如何使用Rails + Mongo解决URL性能低下问题

Question

我们可以在伸缩/运维问题上使用建议。

我们有一个简单的移动应用程序，它运行在Rails 3.2.12上，使用MongoMapper而不是ActiveRecord。有一个数据库调用偶尔表现不佳，导致用户写入并抱怨。目前还不清楚原因。由于MongoMapper的原因，我们无法安装Mongo，而且Mongo返回的数据也不多(< 200KB)。在控制器中也没有执行太多的逻辑。

随着更多的用户，这个问题似乎变得更加严重。服务器在VPS上运行，VPS与30个节点共享。托管公司表示，平均I/O利用率为12%，远低于临界阈值。

既然我们不能使用NewRelic，那么解决这个问题的最佳方法是什么呢？

下面是explain的输出

User.collection.find({:username_downcase => 'banana2006'}).explain
=> {"cursor"=>"BtreeCursor username_downcase", 
    "isMultiKey"=>false, 
    "n"=>1, 
    "nscannedObjects"=>1, 
    "nscanned"=>1, 
    "nscannedObjectsAllPlans"=>1, 
    "nscannedAllPlans"=>1, 
    "scanAndOrder"=>false, 
    "indexOnly"=>false, 
    "nYields"=>0, 
    "nChunkSkips"=>0, 
    "millis"=>3, 
    "indexBounds"=>{"username_downcase"=>[["banana2006", "banana2006"]]}, 
    "allPlans"=>[{"cursor"=>"BtreeCursor username_downcase", "n"=>1, "nscannedObjects"=>1, "nscanned"=>1, "indexBounds"=>{"username_downcase"=>[["banana2006", "banana2006"]]}}], "oldPlan"=>{"cursor"=>"BtreeCursor username_downcase", "indexBounds"=>{"username_downcase"=>[["banana2006", "banana2006"]]}}, "server"=>"x.com"}

来自db.serverStatus的输出

> db.serverStatus()
{
        "host" : "mongo.x.com",
        "version" : "2.2.0",
        "process" : "mongod",
        "pid" : 15957,
        "uptime" : 5232267,
        "uptimeMillis" : NumberLong("5232267460"),
        "uptimeEstimate" : 5178261,
        "localTime" : ISODate("2013-05-19T19:32:14.561Z"),
        "locks" : {
                "." : {
                        "timeLockedMicros" : {
                                "R" : NumberLong(131563265),
                                "W" : NumberLong("2824934127")
                        },
                        "timeAcquiringMicros" : {
                                "R" : NumberLong(536751143),
                                "W" : NumberLong(644540368)
                        }
                },
                "admin" : {
                        "timeLockedMicros" : {
                                "r" : NumberLong(11906),
                                "w" : NumberLong(0)
                        },
                        "timeAcquiringMicros" : {
                                "r" : NumberLong(424),
                                "w" : NumberLong(0)
                        }
                },
                "local" : {
                        "timeLockedMicros" : {
                                "r" : NumberLong(13829064),
                                "w" : NumberLong(0)
                        },
                        "timeAcquiringMicros" : {
                                "r" : NumberLong(96863334),
                                "w" : NumberLong(0)
                        }
                },
                "x-development" : {
                        "timeLockedMicros" : {
                                "r" : NumberLong(22074626),
                                "w" : NumberLong(645528)
                        },
                        "timeAcquiringMicros" : {
                                "r" : NumberLong(2876041),
                                "w" : NumberLong(3693)
                        }
                },
                "x-production" : {
                        "timeLockedMicros" : {
                                "r" : NumberLong("39251850394"),
                                "w" : NumberLong(1466862624)
                        },
                        "timeAcquiringMicros" : {
                                "r" : NumberLong("17410130690"),
                                "w" : NumberLong(858232658)
                        }
                },
                "z-development" : {
                        "timeLockedMicros" : {
                                "r" : NumberLong(1897461),
                                "w" : NumberLong(0)
                        },
                        "timeAcquiringMicros" : {
                                "r" : NumberLong(134836),
                                "w" : NumberLong(0)
                        }
                }
        },
        "globalLock" : {
                "totalTime" : NumberLong("5232267461000"),
                "lockTime" : NumberLong("2824934127"),
                "currentQueue" : {
                        "total" : 0,
                        "readers" : 0,
                        "writers" : 0
                },
                "activeClients" : {
                        "total" : 0,
                        "readers" : 0,
                        "writers" : 0
                }
        },
        "mem" : {
                "bits" : 64,
                "resident" : 87,
                "virtual" : 9071,
                "supported" : true,
                "mapped" : 4207,
                "mappedWithJournal" : 8414
        },
        "connections" : {
                "current" : 3,
                "available" : 9597
        },
        "extra_info" : {
                "note" : "fields vary by platform",
                "heap_usage_bytes" : 198457056,
                "page_faults" : 3176777
        },
        "indexCounters" : {
                "btree" : {
                        "accesses" : 18208995,
                        "hits" : 18208994,
                        "misses" : 0,
                        "resets" : 0,
                        "missRatio" : 0
                }
        },
        "backgroundFlushing" : {
                "flushes" : 87204,
                "total_ms" : 563603,
                "average_ms" : 6.463040686207055,
                "last_ms" : 1,
                "last_finished" : ISODate("2013-05-19T19:31:55.201Z")
        },
        "cursors" : {
                "totalOpen" : 0,
                "clientCursors_size" : 0,
                "timedOut" : 0
        },
        "network" : {
                "bytesIn" : 9286320357,
                "bytesOut" : 148669944094,
                "numRequests" : 5102457
        },
        "opcounters" : {
                "insert" : 0,
                "query" : 3213569,
                "update" : 1989197,
                "delete" : 0,
                "getmore" : 30944,
                "command" : 216139
        },
        "asserts" : {
                "regular" : 0,
                "warning" : 0,
                "msg" : 0,
                "user" : 0,
                "rollovers" : 0
        },
        "writeBacksQueued" : false,
        "dur" : {
                "commits" : 30,
                "journaledMB" : 0.04096,
                "writeToDataFilesMB" : 0.043131,
                "compression" : 0.9447148096039855,
                "commitsInWriteLock" : 0,
                "earlyCommits" : 0,
                "timeMs" : {
                        "dt" : 3069,
                        "prepLogBuffer" : 0,
                        "writeToJournal" : 0,
                        "writeToDataFiles" : 0,
                        "remapPrivateView" : 0
                }
        },
        "recordStats" : {
                "accessesNotInMemory" : 1102532,
                "pageFaultExceptionsThrown" : 657056,
                "admin" : {
                        "accessesNotInMemory" : 0,
                        "pageFaultExceptionsThrown" : 0
                },
                "local" : {
                        "accessesNotInMemory" : 0,
                        "pageFaultExceptionsThrown" : 0
                },
                "x-development" : {
                        "accessesNotInMemory" : 1555,
                        "pageFaultExceptionsThrown" : 1304
                },
                "x-production" : {
                        "accessesNotInMemory" : 1074115,
                        "pageFaultExceptionsThrown" : 639842
                },
                "z-development" : {
                        "accessesNotInMemory" : 0,
                        "pageFaultExceptionsThrown" : 0
                }
        },
        "ok" : 1
}

Answer 1

您可以设置MongoDB profiling，尽管默认的100ms应该足够好。

但是，如果我们不知道Mongo查询是什么，我们就不能告诉你如何优化它们。什么是“偶尔表现不佳的一个数据库调用”？

在注释中，您引用了一个使用snapshot选项的长查询。这可能会导致锁争用，所以如果您可以安全地删除快照选项，您可能会看到改进。

我不知道你为什么说你不能使用新遗物。它可能不会为您提供数据库查询的自动计时，但是因为您知道问题出在哪一个，所以可以使用method tracers来隔离问题，并确保它是数据库而不是代码的其他部分。