Were using SolrCloud 4.10.3 on the Cloudera Platform with a 3 node solr cluster with 2 collections of 3 shards each.
Collection 1: approx size: 15.3 GB Collection 2: size: 1.2GB
Our heap size is 8GB and off heap is 15GB. We have a realtime feed into solr for one of our collections (the other is pretty static). We are constantly getting an out of memory error.
Can anyone help us as to the reason? Should be we having additional shards to spread the load? Or do we need to keep giving more off heap memory? All the cloudera heap graphs show that we are find for heap space (we rarely go above 6.5GB) and GC pauses are not an issue.
Thanks
The best approach should be to upgrade the solr cloud to version
6.2.1.
it also depends on the node architecture if the node is of 32 bit arch. then heap size more than 2gb wont work if the node is of 64 bit arch you can allocate more heap size but can generate gc overhead error.
so better to update solr and add more shards and replicas to avoid the error.
Related
I have an Alfresco 6.2.0 instance on an Ubuntu system using solr specification 6.6.7 and Search Services 1.4.0. I have two cores with currently 155364 documents in the alfresco core and 126054 documents in the archive core. Until today solr hat 1 GB heap space and the last few weeks problems where rising that solr exits with heap space out of memory. Today I raised to 2 Gb, hoping that this is enough.
Is this normal, that solr needs this amount of memory? Are 100.000 documents (no big files excepted the images) really so many, that solr needs more than 1GB? I am just wondering, because the instance is used by a small company.
Thanks,
Florian
1-2 GB heap is not much for Solr, but it is little. In fact, it is the component in the Alfresco architecture that is usually allocated the most memory. The index with the metadata for the full text should fit into th memory if possible to ensure fast searching. You can have a look at Sizing my Alfresco Search Services.
I had to increase the JVM-Memory to 10g from the default value 512m of solr.I changes the values directly in the files ‘solr/bin/solr.cmd‘ and ‘solr/bin/solr.in.cmd‘ and restarted the solr cloud.
All the replica showing statuses as Down mode. And Iam getting error message like status 404 when execute query on the collection.
Nothing is showing in log about the replicas down.
What are steps I need to perform to get the all replicas to Active mode?
I don't really understand why you had to increase the JVM memory from 512 MB directly to 10 GB.
You should know that Solr and Lucene use MMapDirectory as default. This means that all indexes are not loaded in the JVM virtual memory but they are allocated in a dedicated space of memory. This blog post can help you
Considering you have 16GB RAM available, as a first iteration, I'd allocate 4 GB to the JVM so that 12GB remains available for the operating system (and 6GB of index files). Then, by monitoring the system memory and the JVM memory, you can do better tuning.
That being said, I don't think the high JVM allocated memory is enough to break all Solr instances. Can you please verify that you updated only the JVM heap memory value? Can you also verify if the logs show some initialization failures?
There is still some missing information:
How many nodes your SolrCloud is composed of?
How many replicas? And what type of replica?
PS: Considering you are working on solr.cmd and solr.in.cmd I assume your server is Windows, the Linux version invokes the solr.in.sh script.
We are using Flink streaming to run a few jobs on a single cluster. Our jobs are using rocksDB to hold a state.
The cluster is configured to run with a single Jobmanager and 3 Taskmanager on 3 separate VMs.
Each TM is configured to run with 14GB of RAM.
JM is configured to run with 1GB.
We are experiencing 2 memory related issues:
- When running Taskmanager with 8GB heap allocation, the TM ran out of heap memory and we got heap out of memory exception. Our solution to this problem was increasing heap size to 14GB. Seems like this configuration solved the issue, as we no longer crash due to out of heap memory.
- Still, after increasing heap size to 14GB (per TM process) OS runs out of memory and kills the TM process. RES memory is rising over time and reaching ~20GB per TM process.
1. The question is how can we predict the maximal total amount of physical memory and heap size configuration?
2. Due to our memory issues, is it reasonable to use a non default values of Flink managed memory? what will be the guideline in such case?
Further details:
Each Vm is configured with 4 CPUs and 24GB of RAM
Using Flink version: 1.3.2
The total amount of required physical and heap memory is quite difficult to compute since it strongly depends on your user code, your job's topology and which state backend you use.
As a rule of thumb, if you experience OOM and are still using the FileSystemStateBackend or the MemoryStateBackend, then you should switch to RocksDBStateBackend, because it can gracefully spill to disk if the state grows too big.
If you are still experiencing OOM exceptions as you have described, then you should check your user code whether it keeps references to state objects or generates in some other way large objects which cannot be garbage collected. If this is the case, then you should try to refactor your code to rely on Flink's state abstraction, because with RocksDB it can go out of core.
RocksDB itself needs native memory which adds to Flink's memory footprint. This depends on the block cache size, indexes, bloom filters and memtables. You can find out more about these things and how to configure them here.
Last but not least, you should not activate taskmanager.memory.preallocate when running streaming jobs, because streaming jobs currently don't use managed memory. Thus, by activating preallocation, you would allocate memory for Flink's managed memory which is reduces the available heap space.
Using RocksDBStateBackend can lead to significant off-heap/direct memory consumption, up to the available memory on the host. Normally that doesn't cause a problem, when the task manager process is the only big memory consumer. However, if there are other processes with dynamically changing memory allocations, it can lead to out of memory. I came across this post since I'm looking for a way to cap the RocksDBStateBackend memory usage. As of Flink 1.5, there are alternative option sets available here. It appears though that these can only be activated programmatically, not via flink-conf.yaml.
I have 300000 documents stored in solr index. And used 4GB RAM for solr server. But It consumes more than 90% of physical memory. So I moved to my data to a new server which has 16 GB RAM. Again solr consumes more than 90% memory. I don't know how to resolve this issue. I used default MMapDirectory and solr version 4.2.0. Explain me if you have any solution or the reason for this.
MMapDirectory tries to use the OS memory (OS Cache) to the full as much as possible this is normal behaviour, it will try to load the entire index into memory if available. In fact, it is a good thing. Since these memory is available it will try to use it. If another application in the same machine demands more, OS will release it for it. This is one the reason why Solr/Lucene the queries are order of magnitude fast, as most of the call to server ends up memory (depending on the size memory) rather than disk.
JVM memory is a different thing, it can be controlled, only working query response objects and certain cache entries use JVM memory. So JVM size can be configured based on number request and cache entries.
what -Xmx value are you using when invoking the jvm? If you are not using an explicit value, the jvm will set one based on the machine features.
Once you give a max amount of heap to Solr, solr will potentially use all of it, if it needs to, that is how it works. If you to limit to say 2GB use -Xmx=2000m when you invoke the jvm. Not sure how large your docs are, but 300k docs would be considered a smallish index.
Does anyone know how solr4/lucene and the JVM, manages memory?
We have the following case.
We have a 15GB server running only SOLR4/Lucene and the JVM (no custom code)
We had allocated 2GB of memory and the JVM was using 1.9MB. At some point something happened and we run out of memory.
Then we increased the JVM memory to 4GB and we see that gradually, JVM starts to use as much as it can. It is now using 3GB out of the 4GB allocated.
Is that normal JVM memory usage? i.e. Does the JVM always use as much as it can from the allocated space?
Thanks for your help
Facets are using field cache. If you have a lot of fields (even small) which are used for faceting, it very well may be that eventually you hit your memory limit.
if you have a lot of query facets or define a lot of FQ - in your case 1Mb is allocated per each filter cache entry. You have defined 16384 as an upper bound, so you may also hit large numbers there eventually.
Hope this helps.