping a team in Python. The canonical came and stole our best :(
[18:25:21] <joaovagner> PHP community in Brazil is great and is not hard to keep the project with a team keeping. There is no difficulty in keeping projects in PHP due to cost x benefit. In python costs outweigh benefits.
[18:25:59] <GothAlice> (Last PyCon was here, next PyCon will be here, prior to that several years straight of Confoo, etc., etc. Quite the community here.)
[18:27:15] <GothAlice> cheeser: Confoo is almost always cross-scheduled with inclement weather. It's terrible. XD
[18:28:48] <GothAlice> cheeser: “I don’t trust your coffee. Had the last batch sent off to a lab and the results came back: congratulations, your moose is pregnant.”
[18:30:08] <GothAlice> joaovagner: Lived in Panama for a while; it's not the heat that gets me, it's the constant 95% humidity.
[18:32:39] <joaovagner> kkkkkkkkk ciesta who did was my grandfather, of Italian descent. I always thought it a lazy!
[18:33:33] <GothAlice> Science agrees that a nap in the afternoon can actually increase productivity through improved alertness and problem solving abilities. Recharging is good for you.
[18:36:12] <joaovagner> GothAlice: I awake to 05am, I'll run and go to the gym, according to my wife, I take the kids to school, it all before 09am. Most people who know the awakened 09am. I'll have plenty of time to rest when you die. hhahahhahaha
[18:38:06] <unholycrab> heavy write load on the primary. one secondary gets replication lagged, the other secondary doesn't
[18:38:31] <GothAlice> unholycrab: Is there any difference in the hardware the two secondaries are running on?
[18:39:16] <joaovagner> unholycrab: date and time in yours servers, or your router not are not sharing the load
[18:39:39] <GothAlice> unholycrab: There are all sorts of reasons lag might affect only one of the pair.
[18:39:52] <joaovagner> Check communications within the machines, they are getting access and permission required.
[18:40:14] <GothAlice> unholycrab: Make sure ntpd is running on all machines and that they are properly synchronized, time-wise.
[18:40:41] <unholycrab> well, the one was only falling behind while there was heavy write load to the primary
[18:42:58] <GothAlice> joaovagner: amcgregor@chomolungma ~ $ uptime # 11:41:43 up 956 days < maybe a year or two longer than that (this is the uptime from the primary of my larger cluster)
[18:44:39] <joaovagner> I wrote only three applications with Mongo, taking the experience yet. And breaking my head :)
[18:45:14] <GothAlice> MongoDB does open your mind to new possibilities for data design, which can be somewhat mind-blowing when you realize the things you can do. :D
[18:47:47] <joaovagner> GothAlice: indeed, the concept of breaking paradigms NoSQL, did a project a while ago with Java and Neo4j. It's wonderful are old and new concepts while
[18:48:40] <GothAlice> unholycrab: You may be inferring that the kernel has never been upgraded, but it has. Total downtime due to hot upgrades year-to-date of about 15 seconds service level interruption.
[18:48:54] <hexsprite> hey if I want to sort by multiple keys then what's the syntax? for example I want to sort first by "order" and then by "list"... I tried {sort: {order:1, list:1}} but it seems to sort by list first and then order....
[18:49:40] <GothAlice> hexsprite: Depending on your language, your dictionary/mapping/object type might not preserve the order of keys, which would mess up your sort.
[18:50:38] <hexsprite> GothAlice: thanks, i just realized this is probably a meteor specific issue because they wrap the mongodb api... asking on #meteor instead ;)
[18:50:44] <GothAlice> unholycrab: Uninterrupted powered operation. In-place kernel updates don't nuke the allocated stats structure.
[18:53:27] <GothAlice> unholycrab: There are a lot of edge cases to an approach like this, though. Like every counter in the system eventually rolling back around to zero. (Notably network packet and bandwidth counters.)
[19:13:25] <joaovagner> GothAlice: I saw a blurb about jobs in this region, as is the economic situation?
[19:18:08] <kali> mmmm... i was in montreal last week, and i'm glad to be back before november happens :)
[19:19:01] <GothAlice> Montréal gets the extremes of heat in the summer, cold in the winter. I prefer British Columbia; sure, you get rain like it's foggy old England, but it's never too hot and never too cold.
[19:21:11] <kali> cheeser: i actually went there, landed in calgary and drive through the rockies to vancouver
[19:22:25] <GothAlice> Erm… I've taken marked highways up there that were dirt paths. You can avoid it if you want to. ;^P
[19:26:51] <GothAlice> I became so used to the cold just below the arctic circle (Dawson Creek, BC, no relation to Dawsons Creek, the TV show) that at -60°C I was standing outside, barefoot in PJs to smoke. Didn't even notice how cold it was.
[19:27:00] <joaovagner> Devops Senior, 9 years experience in web (php, javascript, api's rest, etc), and infra-estructure (aws, load balancer's, etc)
[19:27:11] <GothAlice> joaovagner: http://www.payscale.com/research/CA/Job=Web_Developer/Salary/4e4b559e/Montréal-QC and http://www.glassdoor.com/Salaries/montreal-software-developer-salary-SRCH_IL.0,8_IM990_KO9,27.htm
[19:28:01] <joaovagner> my country in the recent elections elected a communist government that is destroying our economy. I intend to leave the country for a period escape it. :(
[21:03:37] <bpap> what is the difference between mongodump and mongoexport? more specifically, i want to dump a 7G collection as quickly as possible (consistency not needed). which tool should i use?
[21:05:08] <bpap> ah, so the only difference is the output format? seems like they could have been combined into one tool, no?
[21:08:05] <niczak> I have a replica set with two nodes which are completely in sync. My goal is to make the current secondary the primary and then shutdown the current primary node because we are retiring that server. Following this guide: http://docs.mongodb.org/manual/tutorial/force-member-to-be-primary/ it not working for me, can someone tell me the best/easiest way to make a secondary the primary and remove any references (or just hide) the original primary?
[21:13:25] <kali> niczak: just ask the primary to stepDown(), the secondary will get elected primary.
[21:25:37] <Trindaz> How do I write unit tests for code that accesses a MongoDB instance without needing a running MongoDB instance to run the unit tests
[21:27:10] <GothAlice> Trindaz: Consider https://github.com/marrow/marrow.task/blob/develop/.travis/install.sh which is part of my overall Travis CI integration: https://github.com/marrow/marrow.task/blob/develop/.travis.yml#L10
[21:27:33] <niczak> Kali: Oh wow, that's easy! I actually just changed priorities, removed primary from rs and then secondary became primary and all i well.
[21:27:40] <GothAlice> Trindaz: Locally I use Vagrant to run my tests under a VM (with local mongod).
[21:28:17] <GothAlice> niczak: It's explicitly documented that candidates for election are first ordered by priority; as long as there exists one alive node with a priority of 10, no lower priority nodes will even be considered.
[21:34:13] <Trindaz> what a bummer. I was really hoping there'd be a file-system only node.js library that would let me do this. (https://github.com/openhoat/mongodb-fs looked promising but it broke pretty badly for me)
[21:35:03] <GothAlice> Trindaz: In order to properly mock MongoDB you would have to reimplement nearly all of MongoDB, sadly. It's easier to just spin up mongod on a random local port, expose the port number as an environment variable, and have your tests use it.
[21:38:28] <GothAlice> It's especially easy if your test runner also has good fixture support: https://github.com/marrow/marrow.task/blob/develop/test/model/conftest.py#L9-L14 (with a different scope, I could have this spin up and automatically shut down mongod processes… this just nukes the test db itself between tests.)
[21:42:22] <kali> GothAlice: and any technology distringuisable from magic is insufficiently advanced.
[21:42:29] <GothAlice> https://github.com/mongodb/mongo-python-driver/blob/master/test/__init__.py — pymongo's own tests use environment variables to pass in the test server info.
[21:51:35] <pluffsy> Very newbie question, sorry: I would like to just pass json objects as querys to mongodb. That works fine with simple key value pairs {"x": "y"}, but can I somehow express the $operators as valid json?
[21:54:08] <Boomtime> The $ operators are already valid JSON, perhaps provide an example of what you want to do
[21:54:12] <GothAlice> pluffsy: Valid JSON requires that you quote the strings used as keys, this means "$gt" and such are completely valid keys.
[21:55:28] <pluffsy> ahh, sorry. that explains why I didn’t find anything on google. I was so sure it was invalid, should have tested of course. thanks a lot GothAlice and Boomtime
[22:46:44] <darkblue_b> I have one server running.. it has one port open.. /etc/mongod.conf shows one dir for storage
[22:47:31] <Boomtime> A physical container for collections. Each database gets its own set of files on the file system. A single MongoDB server typically has multiple databases.