Yutaka
October 7, 2022, 4:54am
1
Hello. about Edge Data replication.
I’m using Edge Data replication between InfluxDB OSS and InfluxDB OSS(like a Cloud) on local network.
those PC running on Windows10 Pro (21H1)
Influxdb Version 2.4
How much data can be retained while connectivity is lost between cloud.
I had bad network connection at Cloud side.
after recover connection, some of data are missing on Cloud.
Are there any settings for it?
Hello @Yutaka ,
Hmm I’m not sure. I’m asking around. Thanks for your patience.
Please ping me tomorrow if I don’t respond end of day.
Ah ha! found it
maxQueueSizeBytes: Maximum replication queue size in bytes (default is 67108860
, must be greater than or equal to 33554430
).
1 Like
Yutaka
October 21, 2022, 1:55am
3
Hello @Anaisdg
Thanks for information.
I appreciate your response
I realized that information and documents is on API section by your information.
I confirmed maxQueueSizeBytes will be enough with my situation.
I tested with bad communication condition.
“currentQueueSizeBytes” will reduce within each 10 to 20 sec.
I found following information at Github.
It seems to be similar things of mine.
influxdata:master
← ZuperZee:zuperzee/fix-wrong-max-age-transformation-from-seconds
opened 08:32PM - 29 Aug 22 UTC
- Closes https://github.com/influxdata/influxdb/issues/23681
### Required check… list
- [ ] Sample config files updated (both `/etc` folder and `NewDemoConfig` methods) (influxdb and plutonium)
- [ ] openapi swagger.yml updated (if modified API) - link openapi PR
- [x] Signed [CLA](https://influxdata.com/community/cla/) (if not already signed)
### Description
1-3 sentences describing the PR (or link to well written issue)
Creating a replication stream with maxAgeSeconds (following this guide https://docs.influxdata.com/influxdb/v2.4/write-data/replication/replicate-data/?t=API)
Creates a replication stream with max age in nanoseconds (100_000_000ns = 1s) which makes the maxAgeSeconds=604800 in the docs to be 0.6048ms instead of 168h
Related issue https://github.com/influxdata/influxdb/issues/23681
### Context
Why was this added? What value does it add? What are risks/best practices?
Converts the maxAgeSeconds to be seconds instead of nanoseconds.
### Affected areas (delete section if not relevant):
List of user-visible changes. As a user, what would I need to see in docs?
Examples:
CLI commands, subcommands, and flags
API changes
Configuration (sample config blocks)
A quick workaround was to add 100_000_000 to the value stored in maxAgeSeconds. Using this fix would make those 100_000_000 times longer.
### Severity (delete section if not relevant)
i.e., ("recommend to upgrade immediately", "upgrade at your leisure", etc.)
### Note for reviewers:
Check the semantic commit type:
- Feat: a feature with user-visible changes
- Fix: a bug fix that we might tell a user “upgrade to get this fix for your issue”
- Chore: version bumps, internal doc (e.g. README) changes, code comment updates, code formatting fixes… must not be user facing (except dependency version changes)
- Build: build script changes, CI config changes, build tool updates
- Refactor: non-user-visible refactoring
- Check the PR title: we should be able to put this as a one-liner in the release notes
So I will wait to fix this issue.
Thank you again!!
Hello @Anaisdg ,
I am a InfluxDB v2.6 user having setup remote replication between two InfluxDB 2.6 instances, one acting as local and one more as a cloud. I have the same condition as @Yutaka . The connection between local and cloud is very infrequent and the data replication has to be done for larger queue size. I want to know what is the maximum supported queue size for InfluxDB?
Many thanks in advance and looking forward to your suggestion!
Hello @Anaisdg
I’m also wondering about the maximum supported queue size (and maxAgeSeconds) for InfluxDB?
-frodes-bre