refactor replDataBuffer #36
Conversation
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## cluster-asm #36 +/- ##
==============================================
Coverage ? 68.60%
==============================================
Files ? 125
Lines ? 72994
Branches ? 0
==============================================
Hits ? 50078
Misses ? 22916
Partials ? 0
🚀 New features to boost your workflow:
|
src/replication.c
Outdated
| buf->blocks->free = zfree; | ||
| } | ||
|
|
||
| void replDataBufFree(replDataBuf *buf) { |
There was a problem hiding this comment.
see that there is also peak variable inside replDataBuf. For replicaOf, we don't clear it, so it is kind of a stat. I guess either we should move that variable out of replDataBuf or perhaps we should call this function replDataBufClear() instead. Otherwise, it sounds a bit confusing that we access a variable after "free" call for the struct.
There was a problem hiding this comment.
now i use replDataBufClear that is better than replDataBufFree
we don't clear it, so it is kind of a stat
for ASM, maybe we need to record the peak across all tasks. maybe we need a new struct: asmTaskManager, I want to use a slot map in it to quickly check if a slot is importing or migrating, for request filtering or write pausing, instead of iterating slot ranges.
There was a problem hiding this comment.
yeah, makes sense. perhaps we can reuse bitmap logic and keep track of importing migrating slots.
|
@ShooterIT Looks good to me. Just have minor comments. We can fix those issues later, as you wish. |
No description provided.