3 months ago
Hi, we have moved from MongoCloud to Railway 3 weeks ago, mainly due to the reason that there is no private network peering option on railway to MongoCloud. Using the default deployment for Database with identical resources, our workloads started to struggle. P99 latency on queries over time shot up to 5s+ wait time, and we started noticing a lot of timed out requests due to this.
This is from the Deploy logs "ns": "AggregatorDB.$cmd", "numYields": 0, "protocol": "op_msg", "remote": "[fd12:2822:3278:0:9000:44:7c67:7a22]:60764", "reslen": 60, "storage": {}, "type": "command", "waitForWriteConcernDurationMillis": 4930, "writeConcern": { "provenance": "clientSupplied", "w": "majority", "wtimeout": 0 }
As you see write concern wait was 5s, sometimes this shoots up to 20s..
According to this report your disk is slow : dd if=/dev/urandom of=testfile bs=1M count=200 conv=fdatasync 200+0 records in 200+0 records out 209715200 bytes (210 MB, 200 MiB) copied, 27.2316 s, 7.7 MB/s
We now unfortunately had to move this back to mongo cloud and the performance issue immediately went away despoite us connecting to them over public internet. Is there any way to optimize this so we can move back to railway? Any suggestions?
Project ID: 2a24b8a8-5455-447a-a062-292f785b7f1a
0 Replies
I also see mongo logs like this
I also see this in mongo logs
{ "message": "vm.max_map_count is too low", "attributes": { "attr": { "currentValue": 65530, "maxConns": 838860, "recommendedMinimum": 1677720 }, "c": "NETWORK", "ctx": "initandlisten", "id": 5123300, "level": "info", "s": "W", "t": { "$date": "2025-08-14T04:06:41.530+00:00" }, "tags": [ "startupWarnings" ] }, "tags": { "project": "2a24b8a8-5455-447a-a062-292f785b7f1a", "environment": "899edefe-77a5-42a5-9438-ff11f2784743", "service": "80c4d956-f9cb-4e03-8f36-b9fb8673e882", "deployment": "e6503657-a1fd-4798-9aaf-25034a0a8870", "replica": "eb9d1fb8-8849-4091-90bf-e4e11a84d3bc" }, "timestamp": "2025-08-14T04:06:41.531795842Z" }