Hi 赵科,
Currently, RocketMQ uses mincore system call to determine if the message data
is in memory or swapped out. It seems more reliable than judging hot and cold
data based on the maintenance experience described in your proposal. And
turning off read-ahead for old CommitLog is a good idea, lo
+1
At 2023-03-07 14:55:05, "2017赵科" <464473...@qq.com.INVALID> wrote:
>Hi, RocketMQ Community,
>
>
>As discussed in the previous email, we launched a new RIP[RIP-62] Cold read
>control to support this feature. Now the i...@foxmail.com
> is willing to support the RIP, so I think it is time to star
Hi, thanks for the RIP.
Actually, rocketmq already has two configurations to control the cold read
behavior limitedly: maxTransferCountOnMessageInDisk
and maxTransferBytesOnMessageInDisk.
I have no idea how to handle these two configurations in your RIP, do you
plan to remove them or cooperate wi
+1
> 2023年3月7日 14:55，2017赵科 <464473...@qq.com.INVALID> 写道：
>
> Hi, RocketMQ Community,
>
>
> As discussed in the previous email, we launched a new RIP[RIP-62] Cold read
> control to support this feature. Now the i...@foxmail.com
> is willing to support the RIP, so I think it is time to star
GitHub user socutes added a comment to the discussion: Whether the
ForwardRequestProcessors are being used？
ok
GitHub link:
https://github.com/apache/rocketmq/discussions/6274#discussioncomment-5241433
This is an automatically sent email for dev@rocketmq.apache.org.
To unsubscribe, ple
GitHub user RongtongJin added a comment to the discussion: Whether the
ForwardRequestProcessors are being used？
Hi @socutes , I think this class is useless, you can subbmit a pull request to
delete the class.
GitHub link:
https://github.com/apache/rocketmq/discussions/6274#discussioncomment-
GitHub user RongtongJin edited a comment on the discussion: 关于过期文件删除问题
> 对于长时间下线然后重新上线的Broker，因为CommitLog过期删除策略，始终会保存最后一个文件，如果当前Broker的phyOffset小于主节点的minOffset，这个时候会始终无法进行同步。通常解决办法是人工手动清理掉store文件再重启，我的疑问是能不能在Broker启动load的时候，先手动的检查一下CommitLog的最后一个文件，如果已经过期，那么就主动清空commitlog和consumeQueue。或者说是在主从同步时
GitHub user RongtongJin added a comment to the discussion: 关于过期文件删除问题
> 对于长时间下线然后重新上线的Broker，因为CommitLog过期删除策略，始终会保存最后一个文件，如果当前Broker的phyOffset小于主节点的minOffset，这个时候会始终无法进行同步。通常解决办法是人工手动清理掉store文件再重启，我的疑问是能不能在Broker启动load的时候，先手动的检查一下CommitLog的最后一个文件，如果已经过期，那么就主动清空commitlog和consumeQueue。或者说是在主从同步时，
