I'm using Cypher's LOAD CSV syntax in Neo4J 2.1.2. So far it's been a huge improvement over the more manual ETL process required in previous versions. But I'm running into some behavior in a single case that's not what I'd expect and I wonder if I'm missing something.
The cypher query being used is this:
USING PERIODIC COMMIT 500
LOAD CSV FROM 'file:///Users/James/Desktop/import/dependency_sets_short.csv' AS row
MATCH (s:Sense {uid: toInt(row[4])})
MERGE (ds:DependencySet {label: row[2]}) ON CREATE SET ds.optional=(row[3] = 't')
CREATE (s)-[:has]->(ds)
Here's a couple of lines of the CSV:
227303,1,TO-PURPOSE-NOMINAL,t,73830
334471,1,AT-LOCATION,t,92048
334470,1,AT-TIME,t,92048
334469,1,ON-LOCATION,t,92048
227302,1,TO-PURPOSE-INFINITIVE,t,73830
116008,1,TO-LOCATION,t,68204
116007,1,IN-LOCATION,t,68204
227301,1,TO-LOCATION,t,73830
334468,1,ON-DATE,t,92048
116006,1,AT-LOCATION,t,68204
334467,1,WITH-ASSOCIATE,t,92048
Basically, I'm matching a Sense node (previously imported) based on it's ID value which is the fifth column. Then I'm doing a merge to either get a DependencySet node if it exists, or create it. Finally, I'm creating a has edge between the Sense node and the DependencySet node. So far so good, this all works as expected. What's confusing is the performance as the size of the CSV grows.
CSV Lines Time (msec)
------------------------------
500 480
1000 717
2000 1110
5000 1521
10000 2111
50000 4794
100000 5907
200000 12302
300000 35494
400000 Java heap space error
My expectation is that growth would be more-or-less linear, particularly as I'm committing every 500 lines as recommended by the manual, but it's actually closer to polynomial:
What's worse is that somewhere between 300k and 400k rows, it runs into a Java heap space error. Based on the trend from previous imports, I'd expect the import of 400k to take a bit over a minute. Instead, it churns away for about 5-7 minutes before running into the heap space error. It seems like I could split this file into 300,000-line chunks, but isn't that what "USING PERIODIC COMMIT" is supposed to do, more or less? I suppose I could give Neo4J more memory too, but again, it's not clear why I should have to in this scenario.
Also, to be clear, the lookups on both Sense.uid and DependencySet.label are indexed, so the lookup penalty for these should be pretty small. Here's a snippet from the schema:
Indexes
ON :DependencySet(label) ONLINE (for uniqueness constraint)
ON :Sense(uid) ONLINE (for uniqueness constraint)
Any explanations or thoughts on an alternative approach would be appreciated.
EDIT: The problem definitely seems to be in the MATCH and/or CREATE part of the query. If I remove lines 3 and 5 from the Cypher query it performs fine.
I assume that you've already created all the Sense labeled nodes before running this LOAD CSV import. What I think is going on is that as you are matching nodes with the label Sense into memory and creating relationships from the DependencySet to the Sense node via CREATE (s)-[:HAS]->(ds) you are increasing utilization of the available heap.
Another possibility is that the size of your relationship store in your memory mapped settings needs to be increased. In your scenario it looks like the Sense nodes have a high degree of connectivity to other nodes in the graph. When this happens your relationship store for those nodes require more memory. Eventually when you hit 400k nodes the heap is maxed out. Up until that point it needs to do more garbage collection and reads from disk.
Michael Hunger put together an excellent blog post on memory mapped settings for fast LOAD CSV performance. See here: http://jexp.de/blog/2014/06/load-csv-into-neo4j-quickly-and-successfully/
That should resolve your problem. I don't see anything wrong with your query.
i believe the line
MATCH (s:Sense {uid: toInt(row[4])})
makes the time paradigm. somewhere around the 200 000 in the x line of your graph, you have no longer all the Sense nodes in the memory but some of them must be cached to disk. thus all the increase in time is simply re-loading data from cache to memory and vise-versa (otherwise it will be still linear if kept in memory).
maybe if you could post you server memory settings, we could dig deeper.
to the problem of java heap error refer to Kenny's answer
Related
I have following question. I set up an camel -project to parse certain xml files. I have to selecting take out certain nodes from a file.
I have two files 246kb and 347kb in size. I am extracting a parent-child pair of 250 nodes in the above given example.
With the default factory here are the times. For the 246kb file respt 77secs and 106 secs. I wanted to improve the performance so switched to saxon and the times are as follows 47secs and 54secs. I was able to cut the time down by at least half.
Is it possible to cut the time further, any other factory or optimizations I can use will be appreciated.
I am using XpathBuilder to cut the xpaths out. here is an example. Is it possible to not to have to create XpathBuilder repeatedly, it seems like it has to be constructed for every xpath, I would have one instance and keep pumping the xpaths into it, maybe it will improve performance further.
return XPathBuilder.xpath(nodeXpath)
.saxon()
.namespace(Consts.XPATH_PREFIX, nameSpace)
.evaluate(exchange.getContext(), exchange.getIn().getBody(String.class), String.class);
Adding more details based on Michael's comments. So I am kind of joining them, will become clear with my example below. I am combining them into a json.
So here we go, Lets say we have following mappings for first and second path.
pData.tinf.rexd: bm:Document/bm:xxxxx/bm:PmtInf[{0}]/bm:ReqdExctnDt/text()
pData.tinf.pIdentifi.instId://bm:Document/bm:xxxxx/bm:PmtInf[{0}]/bm:CdtTrfTxInf[{1}]/bm:PmtId/bm:InstrId/text()
This would result in a json as below
pData:{
tinf: {
rexd: <value_from_xml>
}
pIdentifi:{
instId: <value_from_xml>
}
}
Hard to say without seeing your actual XPath expression, but given the file sizes and execution time my guess would be that you're doing a join which is being executed naively as a cartesian product, i.e. with O(n*m) performance. There is probably some way of reorganizing it to have logarithmic performance, but the devil is in the detail. Saxon-EE is quite good at optimizing join queries automatically; if not, there are often ways of doing it manually -- though XSLT gives you more options (e.g. using xsl:key or xsl:merge) than XPath does.
Actually I was able to bring the time down to 10 secs. I am using apache-camel. So I added threads there so that multiple files can be read in separate threads. Once the file was being read, it had serial operation to based on the length of the nodes that had to be traversed. I realized that it was not necessary to be serial here so introduced parrallelStream and that now gave it enough power. One thing to guard agains is not to have a proliferation of threads since that can degrade the performance. So I try to restrict the number of threads to twice or thrice the number of cores on the operating machine.
I have a large dataset (about 1B nodes and a few billion relationships) that I am trying to import into Neo4j. I am using the Neo4j import tool. The nodes finished importing in an hour, however since then, the importer is stuck in a node index preparation phase (unless I am reading the output below incorrectly) for over 12 hours now.
...
Available memory:
Free machine memory: 184.49 GB
Max heap memory : 26.52 GB
Nodes
[>:23.39 MB/s---|PROPERTIE|NODE:|LAB|*v:37.18 MB/s---------------------------------------------] 1B
Done in 1h 7m 18s 54ms
Prepare node index
[*SORT:11.52 GB--------------------------------------------------------------------------------]881M
...
My question is how can I speed this up? I am thinking the following:
1. Split up the import command for nodes and relationships and do the nodes import.
2. Create indexes on the nodes
3. Do merge/match to get rid of dupes
4. Do rels import.
Will this help? Is there something else I should try? Is the heap size too large (I think not, but would like an opinion)?
Thanks.
UPDATE
I also tried importing exactly half that data on the same machine and it gets stuck again in that phase at roughly the same amount of time (proportionally). So I have mostly eliminated disk space and memory as an issue.
I have also checked my headers (since I noticed that other people ran into this problem when they had incorrect headers) and they seem correct to me. Any suggestions on what I else should be looking at?
FURTHER UPDATE
Ok so now it is getting kind of ridiculous. I reduced my data size down to just one large file (about 3G). It only contains nodes of a single kind and only has ids. So the data looks something like this
1|Author
2|Author
3|Author
and the header (in a separate file) looks like this
authorID:ID(Author)|:LABEL
And my import still gets stuck in the sort phase. I am pretty sure I am doing something wrong here. But I really have no clue what. Here is my command line to invoke this
/var/lib/neo4j/bin/neo4j-import --into data/db/graph.db --id-type string --delimiter "|" \
--bad-tolerance 1000000000 --skip-duplicate-nodes true --stacktrace true --ignore-empty-strings true \
--nodes:Author "data/author/author_header_label.csv,data/author/author_half_label.csv.gz"
Most of the options for bad-tolerance and skip-duplicate-nodes are there to see if I can make it get through the import somehow at least once.
I think I found the issue. I was using some of the tips here http://neo4j.com/developer/guide-import-csv/#_super_fast_batch_importer_for_huge_datasets where it says I can re-use the same csv file with different headers -- once for nodes and once for relationships. I underestimated the 1-n (ness) of the data I was using, causing a lot of duplicates on the ID. That stage was basically almost all spent on trying to sort and then dedupe. Re-working my queries to extract the data split into nodes and rels files, fixed that problem. Thanks for looking into this!
So basically, ideally always having separate files for each type of node and rel will give fastest results (at least in my tests).
Have a look at the batch importer I wrote for a stress test:
https://github.com/graphaware/neo4j-stress-test
I used both neo4j index and in memory map between two commit. It is really fast and works for both version of neo4j.
Ignore the tests and get the batch importer.
I am running a bulk import of data into a neo4j instance (I have run against 2.2.0 community and enterprise editions as well as 2.1.7 community) running in server mode. My application creates a bunch of nodes in memory, and will peridoically stop to write a series .csv files and send cypher to the neo4j instance to upload the files. (this was done due to performance issues with running the application using the plain old REST API).
Overall, I'm looking to upload something like 150-5000 million nodes, so this is, in principle, the type of thing that neo4j claims to be able to handle relatively well.
Well, anyway, what I'm noticing when I run this against production data is that the application runs in two states -- one where the csv upload processes between 2k-8k of nodes per second, and one where it processes between 80-200 nodes per second. The two states are interwoven when you look at the upload as a time series, and as time goes on, it spends increasingly long amounts of time in the slow state.
Nodes are created through a series of
MERGE (:{NODE_TYPE} {csvLine.key = n.primaryKey}) on create set [PROPERTY LIST];
statements, and I have indexes on everything that I'm doing merges against. This doesn't feel like a degradation in the insert statements, because the slowdown is not linear, but rather bimodal, this feels like there are garbage collection in the neo4j instance. What is the best way to tune the neo4j JVM garbage collector for frequent bulk inserts?
neo4j.properties:
neostore.nodestore.db.mapped_memory=50M
neostore.relationshipstore.db.mapped_memory=500M
#neostore.relationshipgroupstore.db.mapped_memory=10M
neostore.propertystore.db.mapped_memory=100M
#neostore.propertystore.db.strings.mapped_memory=130M
neostore.propertystore.db.arrays.mapped_memory=130M
neo4j-wrapper.conf:
wrapper.java.additional=-XX:+UseConcMarkSweepGC
wrapper.java.additional=-XX:+CMSClassUnloadingEnabled
wrapper.java.additional=-XX:-OmitStackTraceInFastThrow
wrapper.java.additional=-XX:hashCode=5
wrapper.java.initmemory=8194
wrapper.java.maxmemory=8194
This felt like the sweet spot for both the overall heap memory and the neostore stuff. Increasing the overall heap degraded performance. That said, the neo4j garbage collection logs frequently have that GC (Allocation Failure) message.
EDIT: in response to Michael Hunger:
the machine has 64 GB of RAM, and nothing seems to be maxed out. It also seems like only a small number of cores are being used at any time. Garbage collector profiling shows that the garbage collector seems to be running quite frequently.
The exact cypher statements are, for example:
USING PERIODIC COMMIT 110000 LOAD CSV WITH HEADERS FROM 'file:///home/jschirmer/Event_2015_4_773476.csv' AS csvLine MERGE (s:Event {primaryKey: csvLine.primaryKey}) ON CREATE SET s.checkSum= csvLine.checkSum,s.epochTime= toInt(csvLine.epochTime),s.epochTimeCreated= toInt(csvLine.epochTimeCreated),s.epochTimeUpdated= toInt(csvLine.epochTimeUpdated),s.eventDescription= csvLine.eventDescription,s.fileName= csvLine.fileName,s.ip= csvLine.ip,s.lineNumber= toInt(csvLine.lineNumber),s.port= csvLine.port,s.processPid= csvLine.processPid,s.rawEventLine= csvLine.rawEventLine,s.serverId= csvLine.serverId,s.status= toInt(csvLine.status);
USING PERIODIC COMMIT 110000 LOAD CSV WITH HEADERS FROM 'file:///home/jschirmer/Event__File_2015_4_773476.csv' AS csvLine MATCH (n:SC_CSR{primaryKey: csvLine.Event_id}), (s:File{fileName: csvLine.File_id}) MERGE n-[:DATA_SOURCE]->s;
Though there are serveral such statements being made
I have tried a single concurrent transaction as well as running several (~3) such statements in parallel (which gives a roughly 2x improvement). I've tried tuning the periodic commit frequency, and the size of the file. It seems that this maximizes performance when the csv file is roughly 100k lines, which means that really, the periodic commit can be off.
I have not run profile on the staments. I will do that, but I thought that the eager merget problem was avoided by using MERGE ... on create statements.
IN general your config looks ok, what RAM does your machine have?
For the things you merge against I'd recommend constraint instead of an index.
What's your tx size? And how many concurrent tx do you run?
Instead of your generic merge statement (which wouldn't compile) can you share the concrete statements?
Did you profile the statements? Perhaps you run into the eager pipe problem:
http://www.markhneedham.com/blog/2014/10/23/neo4j-cypher-avoiding-the-eager/
Do you use periodic commit?
How large are you CSV files?
See: http://neo4j.com/developer/guide-import-csv/
I'm trying to query ~8 million nodes in a neo4j database. I can do queries which hit the index for exact matches easily enough, but is there a performant way to do aggregations?
MATCH (r:Resident) RETURN r.forename, count(r.forename) ORDER BY count(r.forename)
This query just sits there until I eventually restart my server. I've read the performance guides and I'm watching vm_stat and it seems to be quickly running out of pages free. I've tried tuning the memory / JVM heap settings to various things, but I'm not sure I completely know what I'm doing ;) I've got an 8 GB MacBook Air with an SSD drive in case that's helpful for suggesting settings. Also, here's my stats on my DB from webadmin:
10,236,226 nodes
56,280,161 properties
10,190,430 relationships
2 relationship types
14,535 MB database disk usage
I inserted 8M nodes with just 1 prop, and got this query down to ~20s without changing the default settings (after warming up the cache--first time took 90s), which is comparable to other databases like postgres (which I also tested).
Some things you could try to do:
raise the sizes on the appropriate files mmio settings (as per the file sizes in data/graph.db/) in conf/neo4j.properties (at the top)
increase the node cache size in neo4j.properties
increase the heap init/max in neo4j-wrapper.conf
make sure you have enough RAM left over
I have written a variety of queries using cypher that take no less than 200ms per query. They're very straightforward, so I'm having trouble identifying where the bottleneck is.
Simple Match with Parameters, 2200ms:
Simple Distinct Match with Parameters, 200ms:
Pathing, 2500ms:
At first I thought the issue was a lack of resources, because I was running neo4j and my application on the same box. While the performance monitor indicated that CPU and memory were largely free'd up and available, I moved the neo4j server to another local box and observed similar latency. Both servers are workstations with fairly new Xeon processors, 12GB memory and SSDs for the data storage. All of the above leads me to believe that the latency isn't due to my hardware. OS is Windows 7.
The graph has less than 200 nodes and less than 200 relationships.
I've attached some queries that I send to neo4j along with the configuration for the server, database, and JVM. No plugins or extensions are loaded.
Pastebin Links:
Database Configuration
Server Configuration
JVM Configuration
[Expanding a bit on a comment I made earlier.]
#TFerrell: Your comments state that "all nodes have labels", and that you tried applying indexes. However, it is not clear if you actually specified the labels in your slow Cypher queries. I noticed from your original question statement that neither of your slower queries actually specified a node label (which presumably should have been "Project").
If your Cypher query does not specify the label for a node, then the DB engine has to test every node, and it also cannot apply an index.
So, please try specifying the correct node label(s) in your slow queries.
Is that the first run or a subsequent run of these queries?
You probably don't have a label on your nodes and no index or unique constraint.
So Neo4j has to scan the whole store for your node pulling everything into memory, loading the properties and checking.
try this:
run until count returns 0:
match (n) where not n:Entity set n:Entity return count(*);
add the constraint
create constraint on (e:Entity) assert e.Id is unique;
run your query again:
match (n:Element {Id:{Id}}) return n
etc.
It seems there is something wrong with the automatic memory mapping calculation when you are on Windows (memory mapping on heap).
I just looked at your messages.log and added up some numbers, so it seems the mmio alone is enough to fill your java heap space (old-gen) leaving no room for the database, caches etc.
Please try to amend that by fixing the mmio config in your conf/neo4j.properties to more sensible values (than the auto-calculation).
For your small store just uncommenting the values starting with #neostore. (i.e. remove the #) should work fine.
Otherwise something like this (fitting for a 3GB heap) for a larger graph (2M nodes, 10M rels, 20M props,10M long strings):
neostore.nodestore.db.mapped_memory=25M
neostore.relationshipstore.db.mapped_memory=250M
neostore.propertystore.db.mapped_memory=250M
neostore.propertystore.db.strings.mapped_memory=250M
neostore.propertystore.db.arrays.mapped_memory=0M
Here are the added numbers:
auto mmio: 134217728 + 134217728 + 536870912 + 536870912 + 1073741824 = 2.3GB
stores sizes: 1073920 + 1073664 + 3221698 + 3221460 + 1073786 = 9MB
JVM max: 3.11 RAM : 13.98 SWAP: 27.97 GB
max heaps: Eden: 1.16, oldgen: 2.33
taken from:
neostore.propertystore.db.strings] brickCount=8 brickSize=134144b mappedMem=134217728b (storeSize=1073920b)
neostore.propertystore.db.arrays] brickCount=8 brickSize=134144b mappedMem=134217728b (storeSize=1073664b)
neostore.propertystore.db] brickCount=6 brickSize=536854b mappedMem=536870912b (storeSize=3221698b)
neostore.relationshipstore.db] brickCount=6 brickSize=536844b mappedMem=536870912b (storeSize=3221460b)
neostore.nodestore.db] brickCount=1 brickSize=1073730b mappedMem=1073741824b (storeSize=1073786b)