Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.
Original topic: undersized region是什么状态的
What are the specific metrics for undersized-region-count
and oversized-region-count
? What issues do they reflect? From the literal meaning, it seems to be the number of regions under a certain size. How large is this size? During the period when the undersized-region-count increased in the screenshot, a large amount of data was being written.
Are the 439 small regions due to setting a single table to one region or limiting TiKV’s split.byte-threshold? The default is 30MB/s, which may cause regions to be unable to merge or limit the throughput of regions, resulting in many small regions being split.
Or the table has pre-allocated many regions, but the table data is not that much.
There are almost no empty regions. You can check if the duration of 999 or 99 has any impact.
The title of this issue was later changed to something more understandable:
add monitor to record some unhealthy region such as need-merge
and need-split
region. #4783
I also have hundreds of undersized regions. In my case, there are partitioned tables with pre-split parameters added.
They won’t be merged and always exist.
Can this thread be closed? Why did you dig up last year’s post?
So, has this issue been resolved? I can’t open the issue~
There are so many metrics in Grafana that I really can’t understand what they mean.
The issue is about monitoring unhealthy regions, with additional monitoring for need-merge and need-split regions. It has already been closed, and the functionality is already available. It’s just that ↓ it’s not mentioned in the documentation.
Region Health Panel
This document hasn’t been updated for a while. I’ll raise an issue to update the document later. This thread can probably be considered concluded for now.
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.