Hi, Please note that ArangoDB 3.5 is EOL. So the only advice I can give you is to try the latest bugfix release of this release series, 3.5.7, or upgrade to a supported version 3.10.x better 3.11.x and check back whether your issue persists. Cheers, Willi
On Wednesday, November 1, 2023 at 4:08:11 AM UTC+1 lcy...@gmail.com wrote: > Hi all, > > I hope this message finds you well. I am writing to seek assistance and > insights regarding the memory usage within our ArangoDB cluster. We are > currently experiencing variations in memory consumption across different > nodes in our cluster, and we would appreciate your expertise in addressing > this issue. > > Here is an overview of our cluster's configuration and the specific > concerns we have: > > Cluster Overview: > 1. ArangoDB Version: 3.5.2 > 1. 6 Nodes in the cluster > 2. Each collection consists of 12 shards with a replication factor of 2 > 3. Each node is equipped with 128GB of RAM > > Specific Concerns: > 1. Some nodes are consistently maintaining RAM usage between 80-90%, while > others remain in the range of 40-50%. We are concerned about the disparity > in memory consumption among nodes. > 2. The largest collection contains approximately 150 million documents, > while the total data volume across all collections in the cluster is around > 400-450 million documents. > 3. Interestingly, we noticed that the three nodes serving as Agency roles > are among the ones with higher memory usage. We are curious if there is any > correlation between the Agency roles and high memory consumption. > 4. When inspecting the distribution of primary shards across nodes, we > observed that nodes with lower memory usage tend to host fewer primary > shards. We wonder if there is a relationship between the distribution of > primary shards and memory usage. > 5. In an attempt to address the memory issues, we manually initiated data > shard migration from the heavily loaded nodes to the less busy ones. > However, the results were not as effective as expected. We would appreciate > any guidance or best practices for optimizing shard migrations within our > cluster. > > We believe that your valuable insights and suggestions can help us better > understand and address the memory-related challenges we are currently > facing. Any advice, recommendations, or potential explanations for the > observed behavior would be greatly appreciated. > > Thank you for your time and assistance. We look forward to your response > and any support you can provide in optimizing our ArangoDB cluster's > performance. > > Best regards, > Kevin -- You received this message because you are subscribed to the Google Groups "ArangoDB" group. To unsubscribe from this group and stop receiving emails from it, send an email to arangodb+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/arangodb/c8d6e65c-aabd-4dc8-af34-72c0ee319bb6n%40googlegroups.com.
