> What are the downsides of increasing cache-max-memory-size. The system will consume more RAM.
> With 128GB setting it to 0 wont create any issues? It's not possible to say there will be no issues. Queries have significant impact on RAM usage, and a spike in queries might lead to an OOM if the cache is also consuming a lot of RAM. Perhaps you can set it to something very high, like 64GB, rather than truly unlimited. > How large can it grow? I don't think there's a technical limit, and the practical limit depends on lots of factors in the write, query, and compaction load. Once you raise the limits you can monitor the cache usage with the memBytes stat on the tsm1_cache measurement in the _internal database. SELECT hostname, path, memBytes FROM _internal..tsm1_cache WHERE time > now() - 10s That will return the node, the path, and the cache memory used for each active shard. You can identify the database and shard ID from the path. > Will it improve performance if its increased? Yes, it should increase write throughput. On Fri, Oct 7, 2016 at 4:26 AM, Dmitriy Akulov <[email protected]> wrote: > What are the downsides of increasing cache-max-memory-size. With 128GB > setting it to 0 wont create any issues? > How large can it grow? > Will it improve performance if its increased? > > Thank you! > > On Fri, Oct 7, 2016 at 2:37 AM, Sean Beckett <[email protected]> wrote: > >> You need to increase cache-max-memory-size. The default is 512mb which is >> fine for most workloads, but is too small for larger write volumes or >> higher cardinality datasets. If you have lots of RAM you could set it to 0 >> to disable the limit and never get that error. You can also examine the >> memBytes stat on cache measurement to get a sense of how big the caches are >> in memory once the limit is raised. >> >> "cache-snapshot-memory-size" shouldn't be changed. Increasing it causes >> the cache in memory to be larger, which slows down snapshot compactions and >> increases startup time. >> >> The aborted compactions messages are informational and can be ignored. We >> will probably move those to trace level logging: https://github.com/in >> fluxdata/influxdb/issues/7425 >> >> On Wed, Oct 5, 2016 at 3:11 PM, Dmitriy Akulov <[email protected]> >> wrote: >> >>> Hello, I dont completely understand what this error means. "write failed >>> for shard 143: engine: cache maximum memory size exceeded" >>> The server has 128GB of RAM and the config has default settings. >>> As far as I understand the "cache-max-memory-size" and >>> "cache-snapshot-memory-size" parameters are not related to RAM. >>> What exactly do they do? How they should be configured? Based on >>> incoming data? Based on RAM? >>> >>> What should I do to fix this? >>> >>> Also what exactly this message means "aborted level 1 group (0). >>> compaction in progress"? >>> >>> Log that contains both errors https://gist.githubusercontent >>> .com/jimaek/5f7fe91cc5f689273bed64401e564db1/raw/5aea6861548 >>> a2f3d86940000f25576402bca29f5/gistfile1.txt >>> >>> Thank you >>> >>> -- >>> Remember to include the InfluxDB version number with all issue reports >>> --- >>> You received this message because you are subscribed to the Google >>> Groups "InfluxDB" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to [email protected]. >>> To post to this group, send email to [email protected]. >>> Visit this group at https://groups.google.com/group/influxdb. >>> To view this discussion on the web visit https://groups.google.com/d/ms >>> gid/influxdb/88679401-4e5b-4cee-9f20-80266333994f%40googlegroups.com >>> <https://groups.google.com/d/msgid/influxdb/88679401-4e5b-4cee-9f20-80266333994f%40googlegroups.com?utm_medium=email&utm_source=footer> >>> . >>> For more options, visit https://groups.google.com/d/optout. >>> >> >> >> >> -- >> Sean Beckett >> Director of Support and Professional Services >> InfluxDB >> >> -- >> Remember to include the InfluxDB version number with all issue reports >> --- >> You received this message because you are subscribed to a topic in the >> Google Groups "InfluxDB" group. >> To unsubscribe from this topic, visit https://groups.google.com/d/to >> pic/influxdb/vy1uhSsW7fw/unsubscribe. >> To unsubscribe from this group and all its topics, send an email to >> [email protected]. >> To post to this group, send email to [email protected]. >> Visit this group at https://groups.google.com/group/influxdb. >> To view this discussion on the web visit https://groups.google.com/d/ms >> gid/influxdb/CALGqCvNycHt1eBV95-JjpFb-DwE95Gps9rpKQgr9pR6sfp >> 9Ang%40mail.gmail.com >> <https://groups.google.com/d/msgid/influxdb/CALGqCvNycHt1eBV95-JjpFb-DwE95Gps9rpKQgr9pR6sfp9Ang%40mail.gmail.com?utm_medium=email&utm_source=footer> >> . >> >> For more options, visit https://groups.google.com/d/optout. >> > > -- > Remember to include the InfluxDB version number with all issue reports > --- > You received this message because you are subscribed to the Google Groups > "InfluxDB" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > Visit this group at https://groups.google.com/group/influxdb. > To view this discussion on the web visit https://groups.google.com/d/ > msgid/influxdb/CACMxjo5PGZcL1YaF47waO%3D1kgizgLKpC2w2S% > 2BvHe13er2aYA4g%40mail.gmail.com > <https://groups.google.com/d/msgid/influxdb/CACMxjo5PGZcL1YaF47waO%3D1kgizgLKpC2w2S%2BvHe13er2aYA4g%40mail.gmail.com?utm_medium=email&utm_source=footer> > . > > For more options, visit https://groups.google.com/d/optout. > -- Sean Beckett Director of Support and Professional Services InfluxDB -- Remember to include the InfluxDB version number with all issue reports --- You received this message because you are subscribed to the Google Groups "InfluxDB" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at https://groups.google.com/group/influxdb. To view this discussion on the web visit https://groups.google.com/d/msgid/influxdb/CALGqCvNatPihEcL88G3MFWsaW5JByX2_Gihnt9vwt2Owut%2BQ%3DA%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
